]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - fs/ext4/ext4.h
ext4: create EXT4_MAX_BLOCKS() macro
[mirror_ubuntu-zesty-kernel.git] / fs / ext4 / ext4.h
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/fscrypto.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #ifdef __KERNEL__
39 #include <linux/compat.h>
40 #endif
41
42 /*
43 * The fourth extended filesystem constants/structures
44 */
45
46 /*
47 * with AGGRESSIVE_CHECK allocator runs consistency checks over
48 * structures. these checks slow things down a lot
49 */
50 #define AGGRESSIVE_CHECK__
51
52 /*
53 * with DOUBLE_CHECK defined mballoc creates persistent in-core
54 * bitmaps, maintains and uses them to check for double allocations
55 */
56 #define DOUBLE_CHECK__
57
58 /*
59 * Define EXT4FS_DEBUG to produce debug messages
60 */
61 #undef EXT4FS_DEBUG
62
63 /*
64 * Debug code
65 */
66 #ifdef EXT4FS_DEBUG
67 #define ext4_debug(f, a...) \
68 do { \
69 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
70 __FILE__, __LINE__, __func__); \
71 printk(KERN_DEBUG f, ## a); \
72 } while (0)
73 #else
74 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
75 #endif
76
77 /*
78 * Turn on EXT_DEBUG to get lots of info about extents operations.
79 */
80 #define EXT_DEBUG__
81 #ifdef EXT_DEBUG
82 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
83 #else
84 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
85 #endif
86
87 /* data type for block offset of block group */
88 typedef int ext4_grpblk_t;
89
90 /* data type for filesystem-wide blocks number */
91 typedef unsigned long long ext4_fsblk_t;
92
93 /* data type for file logical block number */
94 typedef __u32 ext4_lblk_t;
95
96 /* data type for block group number */
97 typedef unsigned int ext4_group_t;
98
99 enum SHIFT_DIRECTION {
100 SHIFT_LEFT = 0,
101 SHIFT_RIGHT,
102 };
103
104 /*
105 * Flags used in mballoc's allocation_context flags field.
106 *
107 * Also used to show what's going on for debugging purposes when the
108 * flag field is exported via the traceport interface
109 */
110
111 /* prefer goal again. length */
112 #define EXT4_MB_HINT_MERGE 0x0001
113 /* blocks already reserved */
114 #define EXT4_MB_HINT_RESERVED 0x0002
115 /* metadata is being allocated */
116 #define EXT4_MB_HINT_METADATA 0x0004
117 /* first blocks in the file */
118 #define EXT4_MB_HINT_FIRST 0x0008
119 /* search for the best chunk */
120 #define EXT4_MB_HINT_BEST 0x0010
121 /* data is being allocated */
122 #define EXT4_MB_HINT_DATA 0x0020
123 /* don't preallocate (for tails) */
124 #define EXT4_MB_HINT_NOPREALLOC 0x0040
125 /* allocate for locality group */
126 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
127 /* allocate goal blocks or none */
128 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
129 /* goal is meaningful */
130 #define EXT4_MB_HINT_TRY_GOAL 0x0200
131 /* blocks already pre-reserved by delayed allocation */
132 #define EXT4_MB_DELALLOC_RESERVED 0x0400
133 /* We are doing stream allocation */
134 #define EXT4_MB_STREAM_ALLOC 0x0800
135 /* Use reserved root blocks if needed */
136 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
137 /* Use blocks from reserved pool */
138 #define EXT4_MB_USE_RESERVED 0x2000
139
140 struct ext4_allocation_request {
141 /* target inode for block we're allocating */
142 struct inode *inode;
143 /* how many blocks we want to allocate */
144 unsigned int len;
145 /* logical block in target inode */
146 ext4_lblk_t logical;
147 /* the closest logical allocated block to the left */
148 ext4_lblk_t lleft;
149 /* the closest logical allocated block to the right */
150 ext4_lblk_t lright;
151 /* phys. target (a hint) */
152 ext4_fsblk_t goal;
153 /* phys. block for the closest logical allocated block to the left */
154 ext4_fsblk_t pleft;
155 /* phys. block for the closest logical allocated block to the right */
156 ext4_fsblk_t pright;
157 /* flags. see above EXT4_MB_HINT_* */
158 unsigned int flags;
159 };
160
161 /*
162 * Logical to physical block mapping, used by ext4_map_blocks()
163 *
164 * This structure is used to pass requests into ext4_map_blocks() as
165 * well as to store the information returned by ext4_map_blocks(). It
166 * takes less room on the stack than a struct buffer_head.
167 */
168 #define EXT4_MAP_NEW (1 << BH_New)
169 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
170 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
171 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
172 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
173 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
174
175 struct ext4_map_blocks {
176 ext4_fsblk_t m_pblk;
177 ext4_lblk_t m_lblk;
178 unsigned int m_len;
179 unsigned int m_flags;
180 };
181
182 /*
183 * Flags for ext4_io_end->flags
184 */
185 #define EXT4_IO_END_UNWRITTEN 0x0001
186
187 /*
188 * For converting unwritten extents on a work queue. 'handle' is used for
189 * buffered writeback.
190 */
191 typedef struct ext4_io_end {
192 struct list_head list; /* per-file finished IO list */
193 handle_t *handle; /* handle reserved for extent
194 * conversion */
195 struct inode *inode; /* file being written to */
196 struct bio *bio; /* Linked list of completed
197 * bios covering the extent */
198 unsigned int flag; /* unwritten or not */
199 atomic_t count; /* reference counter */
200 loff_t offset; /* offset in the file */
201 ssize_t size; /* size of the extent */
202 } ext4_io_end_t;
203
204 struct ext4_io_submit {
205 struct writeback_control *io_wbc;
206 struct bio *io_bio;
207 ext4_io_end_t *io_end;
208 sector_t io_next_block;
209 };
210
211 /*
212 * Special inodes numbers
213 */
214 #define EXT4_BAD_INO 1 /* Bad blocks inode */
215 #define EXT4_ROOT_INO 2 /* Root inode */
216 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
217 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
218 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
219 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
220 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
221 #define EXT4_JOURNAL_INO 8 /* Journal inode */
222
223 /* First non-reserved inode for old ext4 filesystems */
224 #define EXT4_GOOD_OLD_FIRST_INO 11
225
226 /*
227 * Maximal count of links to a file
228 */
229 #define EXT4_LINK_MAX 65000
230
231 /*
232 * Macro-instructions used to manage several block sizes
233 */
234 #define EXT4_MIN_BLOCK_SIZE 1024
235 #define EXT4_MAX_BLOCK_SIZE 65536
236 #define EXT4_MIN_BLOCK_LOG_SIZE 10
237 #define EXT4_MAX_BLOCK_LOG_SIZE 16
238 #ifdef __KERNEL__
239 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
240 #else
241 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
242 #endif
243 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
244 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
245 EXT4_SB(s)->s_cluster_bits)
246 #ifdef __KERNEL__
247 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
248 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
249 #else
250 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
251 #endif
252 #ifdef __KERNEL__
253 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
254 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
255 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
256 #else
257 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
258 EXT4_GOOD_OLD_INODE_SIZE : \
259 (s)->s_inode_size)
260 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
261 EXT4_GOOD_OLD_FIRST_INO : \
262 (s)->s_first_ino)
263 #endif
264 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
265 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
266 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
267 blkbits))
268
269 /* Translate a block number to a cluster number */
270 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
271 /* Translate a cluster number to a block number */
272 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
273 /* Translate # of blks to # of clusters */
274 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
275 (sbi)->s_cluster_bits)
276 /* Mask out the low bits to get the starting block of the cluster */
277 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
278 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
279 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
280 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
281 /* Get the cluster offset */
282 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
283 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
284 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
285 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
286
287 /*
288 * Structure of a blocks group descriptor
289 */
290 struct ext4_group_desc
291 {
292 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
293 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
294 __le32 bg_inode_table_lo; /* Inodes table block */
295 __le16 bg_free_blocks_count_lo;/* Free blocks count */
296 __le16 bg_free_inodes_count_lo;/* Free inodes count */
297 __le16 bg_used_dirs_count_lo; /* Directories count */
298 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
299 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
300 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
301 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
302 __le16 bg_itable_unused_lo; /* Unused inodes count */
303 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
304 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
305 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
306 __le32 bg_inode_table_hi; /* Inodes table block MSB */
307 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
308 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
309 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
310 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
311 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
312 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
313 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
314 __u32 bg_reserved;
315 };
316
317 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
318 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
319 sizeof(__le16))
320 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
321 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
322 sizeof(__le16))
323
324 /*
325 * Structure of a flex block group info
326 */
327
328 struct flex_groups {
329 atomic64_t free_clusters;
330 atomic_t free_inodes;
331 atomic_t used_dirs;
332 };
333
334 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
335 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
336 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
337
338 /*
339 * Macro-instructions used to manage group descriptors
340 */
341 #define EXT4_MIN_DESC_SIZE 32
342 #define EXT4_MIN_DESC_SIZE_64BIT 64
343 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
344 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
345 #ifdef __KERNEL__
346 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
347 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
348 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
349 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
350 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
351 #else
352 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
353 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
354 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
355 #endif
356
357 /*
358 * Constants relative to the data blocks
359 */
360 #define EXT4_NDIR_BLOCKS 12
361 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
362 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
363 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
364 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
365
366 /*
367 * Inode flags
368 */
369 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
370 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
371 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
372 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
373 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
374 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
375 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
376 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
377 /* Reserved for compression usage... */
378 #define EXT4_DIRTY_FL 0x00000100
379 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
380 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
381 /* nb: was previously EXT2_ECOMPR_FL */
382 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
383 /* End compression flags --- maybe not all used */
384 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
385 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
386 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
387 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
388 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
389 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
390 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
391 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
392 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
393 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
394 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
395 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
396 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
397
398 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
399 #define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
400
401 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
402 EXT4_IMMUTABLE_FL | \
403 EXT4_APPEND_FL | \
404 EXT4_NODUMP_FL | \
405 EXT4_NOATIME_FL | \
406 EXT4_PROJINHERIT_FL)
407
408 /* Flags that should be inherited by new inodes from their parent. */
409 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
410 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
411 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
412 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
413 EXT4_PROJINHERIT_FL)
414
415 /* Flags that are appropriate for regular files (all but dir-specific ones). */
416 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
417
418 /* Flags that are appropriate for non-directories/regular files. */
419 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
420
421 /* Mask out flags that are inappropriate for the given type of inode. */
422 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
423 {
424 if (S_ISDIR(mode))
425 return flags;
426 else if (S_ISREG(mode))
427 return flags & EXT4_REG_FLMASK;
428 else
429 return flags & EXT4_OTHER_FLMASK;
430 }
431
432 /*
433 * Inode flags used for atomic set/get
434 */
435 enum {
436 EXT4_INODE_SECRM = 0, /* Secure deletion */
437 EXT4_INODE_UNRM = 1, /* Undelete */
438 EXT4_INODE_COMPR = 2, /* Compress file */
439 EXT4_INODE_SYNC = 3, /* Synchronous updates */
440 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
441 EXT4_INODE_APPEND = 5, /* writes to file may only append */
442 EXT4_INODE_NODUMP = 6, /* do not dump file */
443 EXT4_INODE_NOATIME = 7, /* do not update atime */
444 /* Reserved for compression usage... */
445 EXT4_INODE_DIRTY = 8,
446 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
447 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
448 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
449 /* End compression flags --- maybe not all used */
450 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
451 EXT4_INODE_IMAGIC = 13, /* AFS directory */
452 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
453 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
454 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
455 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
456 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
457 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
458 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
459 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
460 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
461 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
462 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
463 };
464
465 /*
466 * Since it's pretty easy to mix up bit numbers and hex values, we use a
467 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
468 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
469 * any extra space in the compiled kernel image, otherwise, the build will fail.
470 * It's important that these values are the same, since we are using
471 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
472 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
473 * values found in ext2, ext3 and ext4 filesystems, and of course the values
474 * defined in e2fsprogs.
475 *
476 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
477 */
478 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
479 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
480
481 static inline void ext4_check_flag_values(void)
482 {
483 CHECK_FLAG_VALUE(SECRM);
484 CHECK_FLAG_VALUE(UNRM);
485 CHECK_FLAG_VALUE(COMPR);
486 CHECK_FLAG_VALUE(SYNC);
487 CHECK_FLAG_VALUE(IMMUTABLE);
488 CHECK_FLAG_VALUE(APPEND);
489 CHECK_FLAG_VALUE(NODUMP);
490 CHECK_FLAG_VALUE(NOATIME);
491 CHECK_FLAG_VALUE(DIRTY);
492 CHECK_FLAG_VALUE(COMPRBLK);
493 CHECK_FLAG_VALUE(NOCOMPR);
494 CHECK_FLAG_VALUE(ENCRYPT);
495 CHECK_FLAG_VALUE(INDEX);
496 CHECK_FLAG_VALUE(IMAGIC);
497 CHECK_FLAG_VALUE(JOURNAL_DATA);
498 CHECK_FLAG_VALUE(NOTAIL);
499 CHECK_FLAG_VALUE(DIRSYNC);
500 CHECK_FLAG_VALUE(TOPDIR);
501 CHECK_FLAG_VALUE(HUGE_FILE);
502 CHECK_FLAG_VALUE(EXTENTS);
503 CHECK_FLAG_VALUE(EA_INODE);
504 CHECK_FLAG_VALUE(EOFBLOCKS);
505 CHECK_FLAG_VALUE(INLINE_DATA);
506 CHECK_FLAG_VALUE(PROJINHERIT);
507 CHECK_FLAG_VALUE(RESERVED);
508 }
509
510 /* Used to pass group descriptor data when online resize is done */
511 struct ext4_new_group_input {
512 __u32 group; /* Group number for this data */
513 __u64 block_bitmap; /* Absolute block number of block bitmap */
514 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
515 __u64 inode_table; /* Absolute block number of inode table start */
516 __u32 blocks_count; /* Total number of blocks in this group */
517 __u16 reserved_blocks; /* Number of reserved blocks in this group */
518 __u16 unused;
519 };
520
521 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
522 struct compat_ext4_new_group_input {
523 u32 group;
524 compat_u64 block_bitmap;
525 compat_u64 inode_bitmap;
526 compat_u64 inode_table;
527 u32 blocks_count;
528 u16 reserved_blocks;
529 u16 unused;
530 };
531 #endif
532
533 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
534 struct ext4_new_group_data {
535 __u32 group;
536 __u64 block_bitmap;
537 __u64 inode_bitmap;
538 __u64 inode_table;
539 __u32 blocks_count;
540 __u16 reserved_blocks;
541 __u16 unused;
542 __u32 free_blocks_count;
543 };
544
545 /* Indexes used to index group tables in ext4_new_group_data */
546 enum {
547 BLOCK_BITMAP = 0, /* block bitmap */
548 INODE_BITMAP, /* inode bitmap */
549 INODE_TABLE, /* inode tables */
550 GROUP_TABLE_COUNT,
551 };
552
553 /*
554 * Flags used by ext4_map_blocks()
555 */
556 /* Allocate any needed blocks and/or convert an unwritten
557 extent to be an initialized ext4 */
558 #define EXT4_GET_BLOCKS_CREATE 0x0001
559 /* Request the creation of an unwritten extent */
560 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
561 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
562 EXT4_GET_BLOCKS_CREATE)
563 /* Caller is from the delayed allocation writeout path
564 * finally doing the actual allocation of delayed blocks */
565 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
566 /* caller is from the direct IO path, request to creation of an
567 unwritten extents if not allocated, split the unwritten
568 extent if blocks has been preallocated already*/
569 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
570 #define EXT4_GET_BLOCKS_CONVERT 0x0010
571 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
572 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
573 /* Convert extent to initialized after IO complete */
574 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
575 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
576 /* Eventual metadata allocation (due to growing extent tree)
577 * should not fail, so try to use reserved blocks for that.*/
578 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
579 /* Don't normalize allocation size (used for fallocate) */
580 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
581 /* Request will not result in inode size update (user for fallocate) */
582 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
583 /* Convert written extents to unwritten */
584 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
585 /* Write zeros to newly created written extents */
586 #define EXT4_GET_BLOCKS_ZERO 0x0200
587 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
588 EXT4_GET_BLOCKS_ZERO)
589 /* Caller will submit data before dropping transaction handle. This
590 * allows jbd2 to avoid submitting data before commit. */
591 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
592
593 /*
594 * The bit position of these flags must not overlap with any of the
595 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
596 * read_extent_tree_block(), ext4_split_extent_at(),
597 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
598 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
599 * caching the extents when reading from the extent tree while a
600 * truncate or punch hole operation is in progress.
601 */
602 #define EXT4_EX_NOCACHE 0x40000000
603 #define EXT4_EX_FORCE_CACHE 0x20000000
604
605 /*
606 * Flags used by ext4_free_blocks
607 */
608 #define EXT4_FREE_BLOCKS_METADATA 0x0001
609 #define EXT4_FREE_BLOCKS_FORGET 0x0002
610 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
611 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
612 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
613 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
614
615 /*
616 * ioctl commands
617 */
618 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
619 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
620 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
621 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
622 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
623 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
624 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
625 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
626 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
627 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
628 #define EXT4_IOC_MIGRATE _IO('f', 9)
629 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
630 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
631 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
632 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
633 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
634 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
635 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
636 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
637 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
638 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
639
640 #ifndef FS_IOC_FSGETXATTR
641 /* Until the uapi changes get merged for project quota... */
642
643 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
644 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
645
646 /*
647 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
648 */
649 struct fsxattr {
650 __u32 fsx_xflags; /* xflags field value (get/set) */
651 __u32 fsx_extsize; /* extsize field value (get/set)*/
652 __u32 fsx_nextents; /* nextents field value (get) */
653 __u32 fsx_projid; /* project identifier (get/set) */
654 unsigned char fsx_pad[12];
655 };
656
657 /*
658 * Flags for the fsx_xflags field
659 */
660 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
661 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
662 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
663 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
664 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
665 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
666 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
667 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
668 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
669 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
670 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
671 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
672 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
673 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
674 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
675 #endif /* !defined(FS_IOC_FSGETXATTR) */
676
677 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
678 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
679
680 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
681 /*
682 * ioctl commands in 32 bit emulation
683 */
684 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
685 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
686 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
687 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
688 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
689 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
690 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
691 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
692 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
693 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
694 #endif
695
696 /* Max physical block we can address w/o extents */
697 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
698
699 /*
700 * Structure of an inode on the disk
701 */
702 struct ext4_inode {
703 __le16 i_mode; /* File mode */
704 __le16 i_uid; /* Low 16 bits of Owner Uid */
705 __le32 i_size_lo; /* Size in bytes */
706 __le32 i_atime; /* Access time */
707 __le32 i_ctime; /* Inode Change time */
708 __le32 i_mtime; /* Modification time */
709 __le32 i_dtime; /* Deletion Time */
710 __le16 i_gid; /* Low 16 bits of Group Id */
711 __le16 i_links_count; /* Links count */
712 __le32 i_blocks_lo; /* Blocks count */
713 __le32 i_flags; /* File flags */
714 union {
715 struct {
716 __le32 l_i_version;
717 } linux1;
718 struct {
719 __u32 h_i_translator;
720 } hurd1;
721 struct {
722 __u32 m_i_reserved1;
723 } masix1;
724 } osd1; /* OS dependent 1 */
725 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
726 __le32 i_generation; /* File version (for NFS) */
727 __le32 i_file_acl_lo; /* File ACL */
728 __le32 i_size_high;
729 __le32 i_obso_faddr; /* Obsoleted fragment address */
730 union {
731 struct {
732 __le16 l_i_blocks_high; /* were l_i_reserved1 */
733 __le16 l_i_file_acl_high;
734 __le16 l_i_uid_high; /* these 2 fields */
735 __le16 l_i_gid_high; /* were reserved2[0] */
736 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
737 __le16 l_i_reserved;
738 } linux2;
739 struct {
740 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
741 __u16 h_i_mode_high;
742 __u16 h_i_uid_high;
743 __u16 h_i_gid_high;
744 __u32 h_i_author;
745 } hurd2;
746 struct {
747 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
748 __le16 m_i_file_acl_high;
749 __u32 m_i_reserved2[2];
750 } masix2;
751 } osd2; /* OS dependent 2 */
752 __le16 i_extra_isize;
753 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
754 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
755 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
756 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
757 __le32 i_crtime; /* File Creation time */
758 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
759 __le32 i_version_hi; /* high 32 bits for 64-bit version */
760 __le32 i_projid; /* Project ID */
761 };
762
763 struct move_extent {
764 __u32 reserved; /* should be zero */
765 __u32 donor_fd; /* donor file descriptor */
766 __u64 orig_start; /* logical start offset in block for orig */
767 __u64 donor_start; /* logical start offset in block for donor */
768 __u64 len; /* block length to be moved */
769 __u64 moved_len; /* moved block length */
770 };
771
772 #define EXT4_EPOCH_BITS 2
773 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
774 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
775
776 /*
777 * Extended fields will fit into an inode if the filesystem was formatted
778 * with large inodes (-I 256 or larger) and there are not currently any EAs
779 * consuming all of the available space. For new inodes we always reserve
780 * enough space for the kernel's known extended fields, but for inodes
781 * created with an old kernel this might not have been the case. None of
782 * the extended inode fields is critical for correct filesystem operation.
783 * This macro checks if a certain field fits in the inode. Note that
784 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
785 */
786 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
787 ((offsetof(typeof(*ext4_inode), field) + \
788 sizeof((ext4_inode)->field)) \
789 <= (EXT4_GOOD_OLD_INODE_SIZE + \
790 (einode)->i_extra_isize)) \
791
792 /*
793 * We use an encoding that preserves the times for extra epoch "00":
794 *
795 * extra msb of adjust for signed
796 * epoch 32-bit 32-bit tv_sec to
797 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
798 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
799 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
800 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
801 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
802 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
803 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
804 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
805 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
806 *
807 * Note that previous versions of the kernel on 64-bit systems would
808 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
809 * 1970. e2fsck will correct this, assuming that it is run on the
810 * affected filesystem before 2242.
811 */
812
813 static inline __le32 ext4_encode_extra_time(struct timespec *time)
814 {
815 u32 extra = sizeof(time->tv_sec) > 4 ?
816 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
817 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
818 }
819
820 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
821 {
822 if (unlikely(sizeof(time->tv_sec) > 4 &&
823 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
824 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
825 /* Handle legacy encoding of pre-1970 dates with epoch
826 * bits 1,1. We assume that by kernel version 4.20,
827 * everyone will have run fsck over the affected
828 * filesystems to correct the problem. (This
829 * backwards compatibility may be removed before this
830 * time, at the discretion of the ext4 developers.)
831 */
832 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
833 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
834 extra_bits = 0;
835 time->tv_sec += extra_bits << 32;
836 #else
837 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
838 #endif
839 }
840 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
841 }
842
843 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
844 do { \
845 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
846 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
847 (raw_inode)->xtime ## _extra = \
848 ext4_encode_extra_time(&(inode)->xtime); \
849 } while (0)
850
851 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
852 do { \
853 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
854 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
855 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
856 (raw_inode)->xtime ## _extra = \
857 ext4_encode_extra_time(&(einode)->xtime); \
858 } while (0)
859
860 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
861 do { \
862 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
863 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
864 ext4_decode_extra_time(&(inode)->xtime, \
865 raw_inode->xtime ## _extra); \
866 else \
867 (inode)->xtime.tv_nsec = 0; \
868 } while (0)
869
870 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
871 do { \
872 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
873 (einode)->xtime.tv_sec = \
874 (signed)le32_to_cpu((raw_inode)->xtime); \
875 else \
876 (einode)->xtime.tv_sec = 0; \
877 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
878 ext4_decode_extra_time(&(einode)->xtime, \
879 raw_inode->xtime ## _extra); \
880 else \
881 (einode)->xtime.tv_nsec = 0; \
882 } while (0)
883
884 #define i_disk_version osd1.linux1.l_i_version
885
886 #if defined(__KERNEL__) || defined(__linux__)
887 #define i_reserved1 osd1.linux1.l_i_reserved1
888 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
889 #define i_blocks_high osd2.linux2.l_i_blocks_high
890 #define i_uid_low i_uid
891 #define i_gid_low i_gid
892 #define i_uid_high osd2.linux2.l_i_uid_high
893 #define i_gid_high osd2.linux2.l_i_gid_high
894 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
895
896 #elif defined(__GNU__)
897
898 #define i_translator osd1.hurd1.h_i_translator
899 #define i_uid_high osd2.hurd2.h_i_uid_high
900 #define i_gid_high osd2.hurd2.h_i_gid_high
901 #define i_author osd2.hurd2.h_i_author
902
903 #elif defined(__masix__)
904
905 #define i_reserved1 osd1.masix1.m_i_reserved1
906 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
907 #define i_reserved2 osd2.masix2.m_i_reserved2
908
909 #endif /* defined(__KERNEL__) || defined(__linux__) */
910
911 #include "extents_status.h"
912
913 /*
914 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
915 *
916 * These are needed to avoid lockdep false positives when we need to
917 * allocate blocks to the quota inode during ext4_map_blocks(), while
918 * holding i_data_sem for a normal (non-quota) inode. Since we don't
919 * do quota tracking for the quota inode, this avoids deadlock (as
920 * well as infinite recursion, since it isn't turtles all the way
921 * down...)
922 *
923 * I_DATA_SEM_NORMAL - Used for most inodes
924 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
925 * where the second inode has larger inode number
926 * than the first
927 * I_DATA_SEM_QUOTA - Used for quota inodes only
928 */
929 enum {
930 I_DATA_SEM_NORMAL = 0,
931 I_DATA_SEM_OTHER,
932 I_DATA_SEM_QUOTA,
933 };
934
935
936 /*
937 * fourth extended file system inode data in memory
938 */
939 struct ext4_inode_info {
940 __le32 i_data[15]; /* unconverted */
941 __u32 i_dtime;
942 ext4_fsblk_t i_file_acl;
943
944 /*
945 * i_block_group is the number of the block group which contains
946 * this file's inode. Constant across the lifetime of the inode,
947 * it is ued for making block allocation decisions - we try to
948 * place a file's data blocks near its inode block, and new inodes
949 * near to their parent directory's inode.
950 */
951 ext4_group_t i_block_group;
952 ext4_lblk_t i_dir_start_lookup;
953 #if (BITS_PER_LONG < 64)
954 unsigned long i_state_flags; /* Dynamic state flags */
955 #endif
956 unsigned long i_flags;
957
958 /*
959 * Extended attributes can be read independently of the main file
960 * data. Taking i_mutex even when reading would cause contention
961 * between readers of EAs and writers of regular file data, so
962 * instead we synchronize on xattr_sem when reading or changing
963 * EAs.
964 */
965 struct rw_semaphore xattr_sem;
966
967 struct list_head i_orphan; /* unlinked but open inodes */
968
969 /*
970 * i_disksize keeps track of what the inode size is ON DISK, not
971 * in memory. During truncate, i_size is set to the new size by
972 * the VFS prior to calling ext4_truncate(), but the filesystem won't
973 * set i_disksize to 0 until the truncate is actually under way.
974 *
975 * The intent is that i_disksize always represents the blocks which
976 * are used by this file. This allows recovery to restart truncate
977 * on orphans if we crash during truncate. We actually write i_disksize
978 * into the on-disk inode when writing inodes out, instead of i_size.
979 *
980 * The only time when i_disksize and i_size may be different is when
981 * a truncate is in progress. The only things which change i_disksize
982 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
983 */
984 loff_t i_disksize;
985
986 /*
987 * i_data_sem is for serialising ext4_truncate() against
988 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
989 * data tree are chopped off during truncate. We can't do that in
990 * ext4 because whenever we perform intermediate commits during
991 * truncate, the inode and all the metadata blocks *must* be in a
992 * consistent state which allows truncation of the orphans to restart
993 * during recovery. Hence we must fix the get_block-vs-truncate race
994 * by other means, so we have i_data_sem.
995 */
996 struct rw_semaphore i_data_sem;
997 /*
998 * i_mmap_sem is for serializing page faults with truncate / punch hole
999 * operations. We have to make sure that new page cannot be faulted in
1000 * a section of the inode that is being punched. We cannot easily use
1001 * i_data_sem for this since we need protection for the whole punch
1002 * operation and i_data_sem ranks below transaction start so we have
1003 * to occasionally drop it.
1004 */
1005 struct rw_semaphore i_mmap_sem;
1006 struct inode vfs_inode;
1007 struct jbd2_inode *jinode;
1008
1009 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1010
1011 /*
1012 * File creation time. Its function is same as that of
1013 * struct timespec i_{a,c,m}time in the generic inode.
1014 */
1015 struct timespec i_crtime;
1016
1017 /* mballoc */
1018 struct list_head i_prealloc_list;
1019 spinlock_t i_prealloc_lock;
1020
1021 /* extents status tree */
1022 struct ext4_es_tree i_es_tree;
1023 rwlock_t i_es_lock;
1024 struct list_head i_es_list;
1025 unsigned int i_es_all_nr; /* protected by i_es_lock */
1026 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1027 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1028 extents to shrink. Protected by
1029 i_es_lock */
1030
1031 /* ialloc */
1032 ext4_group_t i_last_alloc_group;
1033
1034 /* allocation reservation info for delalloc */
1035 /* In case of bigalloc, these refer to clusters rather than blocks */
1036 unsigned int i_reserved_data_blocks;
1037 unsigned int i_reserved_meta_blocks;
1038 unsigned int i_allocated_meta_blocks;
1039 ext4_lblk_t i_da_metadata_calc_last_lblock;
1040 int i_da_metadata_calc_len;
1041
1042 /* on-disk additional length */
1043 __u16 i_extra_isize;
1044
1045 /* Indicate the inline data space. */
1046 u16 i_inline_off;
1047 u16 i_inline_size;
1048
1049 #ifdef CONFIG_QUOTA
1050 /* quota space reservation, managed internally by quota code */
1051 qsize_t i_reserved_quota;
1052 #endif
1053
1054 /* Lock protecting lists below */
1055 spinlock_t i_completed_io_lock;
1056 /*
1057 * Completed IOs that need unwritten extents handling and have
1058 * transaction reserved
1059 */
1060 struct list_head i_rsv_conversion_list;
1061 struct work_struct i_rsv_conversion_work;
1062 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1063
1064 spinlock_t i_block_reservation_lock;
1065
1066 /*
1067 * Transactions that contain inode's metadata needed to complete
1068 * fsync and fdatasync, respectively.
1069 */
1070 tid_t i_sync_tid;
1071 tid_t i_datasync_tid;
1072
1073 #ifdef CONFIG_QUOTA
1074 struct dquot *i_dquot[MAXQUOTAS];
1075 #endif
1076
1077 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1078 __u32 i_csum_seed;
1079
1080 kprojid_t i_projid;
1081 };
1082
1083 /*
1084 * File system states
1085 */
1086 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1087 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1088 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1089
1090 /*
1091 * Misc. filesystem flags
1092 */
1093 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1094 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1095 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1096
1097 /*
1098 * Mount flags set via mount options or defaults
1099 */
1100 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1101 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1102 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1103 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1104 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1105 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1106 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1107 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1108 #ifdef CONFIG_FS_DAX
1109 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1110 #else
1111 #define EXT4_MOUNT_DAX 0
1112 #endif
1113 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1114 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1115 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1116 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1117 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1118 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1119 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1120 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1121 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1122 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1123 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1124 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1125 * enable enforcement for hidden
1126 * quota files */
1127 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1128 * enforcement for hidden quota
1129 * files */
1130 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1131 * enforcement */
1132 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1133 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1134 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1135 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1136 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1137 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1138 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1139 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1140
1141 /*
1142 * Mount flags set either automatically (could not be set by mount option)
1143 * based on per file system feature or property or in special cases such as
1144 * distinguishing between explicit mount option definition and default.
1145 */
1146 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1147 specified delalloc */
1148 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1149 size of blocksize * 8
1150 blocks */
1151 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1152 file systems */
1153
1154 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1155 specified journal checksum */
1156
1157 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1158 ~EXT4_MOUNT_##opt
1159 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1160 EXT4_MOUNT_##opt
1161 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1162 EXT4_MOUNT_##opt)
1163
1164 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1165 ~EXT4_MOUNT2_##opt
1166 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1167 EXT4_MOUNT2_##opt
1168 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1169 EXT4_MOUNT2_##opt)
1170
1171 #define ext4_test_and_set_bit __test_and_set_bit_le
1172 #define ext4_set_bit __set_bit_le
1173 #define ext4_set_bit_atomic ext2_set_bit_atomic
1174 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1175 #define ext4_clear_bit __clear_bit_le
1176 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1177 #define ext4_test_bit test_bit_le
1178 #define ext4_find_next_zero_bit find_next_zero_bit_le
1179 #define ext4_find_next_bit find_next_bit_le
1180
1181 extern void ext4_set_bits(void *bm, int cur, int len);
1182
1183 /*
1184 * Maximal mount counts between two filesystem checks
1185 */
1186 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1187 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1188
1189 /*
1190 * Behaviour when detecting errors
1191 */
1192 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1193 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1194 #define EXT4_ERRORS_PANIC 3 /* Panic */
1195 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1196
1197 /* Metadata checksum algorithm codes */
1198 #define EXT4_CRC32C_CHKSUM 1
1199
1200 /*
1201 * Structure of the super block
1202 */
1203 struct ext4_super_block {
1204 /*00*/ __le32 s_inodes_count; /* Inodes count */
1205 __le32 s_blocks_count_lo; /* Blocks count */
1206 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1207 __le32 s_free_blocks_count_lo; /* Free blocks count */
1208 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1209 __le32 s_first_data_block; /* First Data Block */
1210 __le32 s_log_block_size; /* Block size */
1211 __le32 s_log_cluster_size; /* Allocation cluster size */
1212 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1213 __le32 s_clusters_per_group; /* # Clusters per group */
1214 __le32 s_inodes_per_group; /* # Inodes per group */
1215 __le32 s_mtime; /* Mount time */
1216 /*30*/ __le32 s_wtime; /* Write time */
1217 __le16 s_mnt_count; /* Mount count */
1218 __le16 s_max_mnt_count; /* Maximal mount count */
1219 __le16 s_magic; /* Magic signature */
1220 __le16 s_state; /* File system state */
1221 __le16 s_errors; /* Behaviour when detecting errors */
1222 __le16 s_minor_rev_level; /* minor revision level */
1223 /*40*/ __le32 s_lastcheck; /* time of last check */
1224 __le32 s_checkinterval; /* max. time between checks */
1225 __le32 s_creator_os; /* OS */
1226 __le32 s_rev_level; /* Revision level */
1227 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1228 __le16 s_def_resgid; /* Default gid for reserved blocks */
1229 /*
1230 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1231 *
1232 * Note: the difference between the compatible feature set and
1233 * the incompatible feature set is that if there is a bit set
1234 * in the incompatible feature set that the kernel doesn't
1235 * know about, it should refuse to mount the filesystem.
1236 *
1237 * e2fsck's requirements are more strict; if it doesn't know
1238 * about a feature in either the compatible or incompatible
1239 * feature set, it must abort and not try to meddle with
1240 * things it doesn't understand...
1241 */
1242 __le32 s_first_ino; /* First non-reserved inode */
1243 __le16 s_inode_size; /* size of inode structure */
1244 __le16 s_block_group_nr; /* block group # of this superblock */
1245 __le32 s_feature_compat; /* compatible feature set */
1246 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1247 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1248 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1249 /*78*/ char s_volume_name[16]; /* volume name */
1250 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1251 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1252 /*
1253 * Performance hints. Directory preallocation should only
1254 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1255 */
1256 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1257 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1258 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1259 /*
1260 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1261 */
1262 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1263 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1264 __le32 s_journal_dev; /* device number of journal file */
1265 __le32 s_last_orphan; /* start of list of inodes to delete */
1266 __le32 s_hash_seed[4]; /* HTREE hash seed */
1267 __u8 s_def_hash_version; /* Default hash version to use */
1268 __u8 s_jnl_backup_type;
1269 __le16 s_desc_size; /* size of group descriptor */
1270 /*100*/ __le32 s_default_mount_opts;
1271 __le32 s_first_meta_bg; /* First metablock block group */
1272 __le32 s_mkfs_time; /* When the filesystem was created */
1273 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1274 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1275 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1276 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1277 __le32 s_free_blocks_count_hi; /* Free blocks count */
1278 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1279 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1280 __le32 s_flags; /* Miscellaneous flags */
1281 __le16 s_raid_stride; /* RAID stride */
1282 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1283 __le64 s_mmp_block; /* Block for multi-mount protection */
1284 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1285 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1286 __u8 s_checksum_type; /* metadata checksum algorithm used */
1287 __u8 s_encryption_level; /* versioning level for encryption */
1288 __u8 s_reserved_pad; /* Padding to next 32bits */
1289 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1290 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1291 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1292 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1293 snapshot's future use */
1294 __le32 s_snapshot_list; /* inode number of the head of the
1295 on-disk snapshot list */
1296 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1297 __le32 s_error_count; /* number of fs errors */
1298 __le32 s_first_error_time; /* first time an error happened */
1299 __le32 s_first_error_ino; /* inode involved in first error */
1300 __le64 s_first_error_block; /* block involved of first error */
1301 __u8 s_first_error_func[32]; /* function where the error happened */
1302 __le32 s_first_error_line; /* line number where error happened */
1303 __le32 s_last_error_time; /* most recent time of an error */
1304 __le32 s_last_error_ino; /* inode involved in last error */
1305 __le32 s_last_error_line; /* line number where error happened */
1306 __le64 s_last_error_block; /* block involved of last error */
1307 __u8 s_last_error_func[32]; /* function where the error happened */
1308 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1309 __u8 s_mount_opts[64];
1310 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1311 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1312 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1313 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1314 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1315 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1316 __le32 s_lpf_ino; /* Location of the lost+found inode */
1317 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1318 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1319 __le32 s_reserved[98]; /* Padding to the end of the block */
1320 __le32 s_checksum; /* crc32c(superblock) */
1321 };
1322
1323 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1324
1325 #ifdef __KERNEL__
1326
1327 /*
1328 * run-time mount flags
1329 */
1330 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1331 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1332 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1333
1334 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1335 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1336 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1337 #else
1338 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1339 #endif
1340
1341 /* Number of quota types we support */
1342 #define EXT4_MAXQUOTAS 3
1343
1344 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1345 #define EXT4_KEY_DESC_PREFIX "ext4:"
1346 #define EXT4_KEY_DESC_PREFIX_SIZE 5
1347 #endif
1348
1349 /*
1350 * fourth extended-fs super-block data in memory
1351 */
1352 struct ext4_sb_info {
1353 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1354 unsigned long s_inodes_per_block;/* Number of inodes per block */
1355 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1356 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1357 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1358 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1359 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1360 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1361 ext4_group_t s_groups_count; /* Number of groups in the fs */
1362 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1363 unsigned long s_overhead; /* # of fs overhead clusters */
1364 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1365 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1366 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1367 struct buffer_head * s_sbh; /* Buffer containing the super block */
1368 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1369 struct buffer_head **s_group_desc;
1370 unsigned int s_mount_opt;
1371 unsigned int s_mount_opt2;
1372 unsigned int s_mount_flags;
1373 unsigned int s_def_mount_opt;
1374 ext4_fsblk_t s_sb_block;
1375 atomic64_t s_resv_clusters;
1376 kuid_t s_resuid;
1377 kgid_t s_resgid;
1378 unsigned short s_mount_state;
1379 unsigned short s_pad;
1380 int s_addr_per_block_bits;
1381 int s_desc_per_block_bits;
1382 int s_inode_size;
1383 int s_first_ino;
1384 unsigned int s_inode_readahead_blks;
1385 unsigned int s_inode_goal;
1386 spinlock_t s_next_gen_lock;
1387 u32 s_next_generation;
1388 u32 s_hash_seed[4];
1389 int s_def_hash_version;
1390 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1391 struct percpu_counter s_freeclusters_counter;
1392 struct percpu_counter s_freeinodes_counter;
1393 struct percpu_counter s_dirs_counter;
1394 struct percpu_counter s_dirtyclusters_counter;
1395 struct blockgroup_lock *s_blockgroup_lock;
1396 struct proc_dir_entry *s_proc;
1397 struct kobject s_kobj;
1398 struct completion s_kobj_unregister;
1399 struct super_block *s_sb;
1400
1401 /* Journaling */
1402 struct journal_s *s_journal;
1403 struct list_head s_orphan;
1404 struct mutex s_orphan_lock;
1405 unsigned long s_resize_flags; /* Flags indicating if there
1406 is a resizer */
1407 unsigned long s_commit_interval;
1408 u32 s_max_batch_time;
1409 u32 s_min_batch_time;
1410 struct block_device *journal_bdev;
1411 #ifdef CONFIG_QUOTA
1412 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1413 int s_jquota_fmt; /* Format of quota to use */
1414 #endif
1415 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1416 struct rb_root system_blks;
1417
1418 #ifdef EXTENTS_STATS
1419 /* ext4 extents stats */
1420 unsigned long s_ext_min;
1421 unsigned long s_ext_max;
1422 unsigned long s_depth_max;
1423 spinlock_t s_ext_stats_lock;
1424 unsigned long s_ext_blocks;
1425 unsigned long s_ext_extents;
1426 #endif
1427
1428 /* for buddy allocator */
1429 struct ext4_group_info ***s_group_info;
1430 struct inode *s_buddy_cache;
1431 spinlock_t s_md_lock;
1432 unsigned short *s_mb_offsets;
1433 unsigned int *s_mb_maxs;
1434 unsigned int s_group_info_size;
1435 unsigned int s_mb_free_pending;
1436
1437 /* tunables */
1438 unsigned long s_stripe;
1439 unsigned int s_mb_stream_request;
1440 unsigned int s_mb_max_to_scan;
1441 unsigned int s_mb_min_to_scan;
1442 unsigned int s_mb_stats;
1443 unsigned int s_mb_order2_reqs;
1444 unsigned int s_mb_group_prealloc;
1445 unsigned int s_max_dir_size_kb;
1446 /* where last allocation was done - for stream allocation */
1447 unsigned long s_mb_last_group;
1448 unsigned long s_mb_last_start;
1449
1450 /* stats for buddy allocator */
1451 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1452 atomic_t s_bal_success; /* we found long enough chunks */
1453 atomic_t s_bal_allocated; /* in blocks */
1454 atomic_t s_bal_ex_scanned; /* total extents scanned */
1455 atomic_t s_bal_goals; /* goal hits */
1456 atomic_t s_bal_breaks; /* too long searches */
1457 atomic_t s_bal_2orders; /* 2^order hits */
1458 spinlock_t s_bal_lock;
1459 unsigned long s_mb_buddies_generated;
1460 unsigned long long s_mb_generation_time;
1461 atomic_t s_mb_lost_chunks;
1462 atomic_t s_mb_preallocated;
1463 atomic_t s_mb_discarded;
1464 atomic_t s_lock_busy;
1465
1466 /* locality groups */
1467 struct ext4_locality_group __percpu *s_locality_groups;
1468
1469 /* for write statistics */
1470 unsigned long s_sectors_written_start;
1471 u64 s_kbytes_written;
1472
1473 /* the size of zero-out chunk */
1474 unsigned int s_extent_max_zeroout_kb;
1475
1476 unsigned int s_log_groups_per_flex;
1477 struct flex_groups *s_flex_groups;
1478 ext4_group_t s_flex_groups_allocated;
1479
1480 /* workqueue for reserved extent conversions (buffered io) */
1481 struct workqueue_struct *rsv_conversion_wq;
1482
1483 /* timer for periodic error stats printing */
1484 struct timer_list s_err_report;
1485
1486 /* Lazy inode table initialization info */
1487 struct ext4_li_request *s_li_request;
1488 /* Wait multiplier for lazy initialization thread */
1489 unsigned int s_li_wait_mult;
1490
1491 /* Kernel thread for multiple mount protection */
1492 struct task_struct *s_mmp_tsk;
1493
1494 /* record the last minlen when FITRIM is called. */
1495 atomic_t s_last_trim_minblks;
1496
1497 /* Reference to checksum algorithm driver via cryptoapi */
1498 struct crypto_shash *s_chksum_driver;
1499
1500 /* Precomputed FS UUID checksum for seeding other checksums */
1501 __u32 s_csum_seed;
1502
1503 /* Reclaim extents from extent status tree */
1504 struct shrinker s_es_shrinker;
1505 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1506 long s_es_nr_inode;
1507 struct ext4_es_stats s_es_stats;
1508 struct mb_cache *s_mb_cache;
1509 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1510
1511 /* Ratelimit ext4 messages. */
1512 struct ratelimit_state s_err_ratelimit_state;
1513 struct ratelimit_state s_warning_ratelimit_state;
1514 struct ratelimit_state s_msg_ratelimit_state;
1515
1516 /* Barrier between changing inodes' journal flags and writepages ops. */
1517 struct percpu_rw_semaphore s_journal_flag_rwsem;
1518
1519 /* Encryption support */
1520 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1521 u8 key_prefix[EXT4_KEY_DESC_PREFIX_SIZE];
1522 u8 key_prefix_size;
1523 #endif
1524 };
1525
1526 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1527 {
1528 return sb->s_fs_info;
1529 }
1530 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1531 {
1532 return container_of(inode, struct ext4_inode_info, vfs_inode);
1533 }
1534
1535 static inline struct timespec ext4_current_time(struct inode *inode)
1536 {
1537 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1538 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1539 }
1540
1541 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1542 {
1543 return ino == EXT4_ROOT_INO ||
1544 ino == EXT4_USR_QUOTA_INO ||
1545 ino == EXT4_GRP_QUOTA_INO ||
1546 ino == EXT4_BOOT_LOADER_INO ||
1547 ino == EXT4_JOURNAL_INO ||
1548 ino == EXT4_RESIZE_INO ||
1549 (ino >= EXT4_FIRST_INO(sb) &&
1550 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1551 }
1552
1553 /*
1554 * Inode dynamic state flags
1555 */
1556 enum {
1557 EXT4_STATE_JDATA, /* journaled data exists */
1558 EXT4_STATE_NEW, /* inode is newly created */
1559 EXT4_STATE_XATTR, /* has in-inode xattrs */
1560 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1561 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1562 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1563 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1564 EXT4_STATE_NEWENTRY, /* File just added to dir */
1565 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1566 nolocking */
1567 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1568 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1569 };
1570
1571 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1572 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1573 { \
1574 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1575 } \
1576 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1577 { \
1578 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1579 } \
1580 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1581 { \
1582 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1583 }
1584
1585 /* Add these declarations here only so that these functions can be
1586 * found by name. Otherwise, they are very hard to locate. */
1587 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1588 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1589 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1590 EXT4_INODE_BIT_FNS(flag, flags, 0)
1591
1592 /* Add these declarations here only so that these functions can be
1593 * found by name. Otherwise, they are very hard to locate. */
1594 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1595 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1596 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1597 #if (BITS_PER_LONG < 64)
1598 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1599
1600 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1601 {
1602 (ei)->i_state_flags = 0;
1603 }
1604 #else
1605 EXT4_INODE_BIT_FNS(state, flags, 32)
1606
1607 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1608 {
1609 /* We depend on the fact that callers will set i_flags */
1610 }
1611 #endif
1612 #else
1613 /* Assume that user mode programs are passing in an ext4fs superblock, not
1614 * a kernel struct super_block. This will allow us to call the feature-test
1615 * macros from user land. */
1616 #define EXT4_SB(sb) (sb)
1617 #endif
1618
1619 /*
1620 * Returns true if the inode is inode is encrypted
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 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1649 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1650 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1651 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1652 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1653 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1654 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1655
1656 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1657 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1658 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1659 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1660 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1661 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1662 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1663 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1664 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1665 /*
1666 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1667 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1668 * all other data structures' checksums. However, the METADATA_CSUM and
1669 * GDT_CSUM bits are mutually exclusive.
1670 */
1671 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1672 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1673 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1674
1675 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1676 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1677 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1678 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1679 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1680 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1681 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1682 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1683 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1684 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1685 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1686 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1687 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1688 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1689 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1690
1691 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1692 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1693 { \
1694 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1695 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1696 } \
1697 static inline void ext4_set_feature_##name(struct super_block *sb) \
1698 { \
1699 EXT4_SB(sb)->s_es->s_feature_compat |= \
1700 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1701 } \
1702 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1703 { \
1704 EXT4_SB(sb)->s_es->s_feature_compat &= \
1705 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1706 }
1707
1708 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1709 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1710 { \
1711 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1712 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1713 } \
1714 static inline void ext4_set_feature_##name(struct super_block *sb) \
1715 { \
1716 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1717 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1718 } \
1719 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1720 { \
1721 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1722 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1723 }
1724
1725 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1726 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1727 { \
1728 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1729 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1730 } \
1731 static inline void ext4_set_feature_##name(struct super_block *sb) \
1732 { \
1733 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1734 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1735 } \
1736 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1737 { \
1738 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1739 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1740 }
1741
1742 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1743 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1744 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1745 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1746 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1747 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1748 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1749
1750 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1751 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1752 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1753 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1754 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1755 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1756 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1757 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1758 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1759 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1760 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1761 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1762
1763 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1764 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1765 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1766 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1769 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1770 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1771 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1772 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1773 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1774 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1775 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1776 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1777 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1778
1779 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1780 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1781 EXT4_FEATURE_INCOMPAT_META_BG)
1782 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1783 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1784 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1785
1786 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1787 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1788 EXT4_FEATURE_INCOMPAT_RECOVER| \
1789 EXT4_FEATURE_INCOMPAT_META_BG)
1790 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1791 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1792 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1793
1794 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1795 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1796 EXT4_FEATURE_INCOMPAT_RECOVER| \
1797 EXT4_FEATURE_INCOMPAT_META_BG| \
1798 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1799 EXT4_FEATURE_INCOMPAT_64BIT| \
1800 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1801 EXT4_FEATURE_INCOMPAT_MMP | \
1802 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1803 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1804 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1805 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1806 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1807 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1808 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1809 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1810 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1811 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1812 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1813 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1814 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1815 EXT4_FEATURE_RO_COMPAT_PROJECT)
1816
1817 #define EXTN_FEATURE_FUNCS(ver) \
1818 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1819 { \
1820 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1821 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1822 } \
1823 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1824 { \
1825 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1826 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1827 } \
1828 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1829 { \
1830 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1831 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1832 }
1833
1834 EXTN_FEATURE_FUNCS(2)
1835 EXTN_FEATURE_FUNCS(3)
1836 EXTN_FEATURE_FUNCS(4)
1837
1838 static inline bool ext4_has_compat_features(struct super_block *sb)
1839 {
1840 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1841 }
1842 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1843 {
1844 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1845 }
1846 static inline bool ext4_has_incompat_features(struct super_block *sb)
1847 {
1848 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1849 }
1850
1851 /*
1852 * Default values for user and/or group using reserved blocks
1853 */
1854 #define EXT4_DEF_RESUID 0
1855 #define EXT4_DEF_RESGID 0
1856
1857 /*
1858 * Default project ID
1859 */
1860 #define EXT4_DEF_PROJID 0
1861
1862 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1863
1864 /*
1865 * Default mount options
1866 */
1867 #define EXT4_DEFM_DEBUG 0x0001
1868 #define EXT4_DEFM_BSDGROUPS 0x0002
1869 #define EXT4_DEFM_XATTR_USER 0x0004
1870 #define EXT4_DEFM_ACL 0x0008
1871 #define EXT4_DEFM_UID16 0x0010
1872 #define EXT4_DEFM_JMODE 0x0060
1873 #define EXT4_DEFM_JMODE_DATA 0x0020
1874 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1875 #define EXT4_DEFM_JMODE_WBACK 0x0060
1876 #define EXT4_DEFM_NOBARRIER 0x0100
1877 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1878 #define EXT4_DEFM_DISCARD 0x0400
1879 #define EXT4_DEFM_NODELALLOC 0x0800
1880
1881 /*
1882 * Default journal batch times
1883 */
1884 #define EXT4_DEF_MIN_BATCH_TIME 0
1885 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1886
1887 /*
1888 * Minimum number of groups in a flexgroup before we separate out
1889 * directories into the first block group of a flexgroup
1890 */
1891 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1892
1893 /*
1894 * Structure of a directory entry
1895 */
1896 #define EXT4_NAME_LEN 255
1897
1898 struct ext4_dir_entry {
1899 __le32 inode; /* Inode number */
1900 __le16 rec_len; /* Directory entry length */
1901 __le16 name_len; /* Name length */
1902 char name[EXT4_NAME_LEN]; /* File name */
1903 };
1904
1905 /*
1906 * The new version of the directory entry. Since EXT4 structures are
1907 * stored in intel byte order, and the name_len field could never be
1908 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1909 * file_type field.
1910 */
1911 struct ext4_dir_entry_2 {
1912 __le32 inode; /* Inode number */
1913 __le16 rec_len; /* Directory entry length */
1914 __u8 name_len; /* Name length */
1915 __u8 file_type;
1916 char name[EXT4_NAME_LEN]; /* File name */
1917 };
1918
1919 /*
1920 * This is a bogus directory entry at the end of each leaf block that
1921 * records checksums.
1922 */
1923 struct ext4_dir_entry_tail {
1924 __le32 det_reserved_zero1; /* Pretend to be unused */
1925 __le16 det_rec_len; /* 12 */
1926 __u8 det_reserved_zero2; /* Zero name length */
1927 __u8 det_reserved_ft; /* 0xDE, fake file type */
1928 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1929 };
1930
1931 #define EXT4_DIRENT_TAIL(block, blocksize) \
1932 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1933 ((blocksize) - \
1934 sizeof(struct ext4_dir_entry_tail))))
1935
1936 /*
1937 * Ext4 directory file types. Only the low 3 bits are used. The
1938 * other bits are reserved for now.
1939 */
1940 #define EXT4_FT_UNKNOWN 0
1941 #define EXT4_FT_REG_FILE 1
1942 #define EXT4_FT_DIR 2
1943 #define EXT4_FT_CHRDEV 3
1944 #define EXT4_FT_BLKDEV 4
1945 #define EXT4_FT_FIFO 5
1946 #define EXT4_FT_SOCK 6
1947 #define EXT4_FT_SYMLINK 7
1948
1949 #define EXT4_FT_MAX 8
1950
1951 #define EXT4_FT_DIR_CSUM 0xDE
1952
1953 /*
1954 * EXT4_DIR_PAD defines the directory entries boundaries
1955 *
1956 * NOTE: It must be a multiple of 4
1957 */
1958 #define EXT4_DIR_PAD 4
1959 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1960 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1961 ~EXT4_DIR_ROUND)
1962 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1963
1964 /*
1965 * If we ever get support for fs block sizes > page_size, we'll need
1966 * to remove the #if statements in the next two functions...
1967 */
1968 static inline unsigned int
1969 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1970 {
1971 unsigned len = le16_to_cpu(dlen);
1972
1973 #if (PAGE_SIZE >= 65536)
1974 if (len == EXT4_MAX_REC_LEN || len == 0)
1975 return blocksize;
1976 return (len & 65532) | ((len & 3) << 16);
1977 #else
1978 return len;
1979 #endif
1980 }
1981
1982 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1983 {
1984 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1985 BUG();
1986 #if (PAGE_SIZE >= 65536)
1987 if (len < 65536)
1988 return cpu_to_le16(len);
1989 if (len == blocksize) {
1990 if (blocksize == 65536)
1991 return cpu_to_le16(EXT4_MAX_REC_LEN);
1992 else
1993 return cpu_to_le16(0);
1994 }
1995 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1996 #else
1997 return cpu_to_le16(len);
1998 #endif
1999 }
2000
2001 /*
2002 * Hash Tree Directory indexing
2003 * (c) Daniel Phillips, 2001
2004 */
2005
2006 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2007 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2008 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2009 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2010
2011 /* Legal values for the dx_root hash_version field: */
2012
2013 #define DX_HASH_LEGACY 0
2014 #define DX_HASH_HALF_MD4 1
2015 #define DX_HASH_TEA 2
2016 #define DX_HASH_LEGACY_UNSIGNED 3
2017 #define DX_HASH_HALF_MD4_UNSIGNED 4
2018 #define DX_HASH_TEA_UNSIGNED 5
2019
2020 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2021 const void *address, unsigned int length)
2022 {
2023 struct {
2024 struct shash_desc shash;
2025 char ctx[4];
2026 } desc;
2027 int err;
2028
2029 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2030
2031 desc.shash.tfm = sbi->s_chksum_driver;
2032 desc.shash.flags = 0;
2033 *(u32 *)desc.ctx = crc;
2034
2035 err = crypto_shash_update(&desc.shash, address, length);
2036 BUG_ON(err);
2037
2038 return *(u32 *)desc.ctx;
2039 }
2040
2041 #ifdef __KERNEL__
2042
2043 /* hash info structure used by the directory hash */
2044 struct dx_hash_info
2045 {
2046 u32 hash;
2047 u32 minor_hash;
2048 int hash_version;
2049 u32 *seed;
2050 };
2051
2052
2053 /* 32 and 64 bit signed EOF for dx directories */
2054 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2055 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2056
2057
2058 /*
2059 * Control parameters used by ext4_htree_next_block
2060 */
2061 #define HASH_NB_ALWAYS 1
2062
2063 struct ext4_filename {
2064 const struct qstr *usr_fname;
2065 struct fscrypt_str disk_name;
2066 struct dx_hash_info hinfo;
2067 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2068 struct fscrypt_str crypto_buf;
2069 #endif
2070 };
2071
2072 #define fname_name(p) ((p)->disk_name.name)
2073 #define fname_len(p) ((p)->disk_name.len)
2074
2075 /*
2076 * Describe an inode's exact location on disk and in memory
2077 */
2078 struct ext4_iloc
2079 {
2080 struct buffer_head *bh;
2081 unsigned long offset;
2082 ext4_group_t block_group;
2083 };
2084
2085 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2086 {
2087 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2088 }
2089
2090 /*
2091 * This structure is stuffed into the struct file's private_data field
2092 * for directories. It is where we put information so that we can do
2093 * readdir operations in hash tree order.
2094 */
2095 struct dir_private_info {
2096 struct rb_root root;
2097 struct rb_node *curr_node;
2098 struct fname *extra_fname;
2099 loff_t last_pos;
2100 __u32 curr_hash;
2101 __u32 curr_minor_hash;
2102 __u32 next_hash;
2103 };
2104
2105 /* calculate the first block number of the group */
2106 static inline ext4_fsblk_t
2107 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2108 {
2109 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2110 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2111 }
2112
2113 /*
2114 * Special error return code only used by dx_probe() and its callers.
2115 */
2116 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2117
2118 /*
2119 * Timeout and state flag for lazy initialization inode thread.
2120 */
2121 #define EXT4_DEF_LI_WAIT_MULT 10
2122 #define EXT4_DEF_LI_MAX_START_DELAY 5
2123 #define EXT4_LAZYINIT_QUIT 0x0001
2124 #define EXT4_LAZYINIT_RUNNING 0x0002
2125
2126 /*
2127 * Lazy inode table initialization info
2128 */
2129 struct ext4_lazy_init {
2130 unsigned long li_state;
2131 struct list_head li_request_list;
2132 struct mutex li_list_mtx;
2133 };
2134
2135 struct ext4_li_request {
2136 struct super_block *lr_super;
2137 struct ext4_sb_info *lr_sbi;
2138 ext4_group_t lr_next_group;
2139 struct list_head lr_request;
2140 unsigned long lr_next_sched;
2141 unsigned long lr_timeout;
2142 };
2143
2144 struct ext4_features {
2145 struct kobject f_kobj;
2146 struct completion f_kobj_unregister;
2147 };
2148
2149 /*
2150 * This structure will be used for multiple mount protection. It will be
2151 * written into the block number saved in the s_mmp_block field in the
2152 * superblock. Programs that check MMP should assume that if
2153 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2154 * to use the filesystem, regardless of how old the timestamp is.
2155 */
2156 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2157 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2158 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2159 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2160
2161 struct mmp_struct {
2162 __le32 mmp_magic; /* Magic number for MMP */
2163 __le32 mmp_seq; /* Sequence no. updated periodically */
2164
2165 /*
2166 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2167 * purposes and do not affect the correctness of the algorithm
2168 */
2169 __le64 mmp_time; /* Time last updated */
2170 char mmp_nodename[64]; /* Node which last updated MMP block */
2171 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2172
2173 /*
2174 * mmp_check_interval is used to verify if the MMP block has been
2175 * updated on the block device. The value is updated based on the
2176 * maximum time to write the MMP block during an update cycle.
2177 */
2178 __le16 mmp_check_interval;
2179
2180 __le16 mmp_pad1;
2181 __le32 mmp_pad2[226];
2182 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2183 };
2184
2185 /* arguments passed to the mmp thread */
2186 struct mmpd_data {
2187 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2188 struct super_block *sb; /* super block of the fs */
2189 };
2190
2191 /*
2192 * Check interval multiplier
2193 * The MMP block is written every update interval and initially checked every
2194 * update interval x the multiplier (the value is then adapted based on the
2195 * write latency). The reason is that writes can be delayed under load and we
2196 * don't want readers to incorrectly assume that the filesystem is no longer
2197 * in use.
2198 */
2199 #define EXT4_MMP_CHECK_MULT 2UL
2200
2201 /*
2202 * Minimum interval for MMP checking in seconds.
2203 */
2204 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2205
2206 /*
2207 * Maximum interval for MMP checking in seconds.
2208 */
2209 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2210
2211 /*
2212 * Function prototypes
2213 */
2214
2215 /*
2216 * Ok, these declarations are also in <linux/kernel.h> but none of the
2217 * ext4 source programs needs to include it so they are duplicated here.
2218 */
2219 # define NORET_TYPE /**/
2220 # define ATTRIB_NORET __attribute__((noreturn))
2221 # define NORET_AND noreturn,
2222
2223 /* bitmap.c */
2224 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2225 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2226 struct ext4_group_desc *gdp,
2227 struct buffer_head *bh, int sz);
2228 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2229 struct ext4_group_desc *gdp,
2230 struct buffer_head *bh, int sz);
2231 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2232 struct ext4_group_desc *gdp,
2233 struct buffer_head *bh);
2234 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2235 struct ext4_group_desc *gdp,
2236 struct buffer_head *bh);
2237
2238 /* balloc.c */
2239 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2240 ext4_fsblk_t blocknr,
2241 ext4_group_t *blockgrpp,
2242 ext4_grpblk_t *offsetp);
2243 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2244 ext4_fsblk_t block);
2245
2246 extern unsigned int ext4_block_group(struct super_block *sb,
2247 ext4_fsblk_t blocknr);
2248 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2249 ext4_fsblk_t blocknr);
2250 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2251 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2252 ext4_group_t group);
2253 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2254 ext4_fsblk_t goal,
2255 unsigned int flags,
2256 unsigned long *count,
2257 int *errp);
2258 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2259 s64 nclusters, unsigned int flags);
2260 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2261 extern void ext4_check_blocks_bitmap(struct super_block *);
2262 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2263 ext4_group_t block_group,
2264 struct buffer_head ** bh);
2265 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2266
2267 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2268 ext4_group_t block_group);
2269 extern int ext4_wait_block_bitmap(struct super_block *sb,
2270 ext4_group_t block_group,
2271 struct buffer_head *bh);
2272 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2273 ext4_group_t block_group);
2274 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2275 ext4_group_t block_group,
2276 struct ext4_group_desc *gdp);
2277 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2278
2279 static inline int ext4_sb_has_crypto(struct super_block *sb)
2280 {
2281 return ext4_has_feature_encrypt(sb);
2282 }
2283
2284 static inline bool ext4_encrypted_inode(struct inode *inode)
2285 {
2286 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2287 }
2288
2289 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2290 static inline int ext4_fname_setup_filename(struct inode *dir,
2291 const struct qstr *iname,
2292 int lookup, struct ext4_filename *fname)
2293 {
2294 struct fscrypt_name name;
2295 int err;
2296
2297 memset(fname, 0, sizeof(struct ext4_filename));
2298
2299 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2300
2301 fname->usr_fname = name.usr_fname;
2302 fname->disk_name = name.disk_name;
2303 fname->hinfo.hash = name.hash;
2304 fname->hinfo.minor_hash = name.minor_hash;
2305 fname->crypto_buf = name.crypto_buf;
2306 return err;
2307 }
2308
2309 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2310 {
2311 struct fscrypt_name name;
2312
2313 name.crypto_buf = fname->crypto_buf;
2314 fscrypt_free_filename(&name);
2315
2316 fname->crypto_buf.name = NULL;
2317 fname->usr_fname = NULL;
2318 fname->disk_name.name = NULL;
2319 }
2320 #else
2321 static inline int ext4_fname_setup_filename(struct inode *dir,
2322 const struct qstr *iname,
2323 int lookup, struct ext4_filename *fname)
2324 {
2325 fname->usr_fname = iname;
2326 fname->disk_name.name = (unsigned char *) iname->name;
2327 fname->disk_name.len = iname->len;
2328 return 0;
2329 }
2330 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2331
2332 #define fscrypt_set_d_op(i)
2333 #define fscrypt_get_ctx fscrypt_notsupp_get_ctx
2334 #define fscrypt_release_ctx fscrypt_notsupp_release_ctx
2335 #define fscrypt_encrypt_page fscrypt_notsupp_encrypt_page
2336 #define fscrypt_decrypt_page fscrypt_notsupp_decrypt_page
2337 #define fscrypt_decrypt_bio_pages fscrypt_notsupp_decrypt_bio_pages
2338 #define fscrypt_pullback_bio_page fscrypt_notsupp_pullback_bio_page
2339 #define fscrypt_restore_control_page fscrypt_notsupp_restore_control_page
2340 #define fscrypt_zeroout_range fscrypt_notsupp_zeroout_range
2341 #define fscrypt_process_policy fscrypt_notsupp_process_policy
2342 #define fscrypt_get_policy fscrypt_notsupp_get_policy
2343 #define fscrypt_has_permitted_context fscrypt_notsupp_has_permitted_context
2344 #define fscrypt_inherit_context fscrypt_notsupp_inherit_context
2345 #define fscrypt_get_encryption_info fscrypt_notsupp_get_encryption_info
2346 #define fscrypt_put_encryption_info fscrypt_notsupp_put_encryption_info
2347 #define fscrypt_setup_filename fscrypt_notsupp_setup_filename
2348 #define fscrypt_free_filename fscrypt_notsupp_free_filename
2349 #define fscrypt_fname_encrypted_size fscrypt_notsupp_fname_encrypted_size
2350 #define fscrypt_fname_alloc_buffer fscrypt_notsupp_fname_alloc_buffer
2351 #define fscrypt_fname_free_buffer fscrypt_notsupp_fname_free_buffer
2352 #define fscrypt_fname_disk_to_usr fscrypt_notsupp_fname_disk_to_usr
2353 #define fscrypt_fname_usr_to_disk fscrypt_notsupp_fname_usr_to_disk
2354 #endif
2355
2356 /* dir.c */
2357 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2358 struct file *,
2359 struct ext4_dir_entry_2 *,
2360 struct buffer_head *, char *, int,
2361 unsigned int);
2362 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2363 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2364 (de), (bh), (buf), (size), (offset)))
2365 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2366 __u32 minor_hash,
2367 struct ext4_dir_entry_2 *dirent,
2368 struct fscrypt_str *ent_name);
2369 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2370 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2371 struct buffer_head *bh,
2372 void *buf, int buf_size,
2373 struct ext4_filename *fname,
2374 struct ext4_dir_entry_2 **dest_de);
2375 int ext4_insert_dentry(struct inode *dir,
2376 struct inode *inode,
2377 struct ext4_dir_entry_2 *de,
2378 int buf_size,
2379 struct ext4_filename *fname);
2380 static inline void ext4_update_dx_flag(struct inode *inode)
2381 {
2382 if (!ext4_has_feature_dir_index(inode->i_sb))
2383 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2384 }
2385 static unsigned char ext4_filetype_table[] = {
2386 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2387 };
2388
2389 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2390 {
2391 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2392 return DT_UNKNOWN;
2393
2394 return ext4_filetype_table[filetype];
2395 }
2396 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2397 void *buf, int buf_size);
2398
2399 /* fsync.c */
2400 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2401
2402 /* hash.c */
2403 extern int ext4fs_dirhash(const char *name, int len, struct
2404 dx_hash_info *hinfo);
2405
2406 /* ialloc.c */
2407 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2408 const struct qstr *qstr, __u32 goal,
2409 uid_t *owner, int handle_type,
2410 unsigned int line_no, int nblocks);
2411
2412 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2413 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2414 0, 0, 0)
2415 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2416 type, nblocks) \
2417 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2418 (type), __LINE__, (nblocks))
2419
2420
2421 extern void ext4_free_inode(handle_t *, struct inode *);
2422 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2423 extern unsigned long ext4_count_free_inodes(struct super_block *);
2424 extern unsigned long ext4_count_dirs(struct super_block *);
2425 extern void ext4_check_inodes_bitmap(struct super_block *);
2426 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2427 extern int ext4_init_inode_table(struct super_block *sb,
2428 ext4_group_t group, int barrier);
2429 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2430
2431 /* mballoc.c */
2432 extern const struct file_operations ext4_seq_mb_groups_fops;
2433 extern long ext4_mb_stats;
2434 extern long ext4_mb_max_to_scan;
2435 extern int ext4_mb_init(struct super_block *);
2436 extern int ext4_mb_release(struct super_block *);
2437 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2438 struct ext4_allocation_request *, int *);
2439 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2440 extern void ext4_discard_preallocations(struct inode *);
2441 extern int __init ext4_init_mballoc(void);
2442 extern void ext4_exit_mballoc(void);
2443 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2444 struct buffer_head *bh, ext4_fsblk_t block,
2445 unsigned long count, int flags);
2446 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2447 ext4_group_t ngroups);
2448 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2449 ext4_group_t i, struct ext4_group_desc *desc);
2450 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2451 ext4_fsblk_t block, unsigned long count);
2452 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2453
2454 /* inode.c */
2455 int ext4_inode_is_fast_symlink(struct inode *inode);
2456 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2457 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2458 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2459 struct buffer_head *bh_result, int create);
2460 int ext4_dax_get_block(struct inode *inode, sector_t iblock,
2461 struct buffer_head *bh_result, int create);
2462 int ext4_get_block(struct inode *inode, sector_t iblock,
2463 struct buffer_head *bh_result, int create);
2464 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2465 struct buffer_head *bh_result, int create);
2466 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2467 struct buffer_head *bh, int create);
2468 int ext4_walk_page_buffers(handle_t *handle,
2469 struct buffer_head *head,
2470 unsigned from,
2471 unsigned to,
2472 int *partial,
2473 int (*fn)(handle_t *handle,
2474 struct buffer_head *bh));
2475 int do_journal_get_write_access(handle_t *handle,
2476 struct buffer_head *bh);
2477 #define FALL_BACK_TO_NONDELALLOC 1
2478 #define CONVERT_INLINE_DATA 2
2479
2480 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2481 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2482 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2483 extern int ext4_setattr(struct dentry *, struct iattr *);
2484 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2485 struct kstat *stat);
2486 extern void ext4_evict_inode(struct inode *);
2487 extern void ext4_clear_inode(struct inode *);
2488 extern int ext4_sync_inode(handle_t *, struct inode *);
2489 extern void ext4_dirty_inode(struct inode *, int);
2490 extern int ext4_change_inode_journal_flag(struct inode *, int);
2491 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2492 extern int ext4_inode_attach_jinode(struct inode *inode);
2493 extern int ext4_can_truncate(struct inode *inode);
2494 extern void ext4_truncate(struct inode *);
2495 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2496 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2497 extern void ext4_set_inode_flags(struct inode *);
2498 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2499 extern int ext4_alloc_da_blocks(struct inode *inode);
2500 extern void ext4_set_aops(struct inode *inode);
2501 extern int ext4_writepage_trans_blocks(struct inode *);
2502 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2503 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2504 loff_t lstart, loff_t lend);
2505 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2506 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2507 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2508 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2509 extern void ext4_da_update_reserve_space(struct inode *inode,
2510 int used, int quota_claim);
2511 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2512 ext4_fsblk_t pblk, ext4_lblk_t len);
2513 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2514 unsigned int map_len,
2515 struct extent_status *result);
2516
2517 /* indirect.c */
2518 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2519 struct ext4_map_blocks *map, int flags);
2520 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2521 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2522 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2523 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2524 ext4_lblk_t start, ext4_lblk_t end);
2525
2526 /* ioctl.c */
2527 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2528 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2529
2530 /* migrate.c */
2531 extern int ext4_ext_migrate(struct inode *);
2532 extern int ext4_ind_migrate(struct inode *inode);
2533
2534 /* namei.c */
2535 extern int ext4_dirent_csum_verify(struct inode *inode,
2536 struct ext4_dir_entry *dirent);
2537 extern int ext4_orphan_add(handle_t *, struct inode *);
2538 extern int ext4_orphan_del(handle_t *, struct inode *);
2539 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2540 __u32 start_minor_hash, __u32 *next_hash);
2541 extern int ext4_search_dir(struct buffer_head *bh,
2542 char *search_buf,
2543 int buf_size,
2544 struct inode *dir,
2545 struct ext4_filename *fname,
2546 const struct qstr *d_name,
2547 unsigned int offset,
2548 struct ext4_dir_entry_2 **res_dir);
2549 extern int ext4_generic_delete_entry(handle_t *handle,
2550 struct inode *dir,
2551 struct ext4_dir_entry_2 *de_del,
2552 struct buffer_head *bh,
2553 void *entry_buf,
2554 int buf_size,
2555 int csum_size);
2556 extern bool ext4_empty_dir(struct inode *inode);
2557
2558 /* resize.c */
2559 extern int ext4_group_add(struct super_block *sb,
2560 struct ext4_new_group_data *input);
2561 extern int ext4_group_extend(struct super_block *sb,
2562 struct ext4_super_block *es,
2563 ext4_fsblk_t n_blocks_count);
2564 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2565
2566 /* super.c */
2567 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2568 extern int ext4_calculate_overhead(struct super_block *sb);
2569 extern void ext4_superblock_csum_set(struct super_block *sb);
2570 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2571 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2572 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2573 ext4_group_t ngroup);
2574 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2575 char nbuf[16]);
2576
2577 extern __printf(4, 5)
2578 void __ext4_error(struct super_block *, const char *, unsigned int,
2579 const char *, ...);
2580 extern __printf(5, 6)
2581 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2582 const char *, ...);
2583 extern __printf(5, 6)
2584 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2585 const char *, ...);
2586 extern void __ext4_std_error(struct super_block *, const char *,
2587 unsigned int, int);
2588 extern __printf(4, 5)
2589 void __ext4_abort(struct super_block *, const char *, unsigned int,
2590 const char *, ...);
2591 extern __printf(4, 5)
2592 void __ext4_warning(struct super_block *, const char *, unsigned int,
2593 const char *, ...);
2594 extern __printf(4, 5)
2595 void __ext4_warning_inode(const struct inode *inode, const char *function,
2596 unsigned int line, const char *fmt, ...);
2597 extern __printf(3, 4)
2598 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2599 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2600 const char *, unsigned int, const char *);
2601 extern __printf(7, 8)
2602 void __ext4_grp_locked_error(const char *, unsigned int,
2603 struct super_block *, ext4_group_t,
2604 unsigned long, ext4_fsblk_t,
2605 const char *, ...);
2606
2607 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2608 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2609
2610 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2611 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2612
2613 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2614 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2615
2616 #ifdef CONFIG_PRINTK
2617
2618 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2619 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2620 #define ext4_error_file(file, func, line, block, fmt, ...) \
2621 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2622 #define ext4_error(sb, fmt, ...) \
2623 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2624 #define ext4_abort(sb, fmt, ...) \
2625 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2626 #define ext4_warning(sb, fmt, ...) \
2627 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2628 #define ext4_warning_inode(inode, fmt, ...) \
2629 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2630 #define ext4_msg(sb, level, fmt, ...) \
2631 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2632 #define dump_mmp_msg(sb, mmp, msg) \
2633 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2634 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2635 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2636 fmt, ##__VA_ARGS__)
2637
2638 #else
2639
2640 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2641 do { \
2642 no_printk(fmt, ##__VA_ARGS__); \
2643 __ext4_error_inode(inode, "", 0, block, " "); \
2644 } while (0)
2645 #define ext4_error_file(file, func, line, block, fmt, ...) \
2646 do { \
2647 no_printk(fmt, ##__VA_ARGS__); \
2648 __ext4_error_file(file, "", 0, block, " "); \
2649 } while (0)
2650 #define ext4_error(sb, fmt, ...) \
2651 do { \
2652 no_printk(fmt, ##__VA_ARGS__); \
2653 __ext4_error(sb, "", 0, " "); \
2654 } while (0)
2655 #define ext4_abort(sb, fmt, ...) \
2656 do { \
2657 no_printk(fmt, ##__VA_ARGS__); \
2658 __ext4_abort(sb, "", 0, " "); \
2659 } while (0)
2660 #define ext4_warning(sb, fmt, ...) \
2661 do { \
2662 no_printk(fmt, ##__VA_ARGS__); \
2663 __ext4_warning(sb, "", 0, " "); \
2664 } while (0)
2665 #define ext4_warning_inode(inode, fmt, ...) \
2666 do { \
2667 no_printk(fmt, ##__VA_ARGS__); \
2668 __ext4_warning_inode(inode, "", 0, " "); \
2669 } while (0)
2670 #define ext4_msg(sb, level, fmt, ...) \
2671 do { \
2672 no_printk(fmt, ##__VA_ARGS__); \
2673 __ext4_msg(sb, "", " "); \
2674 } while (0)
2675 #define dump_mmp_msg(sb, mmp, msg) \
2676 __dump_mmp_msg(sb, mmp, "", 0, "")
2677 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2678 do { \
2679 no_printk(fmt, ##__VA_ARGS__); \
2680 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2681 } while (0)
2682
2683 #endif
2684
2685 extern void ext4_update_dynamic_rev(struct super_block *sb);
2686 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2687 __u32 compat);
2688 extern int ext4_update_rocompat_feature(handle_t *handle,
2689 struct super_block *sb, __u32 rocompat);
2690 extern int ext4_update_incompat_feature(handle_t *handle,
2691 struct super_block *sb, __u32 incompat);
2692 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2693 struct ext4_group_desc *bg);
2694 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2695 struct ext4_group_desc *bg);
2696 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2697 struct ext4_group_desc *bg);
2698 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2699 struct ext4_group_desc *bg);
2700 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2701 struct ext4_group_desc *bg);
2702 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2703 struct ext4_group_desc *bg);
2704 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2705 struct ext4_group_desc *bg);
2706 extern void ext4_block_bitmap_set(struct super_block *sb,
2707 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2708 extern void ext4_inode_bitmap_set(struct super_block *sb,
2709 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2710 extern void ext4_inode_table_set(struct super_block *sb,
2711 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2712 extern void ext4_free_group_clusters_set(struct super_block *sb,
2713 struct ext4_group_desc *bg,
2714 __u32 count);
2715 extern void ext4_free_inodes_set(struct super_block *sb,
2716 struct ext4_group_desc *bg, __u32 count);
2717 extern void ext4_used_dirs_set(struct super_block *sb,
2718 struct ext4_group_desc *bg, __u32 count);
2719 extern void ext4_itable_unused_set(struct super_block *sb,
2720 struct ext4_group_desc *bg, __u32 count);
2721 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2722 struct ext4_group_desc *gdp);
2723 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2724 struct ext4_group_desc *gdp);
2725 extern int ext4_register_li_request(struct super_block *sb,
2726 ext4_group_t first_not_zeroed);
2727
2728 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2729 {
2730 return ext4_has_feature_gdt_csum(sb) ||
2731 EXT4_SB(sb)->s_chksum_driver != NULL;
2732 }
2733
2734 static inline int ext4_has_metadata_csum(struct super_block *sb)
2735 {
2736 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2737 !EXT4_SB(sb)->s_chksum_driver);
2738
2739 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2740 }
2741 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2742 {
2743 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2744 le32_to_cpu(es->s_blocks_count_lo);
2745 }
2746
2747 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2748 {
2749 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2750 le32_to_cpu(es->s_r_blocks_count_lo);
2751 }
2752
2753 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2754 {
2755 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2756 le32_to_cpu(es->s_free_blocks_count_lo);
2757 }
2758
2759 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2760 ext4_fsblk_t blk)
2761 {
2762 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2763 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2764 }
2765
2766 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2767 ext4_fsblk_t blk)
2768 {
2769 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2770 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2771 }
2772
2773 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2774 ext4_fsblk_t blk)
2775 {
2776 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2777 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2778 }
2779
2780 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2781 {
2782 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2783 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2784 le32_to_cpu(raw_inode->i_size_lo);
2785 else
2786 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2787 }
2788
2789 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2790 {
2791 raw_inode->i_size_lo = cpu_to_le32(i_size);
2792 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2793 }
2794
2795 static inline
2796 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2797 ext4_group_t group)
2798 {
2799 struct ext4_group_info ***grp_info;
2800 long indexv, indexh;
2801 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2802 grp_info = EXT4_SB(sb)->s_group_info;
2803 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2804 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2805 return grp_info[indexv][indexh];
2806 }
2807
2808 /*
2809 * Reading s_groups_count requires using smp_rmb() afterwards. See
2810 * the locking protocol documented in the comments of ext4_group_add()
2811 * in resize.c
2812 */
2813 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2814 {
2815 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2816
2817 smp_rmb();
2818 return ngroups;
2819 }
2820
2821 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2822 ext4_group_t block_group)
2823 {
2824 return block_group >> sbi->s_log_groups_per_flex;
2825 }
2826
2827 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2828 {
2829 return 1 << sbi->s_log_groups_per_flex;
2830 }
2831
2832 #define ext4_std_error(sb, errno) \
2833 do { \
2834 if ((errno)) \
2835 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2836 } while (0)
2837
2838 #ifdef CONFIG_SMP
2839 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2840 * counters. So we need to make sure we have free clusters more
2841 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2842 */
2843 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2844 #else
2845 #define EXT4_FREECLUSTERS_WATERMARK 0
2846 #endif
2847
2848 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2849 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2850 {
2851 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2852 !inode_is_locked(inode));
2853 down_write(&EXT4_I(inode)->i_data_sem);
2854 if (newsize > EXT4_I(inode)->i_disksize)
2855 EXT4_I(inode)->i_disksize = newsize;
2856 up_write(&EXT4_I(inode)->i_data_sem);
2857 }
2858
2859 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2860 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2861 {
2862 int changed = 0;
2863
2864 if (newsize > inode->i_size) {
2865 i_size_write(inode, newsize);
2866 changed = 1;
2867 }
2868 if (newsize > EXT4_I(inode)->i_disksize) {
2869 ext4_update_i_disksize(inode, newsize);
2870 changed |= 2;
2871 }
2872 return changed;
2873 }
2874
2875 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2876 loff_t len);
2877
2878 struct ext4_group_info {
2879 unsigned long bb_state;
2880 struct rb_root bb_free_root;
2881 ext4_grpblk_t bb_first_free; /* first free block */
2882 ext4_grpblk_t bb_free; /* total free blocks */
2883 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2884 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2885 struct list_head bb_prealloc_list;
2886 #ifdef DOUBLE_CHECK
2887 void *bb_bitmap;
2888 #endif
2889 struct rw_semaphore alloc_sem;
2890 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2891 * regions, index is order.
2892 * bb_counters[3] = 5 means
2893 * 5 free 8-block regions. */
2894 };
2895
2896 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2897 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2898 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2899 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2900
2901 #define EXT4_MB_GRP_NEED_INIT(grp) \
2902 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2903 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2904 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2905 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2906 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2907
2908 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2909 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2910 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2911 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2912 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2913 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2914
2915 #define EXT4_MAX_CONTENTION 8
2916 #define EXT4_CONTENTION_THRESHOLD 2
2917
2918 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2919 ext4_group_t group)
2920 {
2921 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2922 }
2923
2924 /*
2925 * Returns true if the filesystem is busy enough that attempts to
2926 * access the block group locks has run into contention.
2927 */
2928 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2929 {
2930 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2931 }
2932
2933 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2934 {
2935 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2936 if (spin_trylock(lock))
2937 /*
2938 * We're able to grab the lock right away, so drop the
2939 * lock contention counter.
2940 */
2941 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2942 else {
2943 /*
2944 * The lock is busy, so bump the contention counter,
2945 * and then wait on the spin lock.
2946 */
2947 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2948 EXT4_MAX_CONTENTION);
2949 spin_lock(lock);
2950 }
2951 }
2952
2953 static inline void ext4_unlock_group(struct super_block *sb,
2954 ext4_group_t group)
2955 {
2956 spin_unlock(ext4_group_lock_ptr(sb, group));
2957 }
2958
2959 /*
2960 * Block validity checking
2961 */
2962 #define ext4_check_indirect_blockref(inode, bh) \
2963 ext4_check_blockref(__func__, __LINE__, inode, \
2964 (__le32 *)(bh)->b_data, \
2965 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2966
2967 #define ext4_ind_check_inode(inode) \
2968 ext4_check_blockref(__func__, __LINE__, inode, \
2969 EXT4_I(inode)->i_data, \
2970 EXT4_NDIR_BLOCKS)
2971
2972 /*
2973 * Inodes and files operations
2974 */
2975
2976 /* dir.c */
2977 extern const struct file_operations ext4_dir_operations;
2978
2979 /* file.c */
2980 extern const struct inode_operations ext4_file_inode_operations;
2981 extern const struct file_operations ext4_file_operations;
2982 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2983
2984 /* inline.c */
2985 extern int ext4_get_max_inline_size(struct inode *inode);
2986 extern int ext4_find_inline_data_nolock(struct inode *inode);
2987 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2988 unsigned int len);
2989 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2990
2991 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2992 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2993 struct inode *inode,
2994 loff_t pos, unsigned len,
2995 unsigned flags,
2996 struct page **pagep);
2997 extern int ext4_write_inline_data_end(struct inode *inode,
2998 loff_t pos, unsigned len,
2999 unsigned copied,
3000 struct page *page);
3001 extern struct buffer_head *
3002 ext4_journalled_write_inline_data(struct inode *inode,
3003 unsigned len,
3004 struct page *page);
3005 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3006 struct inode *inode,
3007 loff_t pos, unsigned len,
3008 unsigned flags,
3009 struct page **pagep,
3010 void **fsdata);
3011 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3012 unsigned len, unsigned copied,
3013 struct page *page);
3014 extern int ext4_try_add_inline_entry(handle_t *handle,
3015 struct ext4_filename *fname,
3016 struct inode *dir, struct inode *inode);
3017 extern int ext4_try_create_inline_dir(handle_t *handle,
3018 struct inode *parent,
3019 struct inode *inode);
3020 extern int ext4_read_inline_dir(struct file *filp,
3021 struct dir_context *ctx,
3022 int *has_inline_data);
3023 extern int htree_inlinedir_to_tree(struct file *dir_file,
3024 struct inode *dir, ext4_lblk_t block,
3025 struct dx_hash_info *hinfo,
3026 __u32 start_hash, __u32 start_minor_hash,
3027 int *has_inline_data);
3028 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3029 struct ext4_filename *fname,
3030 const struct qstr *d_name,
3031 struct ext4_dir_entry_2 **res_dir,
3032 int *has_inline_data);
3033 extern int ext4_delete_inline_entry(handle_t *handle,
3034 struct inode *dir,
3035 struct ext4_dir_entry_2 *de_del,
3036 struct buffer_head *bh,
3037 int *has_inline_data);
3038 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3039 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3040 struct ext4_dir_entry_2 **parent_de,
3041 int *retval);
3042 extern int ext4_inline_data_fiemap(struct inode *inode,
3043 struct fiemap_extent_info *fieinfo,
3044 int *has_inline, __u64 start, __u64 len);
3045 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3046 struct inode *inode,
3047 int needed);
3048 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3049
3050 extern int ext4_convert_inline_data(struct inode *inode);
3051
3052 static inline int ext4_has_inline_data(struct inode *inode)
3053 {
3054 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3055 EXT4_I(inode)->i_inline_off;
3056 }
3057
3058 /* namei.c */
3059 extern const struct inode_operations ext4_dir_inode_operations;
3060 extern const struct inode_operations ext4_special_inode_operations;
3061 extern struct dentry *ext4_get_parent(struct dentry *child);
3062 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3063 struct ext4_dir_entry_2 *de,
3064 int blocksize, int csum_size,
3065 unsigned int parent_ino, int dotdot_real_len);
3066 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3067 unsigned int blocksize);
3068 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3069 struct inode *inode,
3070 struct buffer_head *bh);
3071 #define S_SHIFT 12
3072 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3073 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3074 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3075 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3076 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3077 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3078 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3079 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3080 };
3081
3082 static inline void ext4_set_de_type(struct super_block *sb,
3083 struct ext4_dir_entry_2 *de,
3084 umode_t mode) {
3085 if (ext4_has_feature_filetype(sb))
3086 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3087 }
3088
3089 /* readpages.c */
3090 extern int ext4_mpage_readpages(struct address_space *mapping,
3091 struct list_head *pages, struct page *page,
3092 unsigned nr_pages);
3093
3094 /* symlink.c */
3095 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3096 extern const struct inode_operations ext4_symlink_inode_operations;
3097 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3098
3099 /* sysfs.c */
3100 extern int ext4_register_sysfs(struct super_block *sb);
3101 extern void ext4_unregister_sysfs(struct super_block *sb);
3102 extern int __init ext4_init_sysfs(void);
3103 extern void ext4_exit_sysfs(void);
3104
3105 /* block_validity */
3106 extern void ext4_release_system_zone(struct super_block *sb);
3107 extern int ext4_setup_system_zone(struct super_block *sb);
3108 extern int __init ext4_init_system_zone(void);
3109 extern void ext4_exit_system_zone(void);
3110 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3111 ext4_fsblk_t start_blk,
3112 unsigned int count);
3113 extern int ext4_check_blockref(const char *, unsigned int,
3114 struct inode *, __le32 *, unsigned int);
3115
3116 /* extents.c */
3117 struct ext4_ext_path;
3118 struct ext4_extent;
3119
3120 /*
3121 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3122 * __le32.
3123 */
3124 #define EXT_MAX_BLOCKS 0xffffffff
3125
3126 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3127 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3128 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3129 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3130 struct ext4_map_blocks *map, int flags);
3131 extern void ext4_ext_truncate(handle_t *, struct inode *);
3132 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3133 ext4_lblk_t end);
3134 extern void ext4_ext_init(struct super_block *);
3135 extern void ext4_ext_release(struct super_block *);
3136 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3137 loff_t len);
3138 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3139 loff_t offset, ssize_t len);
3140 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3141 struct ext4_map_blocks *map, int flags);
3142 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3143 ext4_lblk_t lblocks);
3144 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3145 int num,
3146 struct ext4_ext_path *path);
3147 extern int ext4_can_extents_be_merged(struct inode *inode,
3148 struct ext4_extent *ex1,
3149 struct ext4_extent *ex2);
3150 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3151 struct ext4_ext_path **,
3152 struct ext4_extent *, int);
3153 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3154 struct ext4_ext_path **,
3155 int flags);
3156 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3157 extern int ext4_ext_check_inode(struct inode *inode);
3158 extern int ext4_find_delalloc_range(struct inode *inode,
3159 ext4_lblk_t lblk_start,
3160 ext4_lblk_t lblk_end);
3161 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3162 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3163 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3164 __u64 start, __u64 len);
3165 extern int ext4_ext_precache(struct inode *inode);
3166 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3167 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3168 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3169 struct inode *inode2, ext4_lblk_t lblk1,
3170 ext4_lblk_t lblk2, ext4_lblk_t count,
3171 int mark_unwritten,int *err);
3172
3173 /* move_extent.c */
3174 extern void ext4_double_down_write_data_sem(struct inode *first,
3175 struct inode *second);
3176 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3177 struct inode *donor_inode);
3178 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3179 __u64 start_orig, __u64 start_donor,
3180 __u64 len, __u64 *moved_len);
3181
3182 /* page-io.c */
3183 extern int __init ext4_init_pageio(void);
3184 extern void ext4_exit_pageio(void);
3185 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3186 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3187 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3188 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3189 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3190 struct writeback_control *wbc);
3191 extern void ext4_end_io_rsv_work(struct work_struct *work);
3192 extern void ext4_io_submit(struct ext4_io_submit *io);
3193 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3194 struct page *page,
3195 int len,
3196 struct writeback_control *wbc,
3197 bool keep_towrite);
3198
3199 /* mmp.c */
3200 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3201
3202 /*
3203 * Add new method to test whether block and inode bitmaps are properly
3204 * initialized. With uninit_bg reading the block from disk is not enough
3205 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3206 */
3207 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3208
3209 static inline int bitmap_uptodate(struct buffer_head *bh)
3210 {
3211 return (buffer_uptodate(bh) &&
3212 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3213 }
3214 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3215 {
3216 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3217 }
3218
3219 /*
3220 * Disable DIO read nolock optimization, so new dioreaders will be forced
3221 * to grab i_mutex
3222 */
3223 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3224 {
3225 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3226 smp_mb();
3227 }
3228 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3229 {
3230 smp_mb();
3231 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3232 }
3233
3234 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3235
3236 /* For ioend & aio unwritten conversion wait queues */
3237 #define EXT4_WQ_HASH_SZ 37
3238 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3239 EXT4_WQ_HASH_SZ])
3240 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3241
3242 #define EXT4_RESIZING 0
3243 extern int ext4_resize_begin(struct super_block *sb);
3244 extern void ext4_resize_end(struct super_block *sb);
3245
3246 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3247 struct ext4_io_end *io_end)
3248 {
3249 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3250 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3251 atomic_inc(&EXT4_I(inode)->i_unwritten);
3252 }
3253 }
3254
3255 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3256 {
3257 struct inode *inode = io_end->inode;
3258
3259 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3260 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3261 /* Wake up anyone waiting on unwritten extent conversion */
3262 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3263 wake_up_all(ext4_ioend_wq(inode));
3264 }
3265 }
3266
3267 static inline bool ext4_aligned_io(struct inode *inode, loff_t off, loff_t len)
3268 {
3269 int blksize = 1 << inode->i_blkbits;
3270
3271 return IS_ALIGNED(off, blksize) && IS_ALIGNED(len, blksize);
3272 }
3273
3274 #endif /* __KERNEL__ */
3275
3276 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3277 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3278
3279 #endif /* _EXT4_H */