]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - fs/ext4/ext4.h
ext4: fix races between changing inode journal mode and ext4_writepages
[mirror_ubuntu-bionic-kernel.git] / fs / ext4 / ext4.h
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/falloc.h>
36 #include <linux/percpu-rwsem.h>
37 #ifdef __KERNEL__
38 #include <linux/compat.h>
39 #endif
40
41 /*
42 * The fourth extended filesystem constants/structures
43 */
44
45 /*
46 * with AGGRESSIVE_CHECK allocator runs consistency checks over
47 * structures. these checks slow things down a lot
48 */
49 #define AGGRESSIVE_CHECK__
50
51 /*
52 * with DOUBLE_CHECK defined mballoc creates persistent in-core
53 * bitmaps, maintains and uses them to check for double allocations
54 */
55 #define DOUBLE_CHECK__
56
57 /*
58 * Define EXT4FS_DEBUG to produce debug messages
59 */
60 #undef EXT4FS_DEBUG
61
62 /*
63 * Debug code
64 */
65 #ifdef EXT4FS_DEBUG
66 #define ext4_debug(f, a...) \
67 do { \
68 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
69 __FILE__, __LINE__, __func__); \
70 printk(KERN_DEBUG f, ## a); \
71 } while (0)
72 #else
73 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
74 #endif
75
76 /*
77 * Turn on EXT_DEBUG to get lots of info about extents operations.
78 */
79 #define EXT_DEBUG__
80 #ifdef EXT_DEBUG
81 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
82 #else
83 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
84 #endif
85
86 /* data type for block offset of block group */
87 typedef int ext4_grpblk_t;
88
89 /* data type for filesystem-wide blocks number */
90 typedef unsigned long long ext4_fsblk_t;
91
92 /* data type for file logical block number */
93 typedef __u32 ext4_lblk_t;
94
95 /* data type for block group number */
96 typedef unsigned int ext4_group_t;
97
98 enum SHIFT_DIRECTION {
99 SHIFT_LEFT = 0,
100 SHIFT_RIGHT,
101 };
102
103 /*
104 * Flags used in mballoc's allocation_context flags field.
105 *
106 * Also used to show what's going on for debugging purposes when the
107 * flag field is exported via the traceport interface
108 */
109
110 /* prefer goal again. length */
111 #define EXT4_MB_HINT_MERGE 0x0001
112 /* blocks already reserved */
113 #define EXT4_MB_HINT_RESERVED 0x0002
114 /* metadata is being allocated */
115 #define EXT4_MB_HINT_METADATA 0x0004
116 /* first blocks in the file */
117 #define EXT4_MB_HINT_FIRST 0x0008
118 /* search for the best chunk */
119 #define EXT4_MB_HINT_BEST 0x0010
120 /* data is being allocated */
121 #define EXT4_MB_HINT_DATA 0x0020
122 /* don't preallocate (for tails) */
123 #define EXT4_MB_HINT_NOPREALLOC 0x0040
124 /* allocate for locality group */
125 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
126 /* allocate goal blocks or none */
127 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
128 /* goal is meaningful */
129 #define EXT4_MB_HINT_TRY_GOAL 0x0200
130 /* blocks already pre-reserved by delayed allocation */
131 #define EXT4_MB_DELALLOC_RESERVED 0x0400
132 /* We are doing stream allocation */
133 #define EXT4_MB_STREAM_ALLOC 0x0800
134 /* Use reserved root blocks if needed */
135 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
136 /* Use blocks from reserved pool */
137 #define EXT4_MB_USE_RESERVED 0x2000
138
139 struct ext4_allocation_request {
140 /* target inode for block we're allocating */
141 struct inode *inode;
142 /* how many blocks we want to allocate */
143 unsigned int len;
144 /* logical block in target inode */
145 ext4_lblk_t logical;
146 /* the closest logical allocated block to the left */
147 ext4_lblk_t lleft;
148 /* the closest logical allocated block to the right */
149 ext4_lblk_t lright;
150 /* phys. target (a hint) */
151 ext4_fsblk_t goal;
152 /* phys. block for the closest logical allocated block to the left */
153 ext4_fsblk_t pleft;
154 /* phys. block for the closest logical allocated block to the right */
155 ext4_fsblk_t pright;
156 /* flags. see above EXT4_MB_HINT_* */
157 unsigned int flags;
158 };
159
160 /*
161 * Logical to physical block mapping, used by ext4_map_blocks()
162 *
163 * This structure is used to pass requests into ext4_map_blocks() as
164 * well as to store the information returned by ext4_map_blocks(). It
165 * takes less room on the stack than a struct buffer_head.
166 */
167 #define EXT4_MAP_NEW (1 << BH_New)
168 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
169 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
170 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
171 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
172 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
173
174 struct ext4_map_blocks {
175 ext4_fsblk_t m_pblk;
176 ext4_lblk_t m_lblk;
177 unsigned int m_len;
178 unsigned int m_flags;
179 };
180
181 /*
182 * Flags for ext4_io_end->flags
183 */
184 #define EXT4_IO_END_UNWRITTEN 0x0001
185
186 /*
187 * For converting unwritten extents on a work queue. 'handle' is used for
188 * buffered writeback.
189 */
190 typedef struct ext4_io_end {
191 struct list_head list; /* per-file finished IO list */
192 handle_t *handle; /* handle reserved for extent
193 * conversion */
194 struct inode *inode; /* file being written to */
195 struct bio *bio; /* Linked list of completed
196 * bios covering the extent */
197 unsigned int flag; /* unwritten or not */
198 atomic_t count; /* reference counter */
199 loff_t offset; /* offset in the file */
200 ssize_t size; /* size of the extent */
201 } ext4_io_end_t;
202
203 struct ext4_io_submit {
204 struct writeback_control *io_wbc;
205 struct bio *io_bio;
206 ext4_io_end_t *io_end;
207 sector_t io_next_block;
208 };
209
210 /*
211 * Special inodes numbers
212 */
213 #define EXT4_BAD_INO 1 /* Bad blocks inode */
214 #define EXT4_ROOT_INO 2 /* Root inode */
215 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
216 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
217 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
218 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
219 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
220 #define EXT4_JOURNAL_INO 8 /* Journal inode */
221
222 /* First non-reserved inode for old ext4 filesystems */
223 #define EXT4_GOOD_OLD_FIRST_INO 11
224
225 /*
226 * Maximal count of links to a file
227 */
228 #define EXT4_LINK_MAX 65000
229
230 /*
231 * Macro-instructions used to manage several block sizes
232 */
233 #define EXT4_MIN_BLOCK_SIZE 1024
234 #define EXT4_MAX_BLOCK_SIZE 65536
235 #define EXT4_MIN_BLOCK_LOG_SIZE 10
236 #define EXT4_MAX_BLOCK_LOG_SIZE 16
237 #ifdef __KERNEL__
238 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
239 #else
240 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
241 #endif
242 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
243 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
244 EXT4_SB(s)->s_cluster_bits)
245 #ifdef __KERNEL__
246 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
247 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
248 #else
249 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
250 #endif
251 #ifdef __KERNEL__
252 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
253 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
254 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
255 #else
256 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
257 EXT4_GOOD_OLD_INODE_SIZE : \
258 (s)->s_inode_size)
259 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
260 EXT4_GOOD_OLD_FIRST_INO : \
261 (s)->s_first_ino)
262 #endif
263 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
264
265 /* Translate a block number to a cluster number */
266 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
267 /* Translate a cluster number to a block number */
268 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
269 /* Translate # of blks to # of clusters */
270 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
271 (sbi)->s_cluster_bits)
272 /* Mask out the low bits to get the starting block of the cluster */
273 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
274 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
275 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
276 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
277 /* Get the cluster offset */
278 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
279 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
280 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
281 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
282
283 /*
284 * Structure of a blocks group descriptor
285 */
286 struct ext4_group_desc
287 {
288 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
289 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
290 __le32 bg_inode_table_lo; /* Inodes table block */
291 __le16 bg_free_blocks_count_lo;/* Free blocks count */
292 __le16 bg_free_inodes_count_lo;/* Free inodes count */
293 __le16 bg_used_dirs_count_lo; /* Directories count */
294 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
295 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
296 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
297 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
298 __le16 bg_itable_unused_lo; /* Unused inodes count */
299 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
300 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
301 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
302 __le32 bg_inode_table_hi; /* Inodes table block MSB */
303 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
304 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
305 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
306 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
307 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
308 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
309 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
310 __u32 bg_reserved;
311 };
312
313 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
314 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
315 sizeof(__le16))
316 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
317 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
318 sizeof(__le16))
319
320 /*
321 * Structure of a flex block group info
322 */
323
324 struct flex_groups {
325 atomic64_t free_clusters;
326 atomic_t free_inodes;
327 atomic_t used_dirs;
328 };
329
330 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
331 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
332 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
333
334 /*
335 * Macro-instructions used to manage group descriptors
336 */
337 #define EXT4_MIN_DESC_SIZE 32
338 #define EXT4_MIN_DESC_SIZE_64BIT 64
339 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
340 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
341 #ifdef __KERNEL__
342 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
343 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
344 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
345 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
346 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
347 #else
348 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
349 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
350 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
351 #endif
352
353 /*
354 * Constants relative to the data blocks
355 */
356 #define EXT4_NDIR_BLOCKS 12
357 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
358 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
359 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
360 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
361
362 /*
363 * Inode flags
364 */
365 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
366 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
367 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
368 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
369 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
370 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
371 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
372 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
373 /* Reserved for compression usage... */
374 #define EXT4_DIRTY_FL 0x00000100
375 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
376 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
377 /* nb: was previously EXT2_ECOMPR_FL */
378 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
379 /* End compression flags --- maybe not all used */
380 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
381 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
382 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
383 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
384 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
385 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
386 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
387 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
388 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
389 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
390 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
391 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
392 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
393
394 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
395 #define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
396
397 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
398 EXT4_IMMUTABLE_FL | \
399 EXT4_APPEND_FL | \
400 EXT4_NODUMP_FL | \
401 EXT4_NOATIME_FL | \
402 EXT4_PROJINHERIT_FL)
403
404 /* Flags that should be inherited by new inodes from their parent. */
405 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
406 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
407 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
408 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
409 EXT4_PROJINHERIT_FL)
410
411 /* Flags that are appropriate for regular files (all but dir-specific ones). */
412 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
413
414 /* Flags that are appropriate for non-directories/regular files. */
415 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
416
417 /* Mask out flags that are inappropriate for the given type of inode. */
418 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
419 {
420 if (S_ISDIR(mode))
421 return flags;
422 else if (S_ISREG(mode))
423 return flags & EXT4_REG_FLMASK;
424 else
425 return flags & EXT4_OTHER_FLMASK;
426 }
427
428 /*
429 * Inode flags used for atomic set/get
430 */
431 enum {
432 EXT4_INODE_SECRM = 0, /* Secure deletion */
433 EXT4_INODE_UNRM = 1, /* Undelete */
434 EXT4_INODE_COMPR = 2, /* Compress file */
435 EXT4_INODE_SYNC = 3, /* Synchronous updates */
436 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
437 EXT4_INODE_APPEND = 5, /* writes to file may only append */
438 EXT4_INODE_NODUMP = 6, /* do not dump file */
439 EXT4_INODE_NOATIME = 7, /* do not update atime */
440 /* Reserved for compression usage... */
441 EXT4_INODE_DIRTY = 8,
442 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
443 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
444 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
445 /* End compression flags --- maybe not all used */
446 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
447 EXT4_INODE_IMAGIC = 13, /* AFS directory */
448 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
449 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
450 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
451 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
452 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
453 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
454 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
455 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
456 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
457 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
458 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
459 };
460
461 /*
462 * Since it's pretty easy to mix up bit numbers and hex values, we use a
463 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
464 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
465 * any extra space in the compiled kernel image, otherwise, the build will fail.
466 * It's important that these values are the same, since we are using
467 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
468 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
469 * values found in ext2, ext3 and ext4 filesystems, and of course the values
470 * defined in e2fsprogs.
471 *
472 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
473 */
474 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
475 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
476
477 static inline void ext4_check_flag_values(void)
478 {
479 CHECK_FLAG_VALUE(SECRM);
480 CHECK_FLAG_VALUE(UNRM);
481 CHECK_FLAG_VALUE(COMPR);
482 CHECK_FLAG_VALUE(SYNC);
483 CHECK_FLAG_VALUE(IMMUTABLE);
484 CHECK_FLAG_VALUE(APPEND);
485 CHECK_FLAG_VALUE(NODUMP);
486 CHECK_FLAG_VALUE(NOATIME);
487 CHECK_FLAG_VALUE(DIRTY);
488 CHECK_FLAG_VALUE(COMPRBLK);
489 CHECK_FLAG_VALUE(NOCOMPR);
490 CHECK_FLAG_VALUE(ENCRYPT);
491 CHECK_FLAG_VALUE(INDEX);
492 CHECK_FLAG_VALUE(IMAGIC);
493 CHECK_FLAG_VALUE(JOURNAL_DATA);
494 CHECK_FLAG_VALUE(NOTAIL);
495 CHECK_FLAG_VALUE(DIRSYNC);
496 CHECK_FLAG_VALUE(TOPDIR);
497 CHECK_FLAG_VALUE(HUGE_FILE);
498 CHECK_FLAG_VALUE(EXTENTS);
499 CHECK_FLAG_VALUE(EA_INODE);
500 CHECK_FLAG_VALUE(EOFBLOCKS);
501 CHECK_FLAG_VALUE(INLINE_DATA);
502 CHECK_FLAG_VALUE(PROJINHERIT);
503 CHECK_FLAG_VALUE(RESERVED);
504 }
505
506 /* Used to pass group descriptor data when online resize is done */
507 struct ext4_new_group_input {
508 __u32 group; /* Group number for this data */
509 __u64 block_bitmap; /* Absolute block number of block bitmap */
510 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
511 __u64 inode_table; /* Absolute block number of inode table start */
512 __u32 blocks_count; /* Total number of blocks in this group */
513 __u16 reserved_blocks; /* Number of reserved blocks in this group */
514 __u16 unused;
515 };
516
517 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
518 struct compat_ext4_new_group_input {
519 u32 group;
520 compat_u64 block_bitmap;
521 compat_u64 inode_bitmap;
522 compat_u64 inode_table;
523 u32 blocks_count;
524 u16 reserved_blocks;
525 u16 unused;
526 };
527 #endif
528
529 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
530 struct ext4_new_group_data {
531 __u32 group;
532 __u64 block_bitmap;
533 __u64 inode_bitmap;
534 __u64 inode_table;
535 __u32 blocks_count;
536 __u16 reserved_blocks;
537 __u16 unused;
538 __u32 free_blocks_count;
539 };
540
541 /* Indexes used to index group tables in ext4_new_group_data */
542 enum {
543 BLOCK_BITMAP = 0, /* block bitmap */
544 INODE_BITMAP, /* inode bitmap */
545 INODE_TABLE, /* inode tables */
546 GROUP_TABLE_COUNT,
547 };
548
549 /*
550 * Flags used by ext4_map_blocks()
551 */
552 /* Allocate any needed blocks and/or convert an unwritten
553 extent to be an initialized ext4 */
554 #define EXT4_GET_BLOCKS_CREATE 0x0001
555 /* Request the creation of an unwritten extent */
556 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
557 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
558 EXT4_GET_BLOCKS_CREATE)
559 /* Caller is from the delayed allocation writeout path
560 * finally doing the actual allocation of delayed blocks */
561 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
562 /* caller is from the direct IO path, request to creation of an
563 unwritten extents if not allocated, split the unwritten
564 extent if blocks has been preallocated already*/
565 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
566 #define EXT4_GET_BLOCKS_CONVERT 0x0010
567 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
568 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
569 /* Convert extent to initialized after IO complete */
570 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
571 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
572 /* Eventual metadata allocation (due to growing extent tree)
573 * should not fail, so try to use reserved blocks for that.*/
574 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
575 /* Don't normalize allocation size (used for fallocate) */
576 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
577 /* Request will not result in inode size update (user for fallocate) */
578 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
579 /* Convert written extents to unwritten */
580 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
581 /* Write zeros to newly created written extents */
582 #define EXT4_GET_BLOCKS_ZERO 0x0200
583 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
584 EXT4_GET_BLOCKS_ZERO)
585 /* Caller will submit data before dropping transaction handle. This
586 * allows jbd2 to avoid submitting data before commit. */
587 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
588
589 /*
590 * The bit position of these flags must not overlap with any of the
591 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
592 * read_extent_tree_block(), ext4_split_extent_at(),
593 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
594 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
595 * caching the extents when reading from the extent tree while a
596 * truncate or punch hole operation is in progress.
597 */
598 #define EXT4_EX_NOCACHE 0x40000000
599 #define EXT4_EX_FORCE_CACHE 0x20000000
600
601 /*
602 * Flags used by ext4_free_blocks
603 */
604 #define EXT4_FREE_BLOCKS_METADATA 0x0001
605 #define EXT4_FREE_BLOCKS_FORGET 0x0002
606 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
607 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
608 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
609 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
610
611 /* Encryption algorithms */
612 #define EXT4_ENCRYPTION_MODE_INVALID 0
613 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
614 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
615 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
616 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
617
618 #include "ext4_crypto.h"
619
620 /*
621 * ioctl commands
622 */
623 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
624 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
625 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
626 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
627 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
628 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
629 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
630 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
631 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
632 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
633 #define EXT4_IOC_MIGRATE _IO('f', 9)
634 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
635 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
636 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
637 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
638 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
639 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
640 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
641 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
642 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
643 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
644
645 #ifndef FS_IOC_FSGETXATTR
646 /* Until the uapi changes get merged for project quota... */
647
648 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
649 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
650
651 /*
652 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
653 */
654 struct fsxattr {
655 __u32 fsx_xflags; /* xflags field value (get/set) */
656 __u32 fsx_extsize; /* extsize field value (get/set)*/
657 __u32 fsx_nextents; /* nextents field value (get) */
658 __u32 fsx_projid; /* project identifier (get/set) */
659 unsigned char fsx_pad[12];
660 };
661
662 /*
663 * Flags for the fsx_xflags field
664 */
665 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
666 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
667 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
668 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
669 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
670 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
671 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
672 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
673 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
674 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
675 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
676 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
677 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
678 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
679 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
680 #endif /* !defined(FS_IOC_FSGETXATTR) */
681
682 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
683 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
684
685 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
686 /*
687 * ioctl commands in 32 bit emulation
688 */
689 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
690 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
691 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
692 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
693 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
694 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
695 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
696 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
697 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
698 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
699 #endif
700
701 /* Max physical block we can address w/o extents */
702 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
703
704 /*
705 * Structure of an inode on the disk
706 */
707 struct ext4_inode {
708 __le16 i_mode; /* File mode */
709 __le16 i_uid; /* Low 16 bits of Owner Uid */
710 __le32 i_size_lo; /* Size in bytes */
711 __le32 i_atime; /* Access time */
712 __le32 i_ctime; /* Inode Change time */
713 __le32 i_mtime; /* Modification time */
714 __le32 i_dtime; /* Deletion Time */
715 __le16 i_gid; /* Low 16 bits of Group Id */
716 __le16 i_links_count; /* Links count */
717 __le32 i_blocks_lo; /* Blocks count */
718 __le32 i_flags; /* File flags */
719 union {
720 struct {
721 __le32 l_i_version;
722 } linux1;
723 struct {
724 __u32 h_i_translator;
725 } hurd1;
726 struct {
727 __u32 m_i_reserved1;
728 } masix1;
729 } osd1; /* OS dependent 1 */
730 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
731 __le32 i_generation; /* File version (for NFS) */
732 __le32 i_file_acl_lo; /* File ACL */
733 __le32 i_size_high;
734 __le32 i_obso_faddr; /* Obsoleted fragment address */
735 union {
736 struct {
737 __le16 l_i_blocks_high; /* were l_i_reserved1 */
738 __le16 l_i_file_acl_high;
739 __le16 l_i_uid_high; /* these 2 fields */
740 __le16 l_i_gid_high; /* were reserved2[0] */
741 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
742 __le16 l_i_reserved;
743 } linux2;
744 struct {
745 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
746 __u16 h_i_mode_high;
747 __u16 h_i_uid_high;
748 __u16 h_i_gid_high;
749 __u32 h_i_author;
750 } hurd2;
751 struct {
752 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
753 __le16 m_i_file_acl_high;
754 __u32 m_i_reserved2[2];
755 } masix2;
756 } osd2; /* OS dependent 2 */
757 __le16 i_extra_isize;
758 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
759 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
760 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
761 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
762 __le32 i_crtime; /* File Creation time */
763 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
764 __le32 i_version_hi; /* high 32 bits for 64-bit version */
765 __le32 i_projid; /* Project ID */
766 };
767
768 struct move_extent {
769 __u32 reserved; /* should be zero */
770 __u32 donor_fd; /* donor file descriptor */
771 __u64 orig_start; /* logical start offset in block for orig */
772 __u64 donor_start; /* logical start offset in block for donor */
773 __u64 len; /* block length to be moved */
774 __u64 moved_len; /* moved block length */
775 };
776
777 #define EXT4_EPOCH_BITS 2
778 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
779 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
780
781 /*
782 * Extended fields will fit into an inode if the filesystem was formatted
783 * with large inodes (-I 256 or larger) and there are not currently any EAs
784 * consuming all of the available space. For new inodes we always reserve
785 * enough space for the kernel's known extended fields, but for inodes
786 * created with an old kernel this might not have been the case. None of
787 * the extended inode fields is critical for correct filesystem operation.
788 * This macro checks if a certain field fits in the inode. Note that
789 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
790 */
791 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
792 ((offsetof(typeof(*ext4_inode), field) + \
793 sizeof((ext4_inode)->field)) \
794 <= (EXT4_GOOD_OLD_INODE_SIZE + \
795 (einode)->i_extra_isize)) \
796
797 /*
798 * We use an encoding that preserves the times for extra epoch "00":
799 *
800 * extra msb of adjust for signed
801 * epoch 32-bit 32-bit tv_sec to
802 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
803 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
804 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
805 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
806 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
807 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
808 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
809 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
810 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
811 *
812 * Note that previous versions of the kernel on 64-bit systems would
813 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
814 * 1970. e2fsck will correct this, assuming that it is run on the
815 * affected filesystem before 2242.
816 */
817
818 static inline __le32 ext4_encode_extra_time(struct timespec *time)
819 {
820 u32 extra = sizeof(time->tv_sec) > 4 ?
821 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
822 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
823 }
824
825 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
826 {
827 if (unlikely(sizeof(time->tv_sec) > 4 &&
828 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
829 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
830 /* Handle legacy encoding of pre-1970 dates with epoch
831 * bits 1,1. We assume that by kernel version 4.20,
832 * everyone will have run fsck over the affected
833 * filesystems to correct the problem. (This
834 * backwards compatibility may be removed before this
835 * time, at the discretion of the ext4 developers.)
836 */
837 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
838 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
839 extra_bits = 0;
840 time->tv_sec += extra_bits << 32;
841 #else
842 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
843 #endif
844 }
845 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
846 }
847
848 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
849 do { \
850 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
851 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
852 (raw_inode)->xtime ## _extra = \
853 ext4_encode_extra_time(&(inode)->xtime); \
854 } while (0)
855
856 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
857 do { \
858 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
859 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
860 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
861 (raw_inode)->xtime ## _extra = \
862 ext4_encode_extra_time(&(einode)->xtime); \
863 } while (0)
864
865 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
866 do { \
867 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
868 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
869 ext4_decode_extra_time(&(inode)->xtime, \
870 raw_inode->xtime ## _extra); \
871 else \
872 (inode)->xtime.tv_nsec = 0; \
873 } while (0)
874
875 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
876 do { \
877 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
878 (einode)->xtime.tv_sec = \
879 (signed)le32_to_cpu((raw_inode)->xtime); \
880 else \
881 (einode)->xtime.tv_sec = 0; \
882 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
883 ext4_decode_extra_time(&(einode)->xtime, \
884 raw_inode->xtime ## _extra); \
885 else \
886 (einode)->xtime.tv_nsec = 0; \
887 } while (0)
888
889 #define i_disk_version osd1.linux1.l_i_version
890
891 #if defined(__KERNEL__) || defined(__linux__)
892 #define i_reserved1 osd1.linux1.l_i_reserved1
893 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
894 #define i_blocks_high osd2.linux2.l_i_blocks_high
895 #define i_uid_low i_uid
896 #define i_gid_low i_gid
897 #define i_uid_high osd2.linux2.l_i_uid_high
898 #define i_gid_high osd2.linux2.l_i_gid_high
899 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
900
901 #elif defined(__GNU__)
902
903 #define i_translator osd1.hurd1.h_i_translator
904 #define i_uid_high osd2.hurd2.h_i_uid_high
905 #define i_gid_high osd2.hurd2.h_i_gid_high
906 #define i_author osd2.hurd2.h_i_author
907
908 #elif defined(__masix__)
909
910 #define i_reserved1 osd1.masix1.m_i_reserved1
911 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
912 #define i_reserved2 osd2.masix2.m_i_reserved2
913
914 #endif /* defined(__KERNEL__) || defined(__linux__) */
915
916 #include "extents_status.h"
917
918 /*
919 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
920 *
921 * These are needed to avoid lockdep false positives when we need to
922 * allocate blocks to the quota inode during ext4_map_blocks(), while
923 * holding i_data_sem for a normal (non-quota) inode. Since we don't
924 * do quota tracking for the quota inode, this avoids deadlock (as
925 * well as infinite recursion, since it isn't turtles all the way
926 * down...)
927 *
928 * I_DATA_SEM_NORMAL - Used for most inodes
929 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
930 * where the second inode has larger inode number
931 * than the first
932 * I_DATA_SEM_QUOTA - Used for quota inodes only
933 */
934 enum {
935 I_DATA_SEM_NORMAL = 0,
936 I_DATA_SEM_OTHER,
937 I_DATA_SEM_QUOTA,
938 };
939
940
941 /*
942 * fourth extended file system inode data in memory
943 */
944 struct ext4_inode_info {
945 __le32 i_data[15]; /* unconverted */
946 __u32 i_dtime;
947 ext4_fsblk_t i_file_acl;
948
949 /*
950 * i_block_group is the number of the block group which contains
951 * this file's inode. Constant across the lifetime of the inode,
952 * it is ued for making block allocation decisions - we try to
953 * place a file's data blocks near its inode block, and new inodes
954 * near to their parent directory's inode.
955 */
956 ext4_group_t i_block_group;
957 ext4_lblk_t i_dir_start_lookup;
958 #if (BITS_PER_LONG < 64)
959 unsigned long i_state_flags; /* Dynamic state flags */
960 #endif
961 unsigned long i_flags;
962
963 /*
964 * Extended attributes can be read independently of the main file
965 * data. Taking i_mutex even when reading would cause contention
966 * between readers of EAs and writers of regular file data, so
967 * instead we synchronize on xattr_sem when reading or changing
968 * EAs.
969 */
970 struct rw_semaphore xattr_sem;
971
972 struct list_head i_orphan; /* unlinked but open inodes */
973
974 /*
975 * i_disksize keeps track of what the inode size is ON DISK, not
976 * in memory. During truncate, i_size is set to the new size by
977 * the VFS prior to calling ext4_truncate(), but the filesystem won't
978 * set i_disksize to 0 until the truncate is actually under way.
979 *
980 * The intent is that i_disksize always represents the blocks which
981 * are used by this file. This allows recovery to restart truncate
982 * on orphans if we crash during truncate. We actually write i_disksize
983 * into the on-disk inode when writing inodes out, instead of i_size.
984 *
985 * The only time when i_disksize and i_size may be different is when
986 * a truncate is in progress. The only things which change i_disksize
987 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
988 */
989 loff_t i_disksize;
990
991 /*
992 * i_data_sem is for serialising ext4_truncate() against
993 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
994 * data tree are chopped off during truncate. We can't do that in
995 * ext4 because whenever we perform intermediate commits during
996 * truncate, the inode and all the metadata blocks *must* be in a
997 * consistent state which allows truncation of the orphans to restart
998 * during recovery. Hence we must fix the get_block-vs-truncate race
999 * by other means, so we have i_data_sem.
1000 */
1001 struct rw_semaphore i_data_sem;
1002 /*
1003 * i_mmap_sem is for serializing page faults with truncate / punch hole
1004 * operations. We have to make sure that new page cannot be faulted in
1005 * a section of the inode that is being punched. We cannot easily use
1006 * i_data_sem for this since we need protection for the whole punch
1007 * operation and i_data_sem ranks below transaction start so we have
1008 * to occasionally drop it.
1009 */
1010 struct rw_semaphore i_mmap_sem;
1011 struct inode vfs_inode;
1012 struct jbd2_inode *jinode;
1013
1014 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1015
1016 /*
1017 * File creation time. Its function is same as that of
1018 * struct timespec i_{a,c,m}time in the generic inode.
1019 */
1020 struct timespec i_crtime;
1021
1022 /* mballoc */
1023 struct list_head i_prealloc_list;
1024 spinlock_t i_prealloc_lock;
1025
1026 /* extents status tree */
1027 struct ext4_es_tree i_es_tree;
1028 rwlock_t i_es_lock;
1029 struct list_head i_es_list;
1030 unsigned int i_es_all_nr; /* protected by i_es_lock */
1031 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1032 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1033 extents to shrink. Protected by
1034 i_es_lock */
1035
1036 /* ialloc */
1037 ext4_group_t i_last_alloc_group;
1038
1039 /* allocation reservation info for delalloc */
1040 /* In case of bigalloc, these refer to clusters rather than blocks */
1041 unsigned int i_reserved_data_blocks;
1042 unsigned int i_reserved_meta_blocks;
1043 unsigned int i_allocated_meta_blocks;
1044 ext4_lblk_t i_da_metadata_calc_last_lblock;
1045 int i_da_metadata_calc_len;
1046
1047 /* on-disk additional length */
1048 __u16 i_extra_isize;
1049
1050 /* Indicate the inline data space. */
1051 u16 i_inline_off;
1052 u16 i_inline_size;
1053
1054 #ifdef CONFIG_QUOTA
1055 /* quota space reservation, managed internally by quota code */
1056 qsize_t i_reserved_quota;
1057 #endif
1058
1059 /* Lock protecting lists below */
1060 spinlock_t i_completed_io_lock;
1061 /*
1062 * Completed IOs that need unwritten extents handling and have
1063 * transaction reserved
1064 */
1065 struct list_head i_rsv_conversion_list;
1066 struct work_struct i_rsv_conversion_work;
1067 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1068
1069 spinlock_t i_block_reservation_lock;
1070
1071 /*
1072 * Transactions that contain inode's metadata needed to complete
1073 * fsync and fdatasync, respectively.
1074 */
1075 tid_t i_sync_tid;
1076 tid_t i_datasync_tid;
1077
1078 #ifdef CONFIG_QUOTA
1079 struct dquot *i_dquot[MAXQUOTAS];
1080 #endif
1081
1082 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1083 __u32 i_csum_seed;
1084
1085 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1086 /* Encryption params */
1087 struct ext4_crypt_info *i_crypt_info;
1088 #endif
1089 kprojid_t i_projid;
1090 };
1091
1092 /*
1093 * File system states
1094 */
1095 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1096 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1097 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1098
1099 /*
1100 * Misc. filesystem flags
1101 */
1102 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1103 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1104 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1105
1106 /*
1107 * Mount flags set via mount options or defaults
1108 */
1109 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1110 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1111 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1112 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1113 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1114 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1115 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1116 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1117 #ifdef CONFIG_FS_DAX
1118 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1119 #else
1120 #define EXT4_MOUNT_DAX 0
1121 #endif
1122 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1123 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1124 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1125 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1126 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1127 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1128 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1129 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1130 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1131 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1132 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1133 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1134 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1135 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1136 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1137 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1138 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1139 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1140 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1141 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1142 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1143
1144 /*
1145 * Mount flags set either automatically (could not be set by mount option)
1146 * based on per file system feature or property or in special cases such as
1147 * distinguishing between explicit mount option definition and default.
1148 */
1149 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1150 specified delalloc */
1151 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1152 size of blocksize * 8
1153 blocks */
1154 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1155 file systems */
1156
1157 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1158 specified journal checksum */
1159
1160 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1161 ~EXT4_MOUNT_##opt
1162 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1163 EXT4_MOUNT_##opt
1164 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1165 EXT4_MOUNT_##opt)
1166
1167 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1168 ~EXT4_MOUNT2_##opt
1169 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1170 EXT4_MOUNT2_##opt
1171 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1172 EXT4_MOUNT2_##opt)
1173
1174 #define ext4_test_and_set_bit __test_and_set_bit_le
1175 #define ext4_set_bit __set_bit_le
1176 #define ext4_set_bit_atomic ext2_set_bit_atomic
1177 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1178 #define ext4_clear_bit __clear_bit_le
1179 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1180 #define ext4_test_bit test_bit_le
1181 #define ext4_find_next_zero_bit find_next_zero_bit_le
1182 #define ext4_find_next_bit find_next_bit_le
1183
1184 extern void ext4_set_bits(void *bm, int cur, int len);
1185
1186 /*
1187 * Maximal mount counts between two filesystem checks
1188 */
1189 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1190 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1191
1192 /*
1193 * Behaviour when detecting errors
1194 */
1195 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1196 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1197 #define EXT4_ERRORS_PANIC 3 /* Panic */
1198 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1199
1200 /* Metadata checksum algorithm codes */
1201 #define EXT4_CRC32C_CHKSUM 1
1202
1203 /*
1204 * Structure of the super block
1205 */
1206 struct ext4_super_block {
1207 /*00*/ __le32 s_inodes_count; /* Inodes count */
1208 __le32 s_blocks_count_lo; /* Blocks count */
1209 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1210 __le32 s_free_blocks_count_lo; /* Free blocks count */
1211 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1212 __le32 s_first_data_block; /* First Data Block */
1213 __le32 s_log_block_size; /* Block size */
1214 __le32 s_log_cluster_size; /* Allocation cluster size */
1215 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1216 __le32 s_clusters_per_group; /* # Clusters per group */
1217 __le32 s_inodes_per_group; /* # Inodes per group */
1218 __le32 s_mtime; /* Mount time */
1219 /*30*/ __le32 s_wtime; /* Write time */
1220 __le16 s_mnt_count; /* Mount count */
1221 __le16 s_max_mnt_count; /* Maximal mount count */
1222 __le16 s_magic; /* Magic signature */
1223 __le16 s_state; /* File system state */
1224 __le16 s_errors; /* Behaviour when detecting errors */
1225 __le16 s_minor_rev_level; /* minor revision level */
1226 /*40*/ __le32 s_lastcheck; /* time of last check */
1227 __le32 s_checkinterval; /* max. time between checks */
1228 __le32 s_creator_os; /* OS */
1229 __le32 s_rev_level; /* Revision level */
1230 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1231 __le16 s_def_resgid; /* Default gid for reserved blocks */
1232 /*
1233 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1234 *
1235 * Note: the difference between the compatible feature set and
1236 * the incompatible feature set is that if there is a bit set
1237 * in the incompatible feature set that the kernel doesn't
1238 * know about, it should refuse to mount the filesystem.
1239 *
1240 * e2fsck's requirements are more strict; if it doesn't know
1241 * about a feature in either the compatible or incompatible
1242 * feature set, it must abort and not try to meddle with
1243 * things it doesn't understand...
1244 */
1245 __le32 s_first_ino; /* First non-reserved inode */
1246 __le16 s_inode_size; /* size of inode structure */
1247 __le16 s_block_group_nr; /* block group # of this superblock */
1248 __le32 s_feature_compat; /* compatible feature set */
1249 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1250 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1251 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1252 /*78*/ char s_volume_name[16]; /* volume name */
1253 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1254 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1255 /*
1256 * Performance hints. Directory preallocation should only
1257 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1258 */
1259 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1260 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1261 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1262 /*
1263 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1264 */
1265 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1266 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1267 __le32 s_journal_dev; /* device number of journal file */
1268 __le32 s_last_orphan; /* start of list of inodes to delete */
1269 __le32 s_hash_seed[4]; /* HTREE hash seed */
1270 __u8 s_def_hash_version; /* Default hash version to use */
1271 __u8 s_jnl_backup_type;
1272 __le16 s_desc_size; /* size of group descriptor */
1273 /*100*/ __le32 s_default_mount_opts;
1274 __le32 s_first_meta_bg; /* First metablock block group */
1275 __le32 s_mkfs_time; /* When the filesystem was created */
1276 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1277 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1278 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1279 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1280 __le32 s_free_blocks_count_hi; /* Free blocks count */
1281 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1282 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1283 __le32 s_flags; /* Miscellaneous flags */
1284 __le16 s_raid_stride; /* RAID stride */
1285 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1286 __le64 s_mmp_block; /* Block for multi-mount protection */
1287 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1288 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1289 __u8 s_checksum_type; /* metadata checksum algorithm used */
1290 __u8 s_encryption_level; /* versioning level for encryption */
1291 __u8 s_reserved_pad; /* Padding to next 32bits */
1292 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1293 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1294 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1295 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1296 snapshot's future use */
1297 __le32 s_snapshot_list; /* inode number of the head of the
1298 on-disk snapshot list */
1299 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1300 __le32 s_error_count; /* number of fs errors */
1301 __le32 s_first_error_time; /* first time an error happened */
1302 __le32 s_first_error_ino; /* inode involved in first error */
1303 __le64 s_first_error_block; /* block involved of first error */
1304 __u8 s_first_error_func[32]; /* function where the error happened */
1305 __le32 s_first_error_line; /* line number where error happened */
1306 __le32 s_last_error_time; /* most recent time of an error */
1307 __le32 s_last_error_ino; /* inode involved in last error */
1308 __le32 s_last_error_line; /* line number where error happened */
1309 __le64 s_last_error_block; /* block involved of last error */
1310 __u8 s_last_error_func[32]; /* function where the error happened */
1311 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1312 __u8 s_mount_opts[64];
1313 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1314 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1315 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1316 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1317 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1318 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1319 __le32 s_lpf_ino; /* Location of the lost+found inode */
1320 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1321 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1322 __le32 s_reserved[98]; /* Padding to the end of the block */
1323 __le32 s_checksum; /* crc32c(superblock) */
1324 };
1325
1326 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1327
1328 #ifdef __KERNEL__
1329
1330 /*
1331 * run-time mount flags
1332 */
1333 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1334 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1335 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1336
1337 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1338 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1339 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1340 #else
1341 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1342 #endif
1343
1344 /* Number of quota types we support */
1345 #define EXT4_MAXQUOTAS 3
1346
1347 /*
1348 * fourth extended-fs super-block data in memory
1349 */
1350 struct ext4_sb_info {
1351 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1352 unsigned long s_inodes_per_block;/* Number of inodes per block */
1353 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1354 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1355 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1356 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1357 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1358 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1359 ext4_group_t s_groups_count; /* Number of groups in the fs */
1360 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1361 unsigned long s_overhead; /* # of fs overhead clusters */
1362 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1363 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1364 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1365 struct buffer_head * s_sbh; /* Buffer containing the super block */
1366 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1367 struct buffer_head **s_group_desc;
1368 unsigned int s_mount_opt;
1369 unsigned int s_mount_opt2;
1370 unsigned int s_mount_flags;
1371 unsigned int s_def_mount_opt;
1372 ext4_fsblk_t s_sb_block;
1373 atomic64_t s_resv_clusters;
1374 kuid_t s_resuid;
1375 kgid_t s_resgid;
1376 unsigned short s_mount_state;
1377 unsigned short s_pad;
1378 int s_addr_per_block_bits;
1379 int s_desc_per_block_bits;
1380 int s_inode_size;
1381 int s_first_ino;
1382 unsigned int s_inode_readahead_blks;
1383 unsigned int s_inode_goal;
1384 spinlock_t s_next_gen_lock;
1385 u32 s_next_generation;
1386 u32 s_hash_seed[4];
1387 int s_def_hash_version;
1388 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1389 struct percpu_counter s_freeclusters_counter;
1390 struct percpu_counter s_freeinodes_counter;
1391 struct percpu_counter s_dirs_counter;
1392 struct percpu_counter s_dirtyclusters_counter;
1393 struct blockgroup_lock *s_blockgroup_lock;
1394 struct proc_dir_entry *s_proc;
1395 struct kobject s_kobj;
1396 struct completion s_kobj_unregister;
1397 struct super_block *s_sb;
1398
1399 /* Journaling */
1400 struct journal_s *s_journal;
1401 struct list_head s_orphan;
1402 struct mutex s_orphan_lock;
1403 unsigned long s_resize_flags; /* Flags indicating if there
1404 is a resizer */
1405 unsigned long s_commit_interval;
1406 u32 s_max_batch_time;
1407 u32 s_min_batch_time;
1408 struct block_device *journal_bdev;
1409 #ifdef CONFIG_QUOTA
1410 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1411 int s_jquota_fmt; /* Format of quota to use */
1412 #endif
1413 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1414 struct rb_root system_blks;
1415
1416 #ifdef EXTENTS_STATS
1417 /* ext4 extents stats */
1418 unsigned long s_ext_min;
1419 unsigned long s_ext_max;
1420 unsigned long s_depth_max;
1421 spinlock_t s_ext_stats_lock;
1422 unsigned long s_ext_blocks;
1423 unsigned long s_ext_extents;
1424 #endif
1425
1426 /* for buddy allocator */
1427 struct ext4_group_info ***s_group_info;
1428 struct inode *s_buddy_cache;
1429 spinlock_t s_md_lock;
1430 unsigned short *s_mb_offsets;
1431 unsigned int *s_mb_maxs;
1432 unsigned int s_group_info_size;
1433
1434 /* tunables */
1435 unsigned long s_stripe;
1436 unsigned int s_mb_stream_request;
1437 unsigned int s_mb_max_to_scan;
1438 unsigned int s_mb_min_to_scan;
1439 unsigned int s_mb_stats;
1440 unsigned int s_mb_order2_reqs;
1441 unsigned int s_mb_group_prealloc;
1442 unsigned int s_max_dir_size_kb;
1443 /* where last allocation was done - for stream allocation */
1444 unsigned long s_mb_last_group;
1445 unsigned long s_mb_last_start;
1446
1447 /* stats for buddy allocator */
1448 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1449 atomic_t s_bal_success; /* we found long enough chunks */
1450 atomic_t s_bal_allocated; /* in blocks */
1451 atomic_t s_bal_ex_scanned; /* total extents scanned */
1452 atomic_t s_bal_goals; /* goal hits */
1453 atomic_t s_bal_breaks; /* too long searches */
1454 atomic_t s_bal_2orders; /* 2^order hits */
1455 spinlock_t s_bal_lock;
1456 unsigned long s_mb_buddies_generated;
1457 unsigned long long s_mb_generation_time;
1458 atomic_t s_mb_lost_chunks;
1459 atomic_t s_mb_preallocated;
1460 atomic_t s_mb_discarded;
1461 atomic_t s_lock_busy;
1462
1463 /* locality groups */
1464 struct ext4_locality_group __percpu *s_locality_groups;
1465
1466 /* for write statistics */
1467 unsigned long s_sectors_written_start;
1468 u64 s_kbytes_written;
1469
1470 /* the size of zero-out chunk */
1471 unsigned int s_extent_max_zeroout_kb;
1472
1473 unsigned int s_log_groups_per_flex;
1474 struct flex_groups *s_flex_groups;
1475 ext4_group_t s_flex_groups_allocated;
1476
1477 /* workqueue for reserved extent conversions (buffered io) */
1478 struct workqueue_struct *rsv_conversion_wq;
1479
1480 /* timer for periodic error stats printing */
1481 struct timer_list s_err_report;
1482
1483 /* Lazy inode table initialization info */
1484 struct ext4_li_request *s_li_request;
1485 /* Wait multiplier for lazy initialization thread */
1486 unsigned int s_li_wait_mult;
1487
1488 /* Kernel thread for multiple mount protection */
1489 struct task_struct *s_mmp_tsk;
1490
1491 /* record the last minlen when FITRIM is called. */
1492 atomic_t s_last_trim_minblks;
1493
1494 /* Reference to checksum algorithm driver via cryptoapi */
1495 struct crypto_shash *s_chksum_driver;
1496
1497 /* Precomputed FS UUID checksum for seeding other checksums */
1498 __u32 s_csum_seed;
1499
1500 /* Reclaim extents from extent status tree */
1501 struct shrinker s_es_shrinker;
1502 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1503 long s_es_nr_inode;
1504 struct ext4_es_stats s_es_stats;
1505 struct mb_cache *s_mb_cache;
1506 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1507
1508 /* Ratelimit ext4 messages. */
1509 struct ratelimit_state s_err_ratelimit_state;
1510 struct ratelimit_state s_warning_ratelimit_state;
1511 struct ratelimit_state s_msg_ratelimit_state;
1512
1513 /* Barrier between changing inodes' journal flags and writepages ops. */
1514 struct percpu_rw_semaphore s_journal_flag_rwsem;
1515 };
1516
1517 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1518 {
1519 return sb->s_fs_info;
1520 }
1521 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1522 {
1523 return container_of(inode, struct ext4_inode_info, vfs_inode);
1524 }
1525
1526 static inline struct timespec ext4_current_time(struct inode *inode)
1527 {
1528 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1529 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1530 }
1531
1532 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1533 {
1534 return ino == EXT4_ROOT_INO ||
1535 ino == EXT4_USR_QUOTA_INO ||
1536 ino == EXT4_GRP_QUOTA_INO ||
1537 ino == EXT4_BOOT_LOADER_INO ||
1538 ino == EXT4_JOURNAL_INO ||
1539 ino == EXT4_RESIZE_INO ||
1540 (ino >= EXT4_FIRST_INO(sb) &&
1541 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1542 }
1543
1544 /*
1545 * Inode dynamic state flags
1546 */
1547 enum {
1548 EXT4_STATE_JDATA, /* journaled data exists */
1549 EXT4_STATE_NEW, /* inode is newly created */
1550 EXT4_STATE_XATTR, /* has in-inode xattrs */
1551 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1552 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1553 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1554 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1555 EXT4_STATE_NEWENTRY, /* File just added to dir */
1556 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1557 nolocking */
1558 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1559 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1560 };
1561
1562 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1563 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1564 { \
1565 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1566 } \
1567 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1568 { \
1569 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1570 } \
1571 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1572 { \
1573 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1574 }
1575
1576 /* Add these declarations here only so that these functions can be
1577 * found by name. Otherwise, they are very hard to locate. */
1578 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1579 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1580 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1581 EXT4_INODE_BIT_FNS(flag, flags, 0)
1582
1583 /* Add these declarations here only so that these functions can be
1584 * found by name. Otherwise, they are very hard to locate. */
1585 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1586 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1587 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1588 #if (BITS_PER_LONG < 64)
1589 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1590
1591 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1592 {
1593 (ei)->i_state_flags = 0;
1594 }
1595 #else
1596 EXT4_INODE_BIT_FNS(state, flags, 32)
1597
1598 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1599 {
1600 /* We depend on the fact that callers will set i_flags */
1601 }
1602 #endif
1603 #else
1604 /* Assume that user mode programs are passing in an ext4fs superblock, not
1605 * a kernel struct super_block. This will allow us to call the feature-test
1606 * macros from user land. */
1607 #define EXT4_SB(sb) (sb)
1608 #endif
1609
1610 /*
1611 * Returns true if the inode is inode is encrypted
1612 */
1613 static inline int ext4_encrypted_inode(struct inode *inode)
1614 {
1615 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1616 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1617 #else
1618 return 0;
1619 #endif
1620 }
1621
1622 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1623
1624 /*
1625 * Codes for operating systems
1626 */
1627 #define EXT4_OS_LINUX 0
1628 #define EXT4_OS_HURD 1
1629 #define EXT4_OS_MASIX 2
1630 #define EXT4_OS_FREEBSD 3
1631 #define EXT4_OS_LITES 4
1632
1633 /*
1634 * Revision levels
1635 */
1636 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1637 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1638
1639 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1640 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1641
1642 #define EXT4_GOOD_OLD_INODE_SIZE 128
1643
1644 /*
1645 * Feature set definitions
1646 */
1647
1648 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1649 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1650 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1651 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1652 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1653 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1654 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1655 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1656 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1657 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1658 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1659 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1660 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1661 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1662 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1663 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1664 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1665 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1666 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1667
1668 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1669 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1670 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1671 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1672 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1673 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1674 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1675
1676 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1677 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1678 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1679 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1680 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1681 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1682 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1683 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1684 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1685 /*
1686 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1687 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1688 * all other data structures' checksums. However, the METADATA_CSUM and
1689 * GDT_CSUM bits are mutually exclusive.
1690 */
1691 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1692 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1693 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1694
1695 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1696 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1697 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1698 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1699 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1700 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1701 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1702 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1703 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1704 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1705 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1706 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1707 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1708 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1709 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1710
1711 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1712 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1713 { \
1714 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1715 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1716 } \
1717 static inline void ext4_set_feature_##name(struct super_block *sb) \
1718 { \
1719 EXT4_SB(sb)->s_es->s_feature_compat |= \
1720 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1721 } \
1722 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1723 { \
1724 EXT4_SB(sb)->s_es->s_feature_compat &= \
1725 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1726 }
1727
1728 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1729 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1730 { \
1731 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1732 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1733 } \
1734 static inline void ext4_set_feature_##name(struct super_block *sb) \
1735 { \
1736 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1737 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1738 } \
1739 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1740 { \
1741 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1742 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1743 }
1744
1745 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1746 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1747 { \
1748 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1749 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1750 } \
1751 static inline void ext4_set_feature_##name(struct super_block *sb) \
1752 { \
1753 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1754 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1755 } \
1756 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1757 { \
1758 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1759 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1760 }
1761
1762 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1763 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1764 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1765 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1766 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1767 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1768 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1769
1770 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1771 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1772 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1773 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1774 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1775 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1776 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1777 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1778 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1779 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1780 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1781 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1782
1783 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1784 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1785 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1786 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1787 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1788 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1789 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1790 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1791 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1792 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1793 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1794 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1795 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1796 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1797 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1798
1799 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1800 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1801 EXT4_FEATURE_INCOMPAT_META_BG)
1802 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1803 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1804 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1805
1806 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1807 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1808 EXT4_FEATURE_INCOMPAT_RECOVER| \
1809 EXT4_FEATURE_INCOMPAT_META_BG)
1810 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1811 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1812 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1813
1814 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1815 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1816 EXT4_FEATURE_INCOMPAT_RECOVER| \
1817 EXT4_FEATURE_INCOMPAT_META_BG| \
1818 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1819 EXT4_FEATURE_INCOMPAT_64BIT| \
1820 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1821 EXT4_FEATURE_INCOMPAT_MMP | \
1822 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1823 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1824 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1825 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1826 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1827 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1828 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1829 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1830 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1831 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1832 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1833 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1834 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1835 EXT4_FEATURE_RO_COMPAT_PROJECT)
1836
1837 #define EXTN_FEATURE_FUNCS(ver) \
1838 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1839 { \
1840 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1841 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1842 } \
1843 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1844 { \
1845 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1846 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1847 } \
1848 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1849 { \
1850 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1851 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1852 }
1853
1854 EXTN_FEATURE_FUNCS(2)
1855 EXTN_FEATURE_FUNCS(3)
1856 EXTN_FEATURE_FUNCS(4)
1857
1858 static inline bool ext4_has_compat_features(struct super_block *sb)
1859 {
1860 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1861 }
1862 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1863 {
1864 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1865 }
1866 static inline bool ext4_has_incompat_features(struct super_block *sb)
1867 {
1868 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1869 }
1870
1871 /*
1872 * Default values for user and/or group using reserved blocks
1873 */
1874 #define EXT4_DEF_RESUID 0
1875 #define EXT4_DEF_RESGID 0
1876
1877 /*
1878 * Default project ID
1879 */
1880 #define EXT4_DEF_PROJID 0
1881
1882 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1883
1884 /*
1885 * Default mount options
1886 */
1887 #define EXT4_DEFM_DEBUG 0x0001
1888 #define EXT4_DEFM_BSDGROUPS 0x0002
1889 #define EXT4_DEFM_XATTR_USER 0x0004
1890 #define EXT4_DEFM_ACL 0x0008
1891 #define EXT4_DEFM_UID16 0x0010
1892 #define EXT4_DEFM_JMODE 0x0060
1893 #define EXT4_DEFM_JMODE_DATA 0x0020
1894 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1895 #define EXT4_DEFM_JMODE_WBACK 0x0060
1896 #define EXT4_DEFM_NOBARRIER 0x0100
1897 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1898 #define EXT4_DEFM_DISCARD 0x0400
1899 #define EXT4_DEFM_NODELALLOC 0x0800
1900
1901 /*
1902 * Default journal batch times
1903 */
1904 #define EXT4_DEF_MIN_BATCH_TIME 0
1905 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1906
1907 /*
1908 * Minimum number of groups in a flexgroup before we separate out
1909 * directories into the first block group of a flexgroup
1910 */
1911 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1912
1913 /*
1914 * Structure of a directory entry
1915 */
1916 #define EXT4_NAME_LEN 255
1917
1918 struct ext4_dir_entry {
1919 __le32 inode; /* Inode number */
1920 __le16 rec_len; /* Directory entry length */
1921 __le16 name_len; /* Name length */
1922 char name[EXT4_NAME_LEN]; /* File name */
1923 };
1924
1925 /*
1926 * The new version of the directory entry. Since EXT4 structures are
1927 * stored in intel byte order, and the name_len field could never be
1928 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1929 * file_type field.
1930 */
1931 struct ext4_dir_entry_2 {
1932 __le32 inode; /* Inode number */
1933 __le16 rec_len; /* Directory entry length */
1934 __u8 name_len; /* Name length */
1935 __u8 file_type;
1936 char name[EXT4_NAME_LEN]; /* File name */
1937 };
1938
1939 /*
1940 * This is a bogus directory entry at the end of each leaf block that
1941 * records checksums.
1942 */
1943 struct ext4_dir_entry_tail {
1944 __le32 det_reserved_zero1; /* Pretend to be unused */
1945 __le16 det_rec_len; /* 12 */
1946 __u8 det_reserved_zero2; /* Zero name length */
1947 __u8 det_reserved_ft; /* 0xDE, fake file type */
1948 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1949 };
1950
1951 #define EXT4_DIRENT_TAIL(block, blocksize) \
1952 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1953 ((blocksize) - \
1954 sizeof(struct ext4_dir_entry_tail))))
1955
1956 /*
1957 * Ext4 directory file types. Only the low 3 bits are used. The
1958 * other bits are reserved for now.
1959 */
1960 #define EXT4_FT_UNKNOWN 0
1961 #define EXT4_FT_REG_FILE 1
1962 #define EXT4_FT_DIR 2
1963 #define EXT4_FT_CHRDEV 3
1964 #define EXT4_FT_BLKDEV 4
1965 #define EXT4_FT_FIFO 5
1966 #define EXT4_FT_SOCK 6
1967 #define EXT4_FT_SYMLINK 7
1968
1969 #define EXT4_FT_MAX 8
1970
1971 #define EXT4_FT_DIR_CSUM 0xDE
1972
1973 /*
1974 * EXT4_DIR_PAD defines the directory entries boundaries
1975 *
1976 * NOTE: It must be a multiple of 4
1977 */
1978 #define EXT4_DIR_PAD 4
1979 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1980 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1981 ~EXT4_DIR_ROUND)
1982 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1983
1984 /*
1985 * If we ever get support for fs block sizes > page_size, we'll need
1986 * to remove the #if statements in the next two functions...
1987 */
1988 static inline unsigned int
1989 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1990 {
1991 unsigned len = le16_to_cpu(dlen);
1992
1993 #if (PAGE_SIZE >= 65536)
1994 if (len == EXT4_MAX_REC_LEN || len == 0)
1995 return blocksize;
1996 return (len & 65532) | ((len & 3) << 16);
1997 #else
1998 return len;
1999 #endif
2000 }
2001
2002 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2003 {
2004 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2005 BUG();
2006 #if (PAGE_SIZE >= 65536)
2007 if (len < 65536)
2008 return cpu_to_le16(len);
2009 if (len == blocksize) {
2010 if (blocksize == 65536)
2011 return cpu_to_le16(EXT4_MAX_REC_LEN);
2012 else
2013 return cpu_to_le16(0);
2014 }
2015 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2016 #else
2017 return cpu_to_le16(len);
2018 #endif
2019 }
2020
2021 /*
2022 * Hash Tree Directory indexing
2023 * (c) Daniel Phillips, 2001
2024 */
2025
2026 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2027 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2028 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2029 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2030
2031 /* Legal values for the dx_root hash_version field: */
2032
2033 #define DX_HASH_LEGACY 0
2034 #define DX_HASH_HALF_MD4 1
2035 #define DX_HASH_TEA 2
2036 #define DX_HASH_LEGACY_UNSIGNED 3
2037 #define DX_HASH_HALF_MD4_UNSIGNED 4
2038 #define DX_HASH_TEA_UNSIGNED 5
2039
2040 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2041 const void *address, unsigned int length)
2042 {
2043 struct {
2044 struct shash_desc shash;
2045 char ctx[4];
2046 } desc;
2047 int err;
2048
2049 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2050
2051 desc.shash.tfm = sbi->s_chksum_driver;
2052 desc.shash.flags = 0;
2053 *(u32 *)desc.ctx = crc;
2054
2055 err = crypto_shash_update(&desc.shash, address, length);
2056 BUG_ON(err);
2057
2058 return *(u32 *)desc.ctx;
2059 }
2060
2061 #ifdef __KERNEL__
2062
2063 /* hash info structure used by the directory hash */
2064 struct dx_hash_info
2065 {
2066 u32 hash;
2067 u32 minor_hash;
2068 int hash_version;
2069 u32 *seed;
2070 };
2071
2072
2073 /* 32 and 64 bit signed EOF for dx directories */
2074 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2075 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2076
2077
2078 /*
2079 * Control parameters used by ext4_htree_next_block
2080 */
2081 #define HASH_NB_ALWAYS 1
2082
2083 struct ext4_filename {
2084 const struct qstr *usr_fname;
2085 struct ext4_str disk_name;
2086 struct dx_hash_info hinfo;
2087 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2088 struct ext4_str crypto_buf;
2089 #endif
2090 };
2091
2092 #define fname_name(p) ((p)->disk_name.name)
2093 #define fname_len(p) ((p)->disk_name.len)
2094
2095 /*
2096 * Describe an inode's exact location on disk and in memory
2097 */
2098 struct ext4_iloc
2099 {
2100 struct buffer_head *bh;
2101 unsigned long offset;
2102 ext4_group_t block_group;
2103 };
2104
2105 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2106 {
2107 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2108 }
2109
2110 /*
2111 * This structure is stuffed into the struct file's private_data field
2112 * for directories. It is where we put information so that we can do
2113 * readdir operations in hash tree order.
2114 */
2115 struct dir_private_info {
2116 struct rb_root root;
2117 struct rb_node *curr_node;
2118 struct fname *extra_fname;
2119 loff_t last_pos;
2120 __u32 curr_hash;
2121 __u32 curr_minor_hash;
2122 __u32 next_hash;
2123 };
2124
2125 /* calculate the first block number of the group */
2126 static inline ext4_fsblk_t
2127 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2128 {
2129 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2130 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2131 }
2132
2133 /*
2134 * Special error return code only used by dx_probe() and its callers.
2135 */
2136 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2137
2138 /*
2139 * Timeout and state flag for lazy initialization inode thread.
2140 */
2141 #define EXT4_DEF_LI_WAIT_MULT 10
2142 #define EXT4_DEF_LI_MAX_START_DELAY 5
2143 #define EXT4_LAZYINIT_QUIT 0x0001
2144 #define EXT4_LAZYINIT_RUNNING 0x0002
2145
2146 /*
2147 * Lazy inode table initialization info
2148 */
2149 struct ext4_lazy_init {
2150 unsigned long li_state;
2151 struct list_head li_request_list;
2152 struct mutex li_list_mtx;
2153 };
2154
2155 struct ext4_li_request {
2156 struct super_block *lr_super;
2157 struct ext4_sb_info *lr_sbi;
2158 ext4_group_t lr_next_group;
2159 struct list_head lr_request;
2160 unsigned long lr_next_sched;
2161 unsigned long lr_timeout;
2162 };
2163
2164 struct ext4_features {
2165 struct kobject f_kobj;
2166 struct completion f_kobj_unregister;
2167 };
2168
2169 /*
2170 * This structure will be used for multiple mount protection. It will be
2171 * written into the block number saved in the s_mmp_block field in the
2172 * superblock. Programs that check MMP should assume that if
2173 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2174 * to use the filesystem, regardless of how old the timestamp is.
2175 */
2176 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2177 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2178 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2179 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2180
2181 struct mmp_struct {
2182 __le32 mmp_magic; /* Magic number for MMP */
2183 __le32 mmp_seq; /* Sequence no. updated periodically */
2184
2185 /*
2186 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2187 * purposes and do not affect the correctness of the algorithm
2188 */
2189 __le64 mmp_time; /* Time last updated */
2190 char mmp_nodename[64]; /* Node which last updated MMP block */
2191 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2192
2193 /*
2194 * mmp_check_interval is used to verify if the MMP block has been
2195 * updated on the block device. The value is updated based on the
2196 * maximum time to write the MMP block during an update cycle.
2197 */
2198 __le16 mmp_check_interval;
2199
2200 __le16 mmp_pad1;
2201 __le32 mmp_pad2[226];
2202 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2203 };
2204
2205 /* arguments passed to the mmp thread */
2206 struct mmpd_data {
2207 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2208 struct super_block *sb; /* super block of the fs */
2209 };
2210
2211 /*
2212 * Check interval multiplier
2213 * The MMP block is written every update interval and initially checked every
2214 * update interval x the multiplier (the value is then adapted based on the
2215 * write latency). The reason is that writes can be delayed under load and we
2216 * don't want readers to incorrectly assume that the filesystem is no longer
2217 * in use.
2218 */
2219 #define EXT4_MMP_CHECK_MULT 2UL
2220
2221 /*
2222 * Minimum interval for MMP checking in seconds.
2223 */
2224 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2225
2226 /*
2227 * Maximum interval for MMP checking in seconds.
2228 */
2229 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2230
2231 /*
2232 * Function prototypes
2233 */
2234
2235 /*
2236 * Ok, these declarations are also in <linux/kernel.h> but none of the
2237 * ext4 source programs needs to include it so they are duplicated here.
2238 */
2239 # define NORET_TYPE /**/
2240 # define ATTRIB_NORET __attribute__((noreturn))
2241 # define NORET_AND noreturn,
2242
2243 /* bitmap.c */
2244 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2245 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2246 struct ext4_group_desc *gdp,
2247 struct buffer_head *bh, int sz);
2248 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2249 struct ext4_group_desc *gdp,
2250 struct buffer_head *bh, int sz);
2251 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2252 struct ext4_group_desc *gdp,
2253 struct buffer_head *bh);
2254 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2255 struct ext4_group_desc *gdp,
2256 struct buffer_head *bh);
2257
2258 /* balloc.c */
2259 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2260 ext4_fsblk_t blocknr,
2261 ext4_group_t *blockgrpp,
2262 ext4_grpblk_t *offsetp);
2263 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2264 ext4_fsblk_t block);
2265
2266 extern unsigned int ext4_block_group(struct super_block *sb,
2267 ext4_fsblk_t blocknr);
2268 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2269 ext4_fsblk_t blocknr);
2270 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2271 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2272 ext4_group_t group);
2273 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2274 ext4_fsblk_t goal,
2275 unsigned int flags,
2276 unsigned long *count,
2277 int *errp);
2278 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2279 s64 nclusters, unsigned int flags);
2280 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2281 extern void ext4_check_blocks_bitmap(struct super_block *);
2282 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2283 ext4_group_t block_group,
2284 struct buffer_head ** bh);
2285 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2286
2287 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2288 ext4_group_t block_group);
2289 extern int ext4_wait_block_bitmap(struct super_block *sb,
2290 ext4_group_t block_group,
2291 struct buffer_head *bh);
2292 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2293 ext4_group_t block_group);
2294 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2295 ext4_group_t block_group,
2296 struct ext4_group_desc *gdp);
2297 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2298
2299 /* crypto_policy.c */
2300 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2301 struct inode *child);
2302 int ext4_inherit_context(struct inode *parent, struct inode *child);
2303 void ext4_to_hex(char *dst, char *src, size_t src_size);
2304 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2305 struct inode *inode);
2306 int ext4_get_policy(struct inode *inode,
2307 struct ext4_encryption_policy *policy);
2308
2309 /* crypto.c */
2310 extern struct kmem_cache *ext4_crypt_info_cachep;
2311 bool ext4_valid_contents_enc_mode(uint32_t mode);
2312 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2313 extern struct workqueue_struct *ext4_read_workqueue;
2314 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode,
2315 gfp_t gfp_flags);
2316 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2317 void ext4_restore_control_page(struct page *data_page);
2318 struct page *ext4_encrypt(struct inode *inode,
2319 struct page *plaintext_page,
2320 gfp_t gfp_flags);
2321 int ext4_decrypt(struct page *page);
2322 int ext4_encrypted_zeroout(struct inode *inode, ext4_lblk_t lblk,
2323 ext4_fsblk_t pblk, ext4_lblk_t len);
2324 extern const struct dentry_operations ext4_encrypted_d_ops;
2325
2326 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2327 int ext4_init_crypto(void);
2328 void ext4_exit_crypto(void);
2329 static inline int ext4_sb_has_crypto(struct super_block *sb)
2330 {
2331 return ext4_has_feature_encrypt(sb);
2332 }
2333 #else
2334 static inline int ext4_init_crypto(void) { return 0; }
2335 static inline void ext4_exit_crypto(void) { }
2336 static inline int ext4_sb_has_crypto(struct super_block *sb)
2337 {
2338 return 0;
2339 }
2340 #endif
2341
2342 /* crypto_fname.c */
2343 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2344 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2345 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2346 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2347 u32 ilen, struct ext4_str *crypto_str);
2348 int _ext4_fname_disk_to_usr(struct inode *inode,
2349 struct dx_hash_info *hinfo,
2350 const struct ext4_str *iname,
2351 struct ext4_str *oname);
2352 int ext4_fname_disk_to_usr(struct inode *inode,
2353 struct dx_hash_info *hinfo,
2354 const struct ext4_dir_entry_2 *de,
2355 struct ext4_str *oname);
2356 int ext4_fname_usr_to_disk(struct inode *inode,
2357 const struct qstr *iname,
2358 struct ext4_str *oname);
2359 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2360 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2361 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2362 int lookup, struct ext4_filename *fname);
2363 void ext4_fname_free_filename(struct ext4_filename *fname);
2364 #else
2365 static inline
2366 int ext4_setup_fname_crypto(struct inode *inode)
2367 {
2368 return 0;
2369 }
2370 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2371 static inline int ext4_fname_setup_filename(struct inode *dir,
2372 const struct qstr *iname,
2373 int lookup, struct ext4_filename *fname)
2374 {
2375 fname->usr_fname = iname;
2376 fname->disk_name.name = (unsigned char *) iname->name;
2377 fname->disk_name.len = iname->len;
2378 return 0;
2379 }
2380 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2381 #endif
2382
2383
2384 /* crypto_key.c */
2385 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2386 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2387 int _ext4_get_encryption_info(struct inode *inode);
2388
2389 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2390 int ext4_has_encryption_key(struct inode *inode);
2391
2392 static inline int ext4_get_encryption_info(struct inode *inode)
2393 {
2394 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2395
2396 if (!ci ||
2397 (ci->ci_keyring_key &&
2398 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2399 (1 << KEY_FLAG_REVOKED) |
2400 (1 << KEY_FLAG_DEAD)))))
2401 return _ext4_get_encryption_info(inode);
2402 return 0;
2403 }
2404
2405 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2406 {
2407 return EXT4_I(inode)->i_crypt_info;
2408 }
2409
2410 #else
2411 static inline int ext4_has_encryption_key(struct inode *inode)
2412 {
2413 return 0;
2414 }
2415 static inline int ext4_get_encryption_info(struct inode *inode)
2416 {
2417 return 0;
2418 }
2419 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2420 {
2421 return NULL;
2422 }
2423 #endif
2424
2425
2426 /* dir.c */
2427 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2428 struct file *,
2429 struct ext4_dir_entry_2 *,
2430 struct buffer_head *, char *, int,
2431 unsigned int);
2432 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2433 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2434 (de), (bh), (buf), (size), (offset)))
2435 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2436 __u32 minor_hash,
2437 struct ext4_dir_entry_2 *dirent,
2438 struct ext4_str *ent_name);
2439 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2440 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2441 struct buffer_head *bh,
2442 void *buf, int buf_size,
2443 struct ext4_filename *fname,
2444 struct ext4_dir_entry_2 **dest_de);
2445 int ext4_insert_dentry(struct inode *dir,
2446 struct inode *inode,
2447 struct ext4_dir_entry_2 *de,
2448 int buf_size,
2449 struct ext4_filename *fname);
2450 static inline void ext4_update_dx_flag(struct inode *inode)
2451 {
2452 if (!ext4_has_feature_dir_index(inode->i_sb))
2453 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2454 }
2455 static unsigned char ext4_filetype_table[] = {
2456 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2457 };
2458
2459 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2460 {
2461 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2462 return DT_UNKNOWN;
2463
2464 return ext4_filetype_table[filetype];
2465 }
2466 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2467 void *buf, int buf_size);
2468
2469 /* fsync.c */
2470 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2471
2472 /* hash.c */
2473 extern int ext4fs_dirhash(const char *name, int len, struct
2474 dx_hash_info *hinfo);
2475
2476 /* ialloc.c */
2477 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2478 const struct qstr *qstr, __u32 goal,
2479 uid_t *owner, int handle_type,
2480 unsigned int line_no, int nblocks);
2481
2482 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2483 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2484 0, 0, 0)
2485 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2486 type, nblocks) \
2487 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2488 (type), __LINE__, (nblocks))
2489
2490
2491 extern void ext4_free_inode(handle_t *, struct inode *);
2492 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2493 extern unsigned long ext4_count_free_inodes(struct super_block *);
2494 extern unsigned long ext4_count_dirs(struct super_block *);
2495 extern void ext4_check_inodes_bitmap(struct super_block *);
2496 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2497 extern int ext4_init_inode_table(struct super_block *sb,
2498 ext4_group_t group, int barrier);
2499 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2500
2501 /* mballoc.c */
2502 extern const struct file_operations ext4_seq_mb_groups_fops;
2503 extern long ext4_mb_stats;
2504 extern long ext4_mb_max_to_scan;
2505 extern int ext4_mb_init(struct super_block *);
2506 extern int ext4_mb_release(struct super_block *);
2507 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2508 struct ext4_allocation_request *, int *);
2509 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2510 extern void ext4_discard_preallocations(struct inode *);
2511 extern int __init ext4_init_mballoc(void);
2512 extern void ext4_exit_mballoc(void);
2513 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2514 struct buffer_head *bh, ext4_fsblk_t block,
2515 unsigned long count, int flags);
2516 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2517 ext4_group_t ngroups);
2518 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2519 ext4_group_t i, struct ext4_group_desc *desc);
2520 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2521 ext4_fsblk_t block, unsigned long count);
2522 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2523
2524 /* inode.c */
2525 int ext4_inode_is_fast_symlink(struct inode *inode);
2526 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2527 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2528 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2529 struct buffer_head *bh_result, int create);
2530 int ext4_dax_mmap_get_block(struct inode *inode, sector_t iblock,
2531 struct buffer_head *bh_result, int create);
2532 int ext4_get_block(struct inode *inode, sector_t iblock,
2533 struct buffer_head *bh_result, int create);
2534 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2535 struct buffer_head *bh_result, int create);
2536 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2537 struct buffer_head *bh, int create);
2538 int ext4_walk_page_buffers(handle_t *handle,
2539 struct buffer_head *head,
2540 unsigned from,
2541 unsigned to,
2542 int *partial,
2543 int (*fn)(handle_t *handle,
2544 struct buffer_head *bh));
2545 int do_journal_get_write_access(handle_t *handle,
2546 struct buffer_head *bh);
2547 #define FALL_BACK_TO_NONDELALLOC 1
2548 #define CONVERT_INLINE_DATA 2
2549
2550 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2551 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2552 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2553 extern int ext4_setattr(struct dentry *, struct iattr *);
2554 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2555 struct kstat *stat);
2556 extern void ext4_evict_inode(struct inode *);
2557 extern void ext4_clear_inode(struct inode *);
2558 extern int ext4_sync_inode(handle_t *, struct inode *);
2559 extern void ext4_dirty_inode(struct inode *, int);
2560 extern int ext4_change_inode_journal_flag(struct inode *, int);
2561 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2562 extern int ext4_inode_attach_jinode(struct inode *inode);
2563 extern int ext4_can_truncate(struct inode *inode);
2564 extern void ext4_truncate(struct inode *);
2565 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2566 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2567 extern void ext4_set_inode_flags(struct inode *);
2568 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2569 extern int ext4_alloc_da_blocks(struct inode *inode);
2570 extern void ext4_set_aops(struct inode *inode);
2571 extern int ext4_writepage_trans_blocks(struct inode *);
2572 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2573 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2574 loff_t lstart, loff_t lend);
2575 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2576 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2577 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2578 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2579 extern void ext4_da_update_reserve_space(struct inode *inode,
2580 int used, int quota_claim);
2581 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2582 ext4_fsblk_t pblk, ext4_lblk_t len);
2583 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2584 unsigned int map_len,
2585 struct extent_status *result);
2586
2587 /* indirect.c */
2588 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2589 struct ext4_map_blocks *map, int flags);
2590 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2591 loff_t offset);
2592 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2593 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2594 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2595 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2596 ext4_lblk_t start, ext4_lblk_t end);
2597
2598 /* ioctl.c */
2599 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2600 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2601
2602 /* migrate.c */
2603 extern int ext4_ext_migrate(struct inode *);
2604 extern int ext4_ind_migrate(struct inode *inode);
2605
2606 /* namei.c */
2607 extern int ext4_dirent_csum_verify(struct inode *inode,
2608 struct ext4_dir_entry *dirent);
2609 extern int ext4_orphan_add(handle_t *, struct inode *);
2610 extern int ext4_orphan_del(handle_t *, struct inode *);
2611 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2612 __u32 start_minor_hash, __u32 *next_hash);
2613 extern int ext4_search_dir(struct buffer_head *bh,
2614 char *search_buf,
2615 int buf_size,
2616 struct inode *dir,
2617 struct ext4_filename *fname,
2618 const struct qstr *d_name,
2619 unsigned int offset,
2620 struct ext4_dir_entry_2 **res_dir);
2621 extern int ext4_generic_delete_entry(handle_t *handle,
2622 struct inode *dir,
2623 struct ext4_dir_entry_2 *de_del,
2624 struct buffer_head *bh,
2625 void *entry_buf,
2626 int buf_size,
2627 int csum_size);
2628 extern int ext4_empty_dir(struct inode *inode);
2629
2630 /* resize.c */
2631 extern int ext4_group_add(struct super_block *sb,
2632 struct ext4_new_group_data *input);
2633 extern int ext4_group_extend(struct super_block *sb,
2634 struct ext4_super_block *es,
2635 ext4_fsblk_t n_blocks_count);
2636 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2637
2638 /* super.c */
2639 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2640 extern int ext4_calculate_overhead(struct super_block *sb);
2641 extern void ext4_superblock_csum_set(struct super_block *sb);
2642 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2643 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2644 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2645 ext4_group_t ngroup);
2646 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2647 char nbuf[16]);
2648
2649 extern __printf(4, 5)
2650 void __ext4_error(struct super_block *, const char *, unsigned int,
2651 const char *, ...);
2652 extern __printf(5, 6)
2653 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2654 const char *, ...);
2655 extern __printf(5, 6)
2656 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2657 const char *, ...);
2658 extern void __ext4_std_error(struct super_block *, const char *,
2659 unsigned int, int);
2660 extern __printf(4, 5)
2661 void __ext4_abort(struct super_block *, const char *, unsigned int,
2662 const char *, ...);
2663 extern __printf(4, 5)
2664 void __ext4_warning(struct super_block *, const char *, unsigned int,
2665 const char *, ...);
2666 extern __printf(4, 5)
2667 void __ext4_warning_inode(const struct inode *inode, const char *function,
2668 unsigned int line, const char *fmt, ...);
2669 extern __printf(3, 4)
2670 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2671 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2672 const char *, unsigned int, const char *);
2673 extern __printf(7, 8)
2674 void __ext4_grp_locked_error(const char *, unsigned int,
2675 struct super_block *, ext4_group_t,
2676 unsigned long, ext4_fsblk_t,
2677 const char *, ...);
2678
2679 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2680 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2681
2682 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2683 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2684
2685 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2686 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2687
2688 #ifdef CONFIG_PRINTK
2689
2690 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2691 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2692 #define ext4_error_file(file, func, line, block, fmt, ...) \
2693 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2694 #define ext4_error(sb, fmt, ...) \
2695 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2696 #define ext4_abort(sb, fmt, ...) \
2697 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2698 #define ext4_warning(sb, fmt, ...) \
2699 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2700 #define ext4_warning_inode(inode, fmt, ...) \
2701 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2702 #define ext4_msg(sb, level, fmt, ...) \
2703 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2704 #define dump_mmp_msg(sb, mmp, msg) \
2705 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2706 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2707 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2708 fmt, ##__VA_ARGS__)
2709
2710 #else
2711
2712 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2713 do { \
2714 no_printk(fmt, ##__VA_ARGS__); \
2715 __ext4_error_inode(inode, "", 0, block, " "); \
2716 } while (0)
2717 #define ext4_error_file(file, func, line, block, fmt, ...) \
2718 do { \
2719 no_printk(fmt, ##__VA_ARGS__); \
2720 __ext4_error_file(file, "", 0, block, " "); \
2721 } while (0)
2722 #define ext4_error(sb, fmt, ...) \
2723 do { \
2724 no_printk(fmt, ##__VA_ARGS__); \
2725 __ext4_error(sb, "", 0, " "); \
2726 } while (0)
2727 #define ext4_abort(sb, fmt, ...) \
2728 do { \
2729 no_printk(fmt, ##__VA_ARGS__); \
2730 __ext4_abort(sb, "", 0, " "); \
2731 } while (0)
2732 #define ext4_warning(sb, fmt, ...) \
2733 do { \
2734 no_printk(fmt, ##__VA_ARGS__); \
2735 __ext4_warning(sb, "", 0, " "); \
2736 } while (0)
2737 #define ext4_warning_inode(inode, fmt, ...) \
2738 do { \
2739 no_printk(fmt, ##__VA_ARGS__); \
2740 __ext4_warning_inode(inode, "", 0, " "); \
2741 } while (0)
2742 #define ext4_msg(sb, level, fmt, ...) \
2743 do { \
2744 no_printk(fmt, ##__VA_ARGS__); \
2745 __ext4_msg(sb, "", " "); \
2746 } while (0)
2747 #define dump_mmp_msg(sb, mmp, msg) \
2748 __dump_mmp_msg(sb, mmp, "", 0, "")
2749 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2750 do { \
2751 no_printk(fmt, ##__VA_ARGS__); \
2752 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2753 } while (0)
2754
2755 #endif
2756
2757 extern void ext4_update_dynamic_rev(struct super_block *sb);
2758 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2759 __u32 compat);
2760 extern int ext4_update_rocompat_feature(handle_t *handle,
2761 struct super_block *sb, __u32 rocompat);
2762 extern int ext4_update_incompat_feature(handle_t *handle,
2763 struct super_block *sb, __u32 incompat);
2764 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2765 struct ext4_group_desc *bg);
2766 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2767 struct ext4_group_desc *bg);
2768 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2769 struct ext4_group_desc *bg);
2770 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2771 struct ext4_group_desc *bg);
2772 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2773 struct ext4_group_desc *bg);
2774 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2775 struct ext4_group_desc *bg);
2776 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2777 struct ext4_group_desc *bg);
2778 extern void ext4_block_bitmap_set(struct super_block *sb,
2779 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2780 extern void ext4_inode_bitmap_set(struct super_block *sb,
2781 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2782 extern void ext4_inode_table_set(struct super_block *sb,
2783 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2784 extern void ext4_free_group_clusters_set(struct super_block *sb,
2785 struct ext4_group_desc *bg,
2786 __u32 count);
2787 extern void ext4_free_inodes_set(struct super_block *sb,
2788 struct ext4_group_desc *bg, __u32 count);
2789 extern void ext4_used_dirs_set(struct super_block *sb,
2790 struct ext4_group_desc *bg, __u32 count);
2791 extern void ext4_itable_unused_set(struct super_block *sb,
2792 struct ext4_group_desc *bg, __u32 count);
2793 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2794 struct ext4_group_desc *gdp);
2795 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2796 struct ext4_group_desc *gdp);
2797 extern int ext4_register_li_request(struct super_block *sb,
2798 ext4_group_t first_not_zeroed);
2799
2800 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2801 {
2802 return ext4_has_feature_gdt_csum(sb) ||
2803 EXT4_SB(sb)->s_chksum_driver != NULL;
2804 }
2805
2806 static inline int ext4_has_metadata_csum(struct super_block *sb)
2807 {
2808 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2809 !EXT4_SB(sb)->s_chksum_driver);
2810
2811 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2812 }
2813 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2814 {
2815 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2816 le32_to_cpu(es->s_blocks_count_lo);
2817 }
2818
2819 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2820 {
2821 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2822 le32_to_cpu(es->s_r_blocks_count_lo);
2823 }
2824
2825 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2826 {
2827 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2828 le32_to_cpu(es->s_free_blocks_count_lo);
2829 }
2830
2831 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2832 ext4_fsblk_t blk)
2833 {
2834 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2835 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2836 }
2837
2838 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2839 ext4_fsblk_t blk)
2840 {
2841 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2842 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2843 }
2844
2845 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2846 ext4_fsblk_t blk)
2847 {
2848 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2849 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2850 }
2851
2852 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2853 {
2854 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2855 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2856 le32_to_cpu(raw_inode->i_size_lo);
2857 else
2858 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2859 }
2860
2861 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2862 {
2863 raw_inode->i_size_lo = cpu_to_le32(i_size);
2864 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2865 }
2866
2867 static inline
2868 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2869 ext4_group_t group)
2870 {
2871 struct ext4_group_info ***grp_info;
2872 long indexv, indexh;
2873 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2874 grp_info = EXT4_SB(sb)->s_group_info;
2875 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2876 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2877 return grp_info[indexv][indexh];
2878 }
2879
2880 /*
2881 * Reading s_groups_count requires using smp_rmb() afterwards. See
2882 * the locking protocol documented in the comments of ext4_group_add()
2883 * in resize.c
2884 */
2885 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2886 {
2887 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2888
2889 smp_rmb();
2890 return ngroups;
2891 }
2892
2893 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2894 ext4_group_t block_group)
2895 {
2896 return block_group >> sbi->s_log_groups_per_flex;
2897 }
2898
2899 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2900 {
2901 return 1 << sbi->s_log_groups_per_flex;
2902 }
2903
2904 #define ext4_std_error(sb, errno) \
2905 do { \
2906 if ((errno)) \
2907 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2908 } while (0)
2909
2910 #ifdef CONFIG_SMP
2911 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2912 * counters. So we need to make sure we have free clusters more
2913 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2914 */
2915 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2916 #else
2917 #define EXT4_FREECLUSTERS_WATERMARK 0
2918 #endif
2919
2920 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2921 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2922 {
2923 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2924 !inode_is_locked(inode));
2925 down_write(&EXT4_I(inode)->i_data_sem);
2926 if (newsize > EXT4_I(inode)->i_disksize)
2927 EXT4_I(inode)->i_disksize = newsize;
2928 up_write(&EXT4_I(inode)->i_data_sem);
2929 }
2930
2931 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2932 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2933 {
2934 int changed = 0;
2935
2936 if (newsize > inode->i_size) {
2937 i_size_write(inode, newsize);
2938 changed = 1;
2939 }
2940 if (newsize > EXT4_I(inode)->i_disksize) {
2941 ext4_update_i_disksize(inode, newsize);
2942 changed |= 2;
2943 }
2944 return changed;
2945 }
2946
2947 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2948 loff_t len);
2949
2950 struct ext4_group_info {
2951 unsigned long bb_state;
2952 struct rb_root bb_free_root;
2953 ext4_grpblk_t bb_first_free; /* first free block */
2954 ext4_grpblk_t bb_free; /* total free blocks */
2955 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2956 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2957 struct list_head bb_prealloc_list;
2958 #ifdef DOUBLE_CHECK
2959 void *bb_bitmap;
2960 #endif
2961 struct rw_semaphore alloc_sem;
2962 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2963 * regions, index is order.
2964 * bb_counters[3] = 5 means
2965 * 5 free 8-block regions. */
2966 };
2967
2968 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2969 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2970 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2971 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2972
2973 #define EXT4_MB_GRP_NEED_INIT(grp) \
2974 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2975 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2976 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2977 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2978 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2979
2980 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2981 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2982 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2983 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2984 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2985 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2986
2987 #define EXT4_MAX_CONTENTION 8
2988 #define EXT4_CONTENTION_THRESHOLD 2
2989
2990 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2991 ext4_group_t group)
2992 {
2993 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2994 }
2995
2996 /*
2997 * Returns true if the filesystem is busy enough that attempts to
2998 * access the block group locks has run into contention.
2999 */
3000 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3001 {
3002 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3003 }
3004
3005 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3006 {
3007 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3008 if (spin_trylock(lock))
3009 /*
3010 * We're able to grab the lock right away, so drop the
3011 * lock contention counter.
3012 */
3013 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3014 else {
3015 /*
3016 * The lock is busy, so bump the contention counter,
3017 * and then wait on the spin lock.
3018 */
3019 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3020 EXT4_MAX_CONTENTION);
3021 spin_lock(lock);
3022 }
3023 }
3024
3025 static inline void ext4_unlock_group(struct super_block *sb,
3026 ext4_group_t group)
3027 {
3028 spin_unlock(ext4_group_lock_ptr(sb, group));
3029 }
3030
3031 /*
3032 * Block validity checking
3033 */
3034 #define ext4_check_indirect_blockref(inode, bh) \
3035 ext4_check_blockref(__func__, __LINE__, inode, \
3036 (__le32 *)(bh)->b_data, \
3037 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3038
3039 #define ext4_ind_check_inode(inode) \
3040 ext4_check_blockref(__func__, __LINE__, inode, \
3041 EXT4_I(inode)->i_data, \
3042 EXT4_NDIR_BLOCKS)
3043
3044 /*
3045 * Inodes and files operations
3046 */
3047
3048 /* dir.c */
3049 extern const struct file_operations ext4_dir_operations;
3050
3051 /* file.c */
3052 extern const struct inode_operations ext4_file_inode_operations;
3053 extern const struct file_operations ext4_file_operations;
3054 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3055
3056 /* inline.c */
3057 extern int ext4_get_max_inline_size(struct inode *inode);
3058 extern int ext4_find_inline_data_nolock(struct inode *inode);
3059 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3060 unsigned int len);
3061 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3062
3063 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3064 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3065 struct inode *inode,
3066 loff_t pos, unsigned len,
3067 unsigned flags,
3068 struct page **pagep);
3069 extern int ext4_write_inline_data_end(struct inode *inode,
3070 loff_t pos, unsigned len,
3071 unsigned copied,
3072 struct page *page);
3073 extern struct buffer_head *
3074 ext4_journalled_write_inline_data(struct inode *inode,
3075 unsigned len,
3076 struct page *page);
3077 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3078 struct inode *inode,
3079 loff_t pos, unsigned len,
3080 unsigned flags,
3081 struct page **pagep,
3082 void **fsdata);
3083 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3084 unsigned len, unsigned copied,
3085 struct page *page);
3086 extern int ext4_try_add_inline_entry(handle_t *handle,
3087 struct ext4_filename *fname,
3088 struct inode *dir, struct inode *inode);
3089 extern int ext4_try_create_inline_dir(handle_t *handle,
3090 struct inode *parent,
3091 struct inode *inode);
3092 extern int ext4_read_inline_dir(struct file *filp,
3093 struct dir_context *ctx,
3094 int *has_inline_data);
3095 extern int htree_inlinedir_to_tree(struct file *dir_file,
3096 struct inode *dir, ext4_lblk_t block,
3097 struct dx_hash_info *hinfo,
3098 __u32 start_hash, __u32 start_minor_hash,
3099 int *has_inline_data);
3100 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3101 struct ext4_filename *fname,
3102 const struct qstr *d_name,
3103 struct ext4_dir_entry_2 **res_dir,
3104 int *has_inline_data);
3105 extern int ext4_delete_inline_entry(handle_t *handle,
3106 struct inode *dir,
3107 struct ext4_dir_entry_2 *de_del,
3108 struct buffer_head *bh,
3109 int *has_inline_data);
3110 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3111 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3112 struct ext4_dir_entry_2 **parent_de,
3113 int *retval);
3114 extern int ext4_inline_data_fiemap(struct inode *inode,
3115 struct fiemap_extent_info *fieinfo,
3116 int *has_inline, __u64 start, __u64 len);
3117 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3118 struct inode *inode,
3119 int needed);
3120 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3121
3122 extern int ext4_convert_inline_data(struct inode *inode);
3123
3124 static inline int ext4_has_inline_data(struct inode *inode)
3125 {
3126 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3127 EXT4_I(inode)->i_inline_off;
3128 }
3129
3130 /* namei.c */
3131 extern const struct inode_operations ext4_dir_inode_operations;
3132 extern const struct inode_operations ext4_special_inode_operations;
3133 extern struct dentry *ext4_get_parent(struct dentry *child);
3134 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3135 struct ext4_dir_entry_2 *de,
3136 int blocksize, int csum_size,
3137 unsigned int parent_ino, int dotdot_real_len);
3138 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3139 unsigned int blocksize);
3140 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3141 struct inode *inode,
3142 struct buffer_head *bh);
3143 #define S_SHIFT 12
3144 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3145 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3146 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3147 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3148 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3149 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3150 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3151 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3152 };
3153
3154 static inline void ext4_set_de_type(struct super_block *sb,
3155 struct ext4_dir_entry_2 *de,
3156 umode_t mode) {
3157 if (ext4_has_feature_filetype(sb))
3158 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3159 }
3160
3161 /* readpages.c */
3162 extern int ext4_mpage_readpages(struct address_space *mapping,
3163 struct list_head *pages, struct page *page,
3164 unsigned nr_pages);
3165
3166 /* symlink.c */
3167 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3168 extern const struct inode_operations ext4_symlink_inode_operations;
3169 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3170
3171 /* sysfs.c */
3172 extern int ext4_register_sysfs(struct super_block *sb);
3173 extern void ext4_unregister_sysfs(struct super_block *sb);
3174 extern int __init ext4_init_sysfs(void);
3175 extern void ext4_exit_sysfs(void);
3176
3177 /* block_validity */
3178 extern void ext4_release_system_zone(struct super_block *sb);
3179 extern int ext4_setup_system_zone(struct super_block *sb);
3180 extern int __init ext4_init_system_zone(void);
3181 extern void ext4_exit_system_zone(void);
3182 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3183 ext4_fsblk_t start_blk,
3184 unsigned int count);
3185 extern int ext4_check_blockref(const char *, unsigned int,
3186 struct inode *, __le32 *, unsigned int);
3187
3188 /* extents.c */
3189 struct ext4_ext_path;
3190 struct ext4_extent;
3191
3192 /*
3193 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3194 * __le32.
3195 */
3196 #define EXT_MAX_BLOCKS 0xffffffff
3197
3198 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3199 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3200 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3201 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3202 struct ext4_map_blocks *map, int flags);
3203 extern void ext4_ext_truncate(handle_t *, struct inode *);
3204 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3205 ext4_lblk_t end);
3206 extern void ext4_ext_init(struct super_block *);
3207 extern void ext4_ext_release(struct super_block *);
3208 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3209 loff_t len);
3210 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3211 loff_t offset, ssize_t len);
3212 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3213 struct ext4_map_blocks *map, int flags);
3214 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3215 ext4_lblk_t lblocks);
3216 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3217 int num,
3218 struct ext4_ext_path *path);
3219 extern int ext4_can_extents_be_merged(struct inode *inode,
3220 struct ext4_extent *ex1,
3221 struct ext4_extent *ex2);
3222 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3223 struct ext4_ext_path **,
3224 struct ext4_extent *, int);
3225 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3226 struct ext4_ext_path **,
3227 int flags);
3228 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3229 extern int ext4_ext_check_inode(struct inode *inode);
3230 extern int ext4_find_delalloc_range(struct inode *inode,
3231 ext4_lblk_t lblk_start,
3232 ext4_lblk_t lblk_end);
3233 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3234 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3235 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3236 __u64 start, __u64 len);
3237 extern int ext4_ext_precache(struct inode *inode);
3238 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3239 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3240 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3241 struct inode *inode2, ext4_lblk_t lblk1,
3242 ext4_lblk_t lblk2, ext4_lblk_t count,
3243 int mark_unwritten,int *err);
3244
3245 /* move_extent.c */
3246 extern void ext4_double_down_write_data_sem(struct inode *first,
3247 struct inode *second);
3248 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3249 struct inode *donor_inode);
3250 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3251 __u64 start_orig, __u64 start_donor,
3252 __u64 len, __u64 *moved_len);
3253
3254 /* page-io.c */
3255 extern int __init ext4_init_pageio(void);
3256 extern void ext4_exit_pageio(void);
3257 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3258 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3259 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3260 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3261 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3262 struct writeback_control *wbc);
3263 extern void ext4_end_io_rsv_work(struct work_struct *work);
3264 extern void ext4_io_submit(struct ext4_io_submit *io);
3265 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3266 struct page *page,
3267 int len,
3268 struct writeback_control *wbc,
3269 bool keep_towrite);
3270
3271 /* mmp.c */
3272 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3273
3274 /*
3275 * Add new method to test whether block and inode bitmaps are properly
3276 * initialized. With uninit_bg reading the block from disk is not enough
3277 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3278 */
3279 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3280
3281 static inline int bitmap_uptodate(struct buffer_head *bh)
3282 {
3283 return (buffer_uptodate(bh) &&
3284 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3285 }
3286 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3287 {
3288 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3289 }
3290
3291 /*
3292 * Disable DIO read nolock optimization, so new dioreaders will be forced
3293 * to grab i_mutex
3294 */
3295 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3296 {
3297 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3298 smp_mb();
3299 }
3300 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3301 {
3302 smp_mb();
3303 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3304 }
3305
3306 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3307
3308 /* For ioend & aio unwritten conversion wait queues */
3309 #define EXT4_WQ_HASH_SZ 37
3310 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3311 EXT4_WQ_HASH_SZ])
3312 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3313
3314 #define EXT4_RESIZING 0
3315 extern int ext4_resize_begin(struct super_block *sb);
3316 extern void ext4_resize_end(struct super_block *sb);
3317
3318 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3319 struct ext4_io_end *io_end)
3320 {
3321 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3322 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3323 atomic_inc(&EXT4_I(inode)->i_unwritten);
3324 }
3325 }
3326
3327 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3328 {
3329 struct inode *inode = io_end->inode;
3330
3331 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3332 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3333 /* Wake up anyone waiting on unwritten extent conversion */
3334 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3335 wake_up_all(ext4_ioend_wq(inode));
3336 }
3337 }
3338
3339 #endif /* __KERNEL__ */
3340
3341 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3342 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3343
3344 #endif /* _EXT4_H */