?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ethtool.h: define INT_MAX for userland
[linux/fpc-iii.git] / fs / ext4 / ext4.h
blob157b458a69d4b7c334f28b80c37fd5d11c9f0c25
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/falloc.h>
36 #ifdef __KERNEL__
37 #include <linux/compat.h>
38 #endif
39
40 /*
41 * The fourth extended filesystem constants/structures
42 */
43
44 /*
45 * Define EXT4FS_DEBUG to produce debug messages
46 */
47 #undef EXT4FS_DEBUG
48
49 /*
50 * Debug code
51 */
52 #ifdef EXT4FS_DEBUG
53 #define ext4_debug(f, a...) \
54 do { \
55 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
56 __FILE__, __LINE__, __func__); \
57 printk(KERN_DEBUG f, ## a); \
58 } while (0)
59 #else
60 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
61 #endif
62
63 /*
64 * Turn on EXT_DEBUG to get lots of info about extents operations.
65 */
66 #define EXT_DEBUG__
67 #ifdef EXT_DEBUG
68 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
69 #else
70 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
71 #endif
72
73 /* data type for block offset of block group */
74 typedef int ext4_grpblk_t;
75
76 /* data type for filesystem-wide blocks number */
77 typedef unsigned long long ext4_fsblk_t;
78
79 /* data type for file logical block number */
80 typedef __u32 ext4_lblk_t;
81
82 /* data type for block group number */
83 typedef unsigned int ext4_group_t;
84
85 enum SHIFT_DIRECTION {
86 SHIFT_LEFT = 0,
87 SHIFT_RIGHT,
88 };
89
90 /*
91 * Flags used in mballoc's allocation_context flags field.
92 *
93 * Also used to show what's going on for debugging purposes when the
94 * flag field is exported via the traceport interface
95 */
96
97 /* prefer goal again. length */
98 #define EXT4_MB_HINT_MERGE 0x0001
99 /* blocks already reserved */
100 #define EXT4_MB_HINT_RESERVED 0x0002
101 /* metadata is being allocated */
102 #define EXT4_MB_HINT_METADATA 0x0004
103 /* first blocks in the file */
104 #define EXT4_MB_HINT_FIRST 0x0008
105 /* search for the best chunk */
106 #define EXT4_MB_HINT_BEST 0x0010
107 /* data is being allocated */
108 #define EXT4_MB_HINT_DATA 0x0020
109 /* don't preallocate (for tails) */
110 #define EXT4_MB_HINT_NOPREALLOC 0x0040
111 /* allocate for locality group */
112 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
113 /* allocate goal blocks or none */
114 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
115 /* goal is meaningful */
116 #define EXT4_MB_HINT_TRY_GOAL 0x0200
117 /* blocks already pre-reserved by delayed allocation */
118 #define EXT4_MB_DELALLOC_RESERVED 0x0400
119 /* We are doing stream allocation */
120 #define EXT4_MB_STREAM_ALLOC 0x0800
121 /* Use reserved root blocks if needed */
122 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
123 /* Use blocks from reserved pool */
124 #define EXT4_MB_USE_RESERVED 0x2000
125
126 struct ext4_allocation_request {
127 /* target inode for block we're allocating */
128 struct inode *inode;
129 /* how many blocks we want to allocate */
130 unsigned int len;
131 /* logical block in target inode */
132 ext4_lblk_t logical;
133 /* the closest logical allocated block to the left */
134 ext4_lblk_t lleft;
135 /* the closest logical allocated block to the right */
136 ext4_lblk_t lright;
137 /* phys. target (a hint) */
138 ext4_fsblk_t goal;
139 /* phys. block for the closest logical allocated block to the left */
140 ext4_fsblk_t pleft;
141 /* phys. block for the closest logical allocated block to the right */
142 ext4_fsblk_t pright;
143 /* flags. see above EXT4_MB_HINT_* */
144 unsigned int flags;
145 };
146
147 /*
148 * Logical to physical block mapping, used by ext4_map_blocks()
149 *
150 * This structure is used to pass requests into ext4_map_blocks() as
151 * well as to store the information returned by ext4_map_blocks(). It
152 * takes less room on the stack than a struct buffer_head.
153 */
154 #define EXT4_MAP_NEW (1 << BH_New)
155 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
156 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
157 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
158 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
159 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
160
161 struct ext4_map_blocks {
162 ext4_fsblk_t m_pblk;
163 ext4_lblk_t m_lblk;
164 unsigned int m_len;
165 unsigned int m_flags;
166 };
167
168 /*
169 * Flags for ext4_io_end->flags
170 */
171 #define EXT4_IO_END_UNWRITTEN 0x0001
172
173 /*
174 * For converting unwritten extents on a work queue. 'handle' is used for
175 * buffered writeback.
176 */
177 typedef struct ext4_io_end {
178 struct list_head list; /* per-file finished IO list */
179 handle_t *handle; /* handle reserved for extent
180 * conversion */
181 struct inode *inode; /* file being written to */
182 struct bio *bio; /* Linked list of completed
183 * bios covering the extent */
184 unsigned int flag; /* unwritten or not */
185 loff_t offset; /* offset in the file */
186 ssize_t size; /* size of the extent */
187 atomic_t count; /* reference counter */
188 } ext4_io_end_t;
189
190 struct ext4_io_submit {
191 struct writeback_control *io_wbc;
192 struct bio *io_bio;
193 ext4_io_end_t *io_end;
194 sector_t io_next_block;
195 };
196
197 /*
198 * Special inodes numbers
199 */
200 #define EXT4_BAD_INO 1 /* Bad blocks inode */
201 #define EXT4_ROOT_INO 2 /* Root inode */
202 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
203 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
204 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
205 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
206 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
207 #define EXT4_JOURNAL_INO 8 /* Journal inode */
208
209 /* First non-reserved inode for old ext4 filesystems */
210 #define EXT4_GOOD_OLD_FIRST_INO 11
211
212 /*
213 * Maximal count of links to a file
214 */
215 #define EXT4_LINK_MAX 65000
216
217 /*
218 * Macro-instructions used to manage several block sizes
219 */
220 #define EXT4_MIN_BLOCK_SIZE 1024
221 #define EXT4_MAX_BLOCK_SIZE 65536
222 #define EXT4_MIN_BLOCK_LOG_SIZE 10
223 #define EXT4_MAX_BLOCK_LOG_SIZE 16
224 #ifdef __KERNEL__
225 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
226 #else
227 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
228 #endif
229 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
230 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
231 EXT4_SB(s)->s_cluster_bits)
232 #ifdef __KERNEL__
233 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
234 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
235 #else
236 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
237 #endif
238 #ifdef __KERNEL__
239 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
240 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
241 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
242 #else
243 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
244 EXT4_GOOD_OLD_INODE_SIZE : \
245 (s)->s_inode_size)
246 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
247 EXT4_GOOD_OLD_FIRST_INO : \
248 (s)->s_first_ino)
249 #endif
250 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
251
252 /* Translate a block number to a cluster number */
253 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
254 /* Translate a cluster number to a block number */
255 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
256 /* Translate # of blks to # of clusters */
257 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
258 (sbi)->s_cluster_bits)
259 /* Mask out the low bits to get the starting block of the cluster */
260 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
261 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
262 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
263 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
264 /* Get the cluster offset */
265 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
266 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
267 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
268 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
269
270 /*
271 * Structure of a blocks group descriptor
272 */
273 struct ext4_group_desc
274 {
275 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
276 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
277 __le32 bg_inode_table_lo; /* Inodes table block */
278 __le16 bg_free_blocks_count_lo;/* Free blocks count */
279 __le16 bg_free_inodes_count_lo;/* Free inodes count */
280 __le16 bg_used_dirs_count_lo; /* Directories count */
281 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
282 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
283 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
284 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
285 __le16 bg_itable_unused_lo; /* Unused inodes count */
286 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
287 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
288 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
289 __le32 bg_inode_table_hi; /* Inodes table block MSB */
290 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
291 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
292 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
293 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
294 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
295 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
296 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
297 __u32 bg_reserved;
298 };
299
300 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
301 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
302 sizeof(__le16))
303 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
304 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
305 sizeof(__le16))
306
307 /*
308 * Structure of a flex block group info
309 */
310
311 struct flex_groups {
312 atomic64_t free_clusters;
313 atomic_t free_inodes;
314 atomic_t used_dirs;
315 };
316
317 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
318 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
319 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
320
321 /*
322 * Macro-instructions used to manage group descriptors
323 */
324 #define EXT4_MIN_DESC_SIZE 32
325 #define EXT4_MIN_DESC_SIZE_64BIT 64
326 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
327 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
328 #ifdef __KERNEL__
329 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
330 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
331 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
332 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
333 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
334 #else
335 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
336 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
337 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
338 #endif
339
340 /*
341 * Constants relative to the data blocks
342 */
343 #define EXT4_NDIR_BLOCKS 12
344 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
345 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
346 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
347 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
348
349 /*
350 * Inode flags
351 */
352 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
353 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
354 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
355 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
356 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
357 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
358 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
359 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
360 /* Reserved for compression usage... */
361 #define EXT4_DIRTY_FL 0x00000100
362 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
363 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
364 /* nb: was previously EXT2_ECOMPR_FL */
365 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
366 /* End compression flags --- maybe not all used */
367 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
368 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
369 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
370 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
371 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
372 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
373 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
374 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
375 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
376 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
377 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
378 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
379 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
380
381 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
382 #define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
383
384 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
385 EXT4_IMMUTABLE_FL | \
386 EXT4_APPEND_FL | \
387 EXT4_NODUMP_FL | \
388 EXT4_NOATIME_FL | \
389 EXT4_PROJINHERIT_FL)
390
391 /* Flags that should be inherited by new inodes from their parent. */
392 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
393 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
394 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
395 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
396 EXT4_PROJINHERIT_FL)
397
398 /* Flags that are appropriate for regular files (all but dir-specific ones). */
399 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
400
401 /* Flags that are appropriate for non-directories/regular files. */
402 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
403
404 /* Mask out flags that are inappropriate for the given type of inode. */
405 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
406 {
407 if (S_ISDIR(mode))
408 return flags;
409 else if (S_ISREG(mode))
410 return flags & EXT4_REG_FLMASK;
411 else
412 return flags & EXT4_OTHER_FLMASK;
413 }
414
415 /*
416 * Inode flags used for atomic set/get
417 */
418 enum {
419 EXT4_INODE_SECRM = 0, /* Secure deletion */
420 EXT4_INODE_UNRM = 1, /* Undelete */
421 EXT4_INODE_COMPR = 2, /* Compress file */
422 EXT4_INODE_SYNC = 3, /* Synchronous updates */
423 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
424 EXT4_INODE_APPEND = 5, /* writes to file may only append */
425 EXT4_INODE_NODUMP = 6, /* do not dump file */
426 EXT4_INODE_NOATIME = 7, /* do not update atime */
427 /* Reserved for compression usage... */
428 EXT4_INODE_DIRTY = 8,
429 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
430 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
431 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
432 /* End compression flags --- maybe not all used */
433 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
434 EXT4_INODE_IMAGIC = 13, /* AFS directory */
435 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
436 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
437 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
438 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
439 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
440 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
441 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
442 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
443 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
444 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
445 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
446 };
447
448 /*
449 * Since it's pretty easy to mix up bit numbers and hex values, we use a
450 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
451 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
452 * any extra space in the compiled kernel image, otherwise, the build will fail.
453 * It's important that these values are the same, since we are using
454 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
455 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
456 * values found in ext2, ext3 and ext4 filesystems, and of course the values
457 * defined in e2fsprogs.
458 *
459 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
460 */
461 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
462 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
463
464 static inline void ext4_check_flag_values(void)
465 {
466 CHECK_FLAG_VALUE(SECRM);
467 CHECK_FLAG_VALUE(UNRM);
468 CHECK_FLAG_VALUE(COMPR);
469 CHECK_FLAG_VALUE(SYNC);
470 CHECK_FLAG_VALUE(IMMUTABLE);
471 CHECK_FLAG_VALUE(APPEND);
472 CHECK_FLAG_VALUE(NODUMP);
473 CHECK_FLAG_VALUE(NOATIME);
474 CHECK_FLAG_VALUE(DIRTY);
475 CHECK_FLAG_VALUE(COMPRBLK);
476 CHECK_FLAG_VALUE(NOCOMPR);
477 CHECK_FLAG_VALUE(ENCRYPT);
478 CHECK_FLAG_VALUE(INDEX);
479 CHECK_FLAG_VALUE(IMAGIC);
480 CHECK_FLAG_VALUE(JOURNAL_DATA);
481 CHECK_FLAG_VALUE(NOTAIL);
482 CHECK_FLAG_VALUE(DIRSYNC);
483 CHECK_FLAG_VALUE(TOPDIR);
484 CHECK_FLAG_VALUE(HUGE_FILE);
485 CHECK_FLAG_VALUE(EXTENTS);
486 CHECK_FLAG_VALUE(EA_INODE);
487 CHECK_FLAG_VALUE(EOFBLOCKS);
488 CHECK_FLAG_VALUE(INLINE_DATA);
489 CHECK_FLAG_VALUE(PROJINHERIT);
490 CHECK_FLAG_VALUE(RESERVED);
491 }
492
493 /* Used to pass group descriptor data when online resize is done */
494 struct ext4_new_group_input {
495 __u32 group; /* Group number for this data */
496 __u64 block_bitmap; /* Absolute block number of block bitmap */
497 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
498 __u64 inode_table; /* Absolute block number of inode table start */
499 __u32 blocks_count; /* Total number of blocks in this group */
500 __u16 reserved_blocks; /* Number of reserved blocks in this group */
501 __u16 unused;
502 };
503
504 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
505 struct compat_ext4_new_group_input {
506 u32 group;
507 compat_u64 block_bitmap;
508 compat_u64 inode_bitmap;
509 compat_u64 inode_table;
510 u32 blocks_count;
511 u16 reserved_blocks;
512 u16 unused;
513 };
514 #endif
515
516 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
517 struct ext4_new_group_data {
518 __u32 group;
519 __u64 block_bitmap;
520 __u64 inode_bitmap;
521 __u64 inode_table;
522 __u32 blocks_count;
523 __u16 reserved_blocks;
524 __u16 unused;
525 __u32 free_blocks_count;
526 };
527
528 /* Indexes used to index group tables in ext4_new_group_data */
529 enum {
530 BLOCK_BITMAP = 0, /* block bitmap */
531 INODE_BITMAP, /* inode bitmap */
532 INODE_TABLE, /* inode tables */
533 GROUP_TABLE_COUNT,
534 };
535
536 /*
537 * Flags used by ext4_map_blocks()
538 */
539 /* Allocate any needed blocks and/or convert an unwritten
540 extent to be an initialized ext4 */
541 #define EXT4_GET_BLOCKS_CREATE 0x0001
542 /* Request the creation of an unwritten extent */
543 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
544 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
545 EXT4_GET_BLOCKS_CREATE)
546 /* Caller is from the delayed allocation writeout path
547 * finally doing the actual allocation of delayed blocks */
548 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
549 /* caller is from the direct IO path, request to creation of an
550 unwritten extents if not allocated, split the unwritten
551 extent if blocks has been preallocated already*/
552 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
553 #define EXT4_GET_BLOCKS_CONVERT 0x0010
554 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
555 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
556 /* Convert extent to initialized after IO complete */
557 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
558 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
559 /* Eventual metadata allocation (due to growing extent tree)
560 * should not fail, so try to use reserved blocks for that.*/
561 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
562 /* Don't normalize allocation size (used for fallocate) */
563 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
564 /* Request will not result in inode size update (user for fallocate) */
565 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
566 /* Convert written extents to unwritten */
567 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
568 /* Write zeros to newly created written extents */
569 #define EXT4_GET_BLOCKS_ZERO 0x0200
570 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
571 EXT4_GET_BLOCKS_ZERO)
572
573 /*
574 * The bit position of these flags must not overlap with any of the
575 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
576 * read_extent_tree_block(), ext4_split_extent_at(),
577 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
578 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
579 * caching the extents when reading from the extent tree while a
580 * truncate or punch hole operation is in progress.
581 */
582 #define EXT4_EX_NOCACHE 0x40000000
583 #define EXT4_EX_FORCE_CACHE 0x20000000
584
585 /*
586 * Flags used by ext4_free_blocks
587 */
588 #define EXT4_FREE_BLOCKS_METADATA 0x0001
589 #define EXT4_FREE_BLOCKS_FORGET 0x0002
590 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
591 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
592 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
593 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
594
595 /* Encryption algorithms */
596 #define EXT4_ENCRYPTION_MODE_INVALID 0
597 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
598 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
599 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
600 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
601
602 #include "ext4_crypto.h"
603
604 /*
605 * ioctl commands
606 */
607 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
608 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
609 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
610 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
611 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
612 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
613 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
614 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
615 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
616 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
617 #define EXT4_IOC_MIGRATE _IO('f', 9)
618 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
619 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
620 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
621 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
622 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
623 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
624 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
625 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
626 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
627 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
628
629 #ifndef FS_IOC_FSGETXATTR
630 /* Until the uapi changes get merged for project quota... */
631
632 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
633 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
634
635 /*
636 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
637 */
638 struct fsxattr {
639 __u32 fsx_xflags; /* xflags field value (get/set) */
640 __u32 fsx_extsize; /* extsize field value (get/set)*/
641 __u32 fsx_nextents; /* nextents field value (get) */
642 __u32 fsx_projid; /* project identifier (get/set) */
643 unsigned char fsx_pad[12];
644 };
645
646 /*
647 * Flags for the fsx_xflags field
648 */
649 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
650 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
651 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
652 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
653 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
654 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
655 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
656 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
657 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
658 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
659 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
660 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
661 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
662 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
663 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
664 #endif /* !defined(FS_IOC_FSGETXATTR) */
665
666 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
667 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
668
669 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
670 /*
671 * ioctl commands in 32 bit emulation
672 */
673 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
674 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
675 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
676 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
677 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
678 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
679 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
680 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
681 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
682 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
683 #endif
684
685 /* Max physical block we can address w/o extents */
686 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
687
688 /*
689 * Structure of an inode on the disk
690 */
691 struct ext4_inode {
692 __le16 i_mode; /* File mode */
693 __le16 i_uid; /* Low 16 bits of Owner Uid */
694 __le32 i_size_lo; /* Size in bytes */
695 __le32 i_atime; /* Access time */
696 __le32 i_ctime; /* Inode Change time */
697 __le32 i_mtime; /* Modification time */
698 __le32 i_dtime; /* Deletion Time */
699 __le16 i_gid; /* Low 16 bits of Group Id */
700 __le16 i_links_count; /* Links count */
701 __le32 i_blocks_lo; /* Blocks count */
702 __le32 i_flags; /* File flags */
703 union {
704 struct {
705 __le32 l_i_version;
706 } linux1;
707 struct {
708 __u32 h_i_translator;
709 } hurd1;
710 struct {
711 __u32 m_i_reserved1;
712 } masix1;
713 } osd1; /* OS dependent 1 */
714 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
715 __le32 i_generation; /* File version (for NFS) */
716 __le32 i_file_acl_lo; /* File ACL */
717 __le32 i_size_high;
718 __le32 i_obso_faddr; /* Obsoleted fragment address */
719 union {
720 struct {
721 __le16 l_i_blocks_high; /* were l_i_reserved1 */
722 __le16 l_i_file_acl_high;
723 __le16 l_i_uid_high; /* these 2 fields */
724 __le16 l_i_gid_high; /* were reserved2[0] */
725 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
726 __le16 l_i_reserved;
727 } linux2;
728 struct {
729 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
730 __u16 h_i_mode_high;
731 __u16 h_i_uid_high;
732 __u16 h_i_gid_high;
733 __u32 h_i_author;
734 } hurd2;
735 struct {
736 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
737 __le16 m_i_file_acl_high;
738 __u32 m_i_reserved2[2];
739 } masix2;
740 } osd2; /* OS dependent 2 */
741 __le16 i_extra_isize;
742 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
743 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
744 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
745 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
746 __le32 i_crtime; /* File Creation time */
747 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
748 __le32 i_version_hi; /* high 32 bits for 64-bit version */
749 __le32 i_projid; /* Project ID */
750 };
751
752 struct move_extent {
753 __u32 reserved; /* should be zero */
754 __u32 donor_fd; /* donor file descriptor */
755 __u64 orig_start; /* logical start offset in block for orig */
756 __u64 donor_start; /* logical start offset in block for donor */
757 __u64 len; /* block length to be moved */
758 __u64 moved_len; /* moved block length */
759 };
760
761 #define EXT4_EPOCH_BITS 2
762 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
763 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
764
765 /*
766 * Extended fields will fit into an inode if the filesystem was formatted
767 * with large inodes (-I 256 or larger) and there are not currently any EAs
768 * consuming all of the available space. For new inodes we always reserve
769 * enough space for the kernel's known extended fields, but for inodes
770 * created with an old kernel this might not have been the case. None of
771 * the extended inode fields is critical for correct filesystem operation.
772 * This macro checks if a certain field fits in the inode. Note that
773 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
774 */
775 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
776 ((offsetof(typeof(*ext4_inode), field) + \
777 sizeof((ext4_inode)->field)) \
778 <= (EXT4_GOOD_OLD_INODE_SIZE + \
779 (einode)->i_extra_isize)) \
780
781 /*
782 * We use an encoding that preserves the times for extra epoch "00":
783 *
784 * extra msb of adjust for signed
785 * epoch 32-bit 32-bit tv_sec to
786 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
787 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
788 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
789 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
790 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
791 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
792 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
793 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
794 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
795 *
796 * Note that previous versions of the kernel on 64-bit systems would
797 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
798 * 1970. e2fsck will correct this, assuming that it is run on the
799 * affected filesystem before 2242.
800 */
801
802 static inline __le32 ext4_encode_extra_time(struct timespec *time)
803 {
804 u32 extra = sizeof(time->tv_sec) > 4 ?
805 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
806 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
807 }
808
809 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
810 {
811 if (unlikely(sizeof(time->tv_sec) > 4 &&
812 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
813 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
814 /* Handle legacy encoding of pre-1970 dates with epoch
815 * bits 1,1. We assume that by kernel version 4.20,
816 * everyone will have run fsck over the affected
817 * filesystems to correct the problem. (This
818 * backwards compatibility may be removed before this
819 * time, at the discretion of the ext4 developers.)
820 */
821 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
822 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
823 extra_bits = 0;
824 time->tv_sec += extra_bits << 32;
825 #else
826 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
827 #endif
828 }
829 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
830 }
831
832 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
833 do { \
834 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
835 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
836 (raw_inode)->xtime ## _extra = \
837 ext4_encode_extra_time(&(inode)->xtime); \
838 } while (0)
839
840 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
841 do { \
842 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
843 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
844 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
845 (raw_inode)->xtime ## _extra = \
846 ext4_encode_extra_time(&(einode)->xtime); \
847 } while (0)
848
849 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
850 do { \
851 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
852 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
853 ext4_decode_extra_time(&(inode)->xtime, \
854 raw_inode->xtime ## _extra); \
855 else \
856 (inode)->xtime.tv_nsec = 0; \
857 } while (0)
858
859 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
860 do { \
861 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
862 (einode)->xtime.tv_sec = \
863 (signed)le32_to_cpu((raw_inode)->xtime); \
864 else \
865 (einode)->xtime.tv_sec = 0; \
866 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
867 ext4_decode_extra_time(&(einode)->xtime, \
868 raw_inode->xtime ## _extra); \
869 else \
870 (einode)->xtime.tv_nsec = 0; \
871 } while (0)
872
873 #define i_disk_version osd1.linux1.l_i_version
874
875 #if defined(__KERNEL__) || defined(__linux__)
876 #define i_reserved1 osd1.linux1.l_i_reserved1
877 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
878 #define i_blocks_high osd2.linux2.l_i_blocks_high
879 #define i_uid_low i_uid
880 #define i_gid_low i_gid
881 #define i_uid_high osd2.linux2.l_i_uid_high
882 #define i_gid_high osd2.linux2.l_i_gid_high
883 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
884
885 #elif defined(__GNU__)
886
887 #define i_translator osd1.hurd1.h_i_translator
888 #define i_uid_high osd2.hurd2.h_i_uid_high
889 #define i_gid_high osd2.hurd2.h_i_gid_high
890 #define i_author osd2.hurd2.h_i_author
891
892 #elif defined(__masix__)
893
894 #define i_reserved1 osd1.masix1.m_i_reserved1
895 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
896 #define i_reserved2 osd2.masix2.m_i_reserved2
897
898 #endif /* defined(__KERNEL__) || defined(__linux__) */
899
900 #include "extents_status.h"
901
902 /*
903 * fourth extended file system inode data in memory
904 */
905 struct ext4_inode_info {
906 __le32 i_data[15]; /* unconverted */
907 __u32 i_dtime;
908 ext4_fsblk_t i_file_acl;
909
910 /*
911 * i_block_group is the number of the block group which contains
912 * this file's inode. Constant across the lifetime of the inode,
913 * it is ued for making block allocation decisions - we try to
914 * place a file's data blocks near its inode block, and new inodes
915 * near to their parent directory's inode.
916 */
917 ext4_group_t i_block_group;
918 ext4_lblk_t i_dir_start_lookup;
919 #if (BITS_PER_LONG < 64)
920 unsigned long i_state_flags; /* Dynamic state flags */
921 #endif
922 unsigned long i_flags;
923
924 /*
925 * Extended attributes can be read independently of the main file
926 * data. Taking i_mutex even when reading would cause contention
927 * between readers of EAs and writers of regular file data, so
928 * instead we synchronize on xattr_sem when reading or changing
929 * EAs.
930 */
931 struct rw_semaphore xattr_sem;
932
933 struct list_head i_orphan; /* unlinked but open inodes */
934
935 /*
936 * i_disksize keeps track of what the inode size is ON DISK, not
937 * in memory. During truncate, i_size is set to the new size by
938 * the VFS prior to calling ext4_truncate(), but the filesystem won't
939 * set i_disksize to 0 until the truncate is actually under way.
940 *
941 * The intent is that i_disksize always represents the blocks which
942 * are used by this file. This allows recovery to restart truncate
943 * on orphans if we crash during truncate. We actually write i_disksize
944 * into the on-disk inode when writing inodes out, instead of i_size.
945 *
946 * The only time when i_disksize and i_size may be different is when
947 * a truncate is in progress. The only things which change i_disksize
948 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
949 */
950 loff_t i_disksize;
951
952 /*
953 * i_data_sem is for serialising ext4_truncate() against
954 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
955 * data tree are chopped off during truncate. We can't do that in
956 * ext4 because whenever we perform intermediate commits during
957 * truncate, the inode and all the metadata blocks *must* be in a
958 * consistent state which allows truncation of the orphans to restart
959 * during recovery. Hence we must fix the get_block-vs-truncate race
960 * by other means, so we have i_data_sem.
961 */
962 struct rw_semaphore i_data_sem;
963 /*
964 * i_mmap_sem is for serializing page faults with truncate / punch hole
965 * operations. We have to make sure that new page cannot be faulted in
966 * a section of the inode that is being punched. We cannot easily use
967 * i_data_sem for this since we need protection for the whole punch
968 * operation and i_data_sem ranks below transaction start so we have
969 * to occasionally drop it.
970 */
971 struct rw_semaphore i_mmap_sem;
972 struct inode vfs_inode;
973 struct jbd2_inode *jinode;
974
975 spinlock_t i_raw_lock; /* protects updates to the raw inode */
976
977 /*
978 * File creation time. Its function is same as that of
979 * struct timespec i_{a,c,m}time in the generic inode.
980 */
981 struct timespec i_crtime;
982
983 /* mballoc */
984 struct list_head i_prealloc_list;
985 spinlock_t i_prealloc_lock;
986
987 /* extents status tree */
988 struct ext4_es_tree i_es_tree;
989 rwlock_t i_es_lock;
990 struct list_head i_es_list;
991 unsigned int i_es_all_nr; /* protected by i_es_lock */
992 unsigned int i_es_shk_nr; /* protected by i_es_lock */
993 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
994 extents to shrink. Protected by
995 i_es_lock */
996
997 /* ialloc */
998 ext4_group_t i_last_alloc_group;
999
1000 /* allocation reservation info for delalloc */
1001 /* In case of bigalloc, these refer to clusters rather than blocks */
1002 unsigned int i_reserved_data_blocks;
1003 unsigned int i_reserved_meta_blocks;
1004 unsigned int i_allocated_meta_blocks;
1005 ext4_lblk_t i_da_metadata_calc_last_lblock;
1006 int i_da_metadata_calc_len;
1007
1008 /* on-disk additional length */
1009 __u16 i_extra_isize;
1010
1011 /* Indicate the inline data space. */
1012 u16 i_inline_off;
1013 u16 i_inline_size;
1014
1015 #ifdef CONFIG_QUOTA
1016 /* quota space reservation, managed internally by quota code */
1017 qsize_t i_reserved_quota;
1018 #endif
1019
1020 /* Lock protecting lists below */
1021 spinlock_t i_completed_io_lock;
1022 /*
1023 * Completed IOs that need unwritten extents handling and have
1024 * transaction reserved
1025 */
1026 struct list_head i_rsv_conversion_list;
1027 /*
1028 * Completed IOs that need unwritten extents handling and don't have
1029 * transaction reserved
1030 */
1031 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
1032 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1033 struct work_struct i_rsv_conversion_work;
1034
1035 spinlock_t i_block_reservation_lock;
1036
1037 /*
1038 * Transactions that contain inode's metadata needed to complete
1039 * fsync and fdatasync, respectively.
1040 */
1041 tid_t i_sync_tid;
1042 tid_t i_datasync_tid;
1043
1044 #ifdef CONFIG_QUOTA
1045 struct dquot *i_dquot[MAXQUOTAS];
1046 #endif
1047
1048 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1049 __u32 i_csum_seed;
1050
1051 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1052 /* Encryption params */
1053 struct ext4_crypt_info *i_crypt_info;
1054 #endif
1055 kprojid_t i_projid;
1056 };
1057
1058 /*
1059 * File system states
1060 */
1061 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1062 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1063 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1064
1065 /*
1066 * Misc. filesystem flags
1067 */
1068 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1069 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1070 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1071
1072 /*
1073 * Mount flags set via mount options or defaults
1074 */
1075 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1076 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1077 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1078 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1079 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1080 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1081 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1082 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1083 #ifdef CONFIG_FS_DAX
1084 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1085 #else
1086 #define EXT4_MOUNT_DAX 0
1087 #endif
1088 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1089 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1090 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1091 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1092 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1093 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1094 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1095 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1096 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1097 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1098 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1099 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1100 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1101 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1102 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1103 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1104 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1105 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1106 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1107 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1108 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1109
1110 /*
1111 * Mount flags set either automatically (could not be set by mount option)
1112 * based on per file system feature or property or in special cases such as
1113 * distinguishing between explicit mount option definition and default.
1114 */
1115 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1116 specified delalloc */
1117 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1118 size of blocksize * 8
1119 blocks */
1120 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1121 file systems */
1122
1123 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1124 specified journal checksum */
1125
1126 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1127 ~EXT4_MOUNT_##opt
1128 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1129 EXT4_MOUNT_##opt
1130 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1131 EXT4_MOUNT_##opt)
1132
1133 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1134 ~EXT4_MOUNT2_##opt
1135 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1136 EXT4_MOUNT2_##opt
1137 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1138 EXT4_MOUNT2_##opt)
1139
1140 #define ext4_test_and_set_bit __test_and_set_bit_le
1141 #define ext4_set_bit __set_bit_le
1142 #define ext4_set_bit_atomic ext2_set_bit_atomic
1143 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1144 #define ext4_clear_bit __clear_bit_le
1145 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1146 #define ext4_test_bit test_bit_le
1147 #define ext4_find_next_zero_bit find_next_zero_bit_le
1148 #define ext4_find_next_bit find_next_bit_le
1149
1150 extern void ext4_set_bits(void *bm, int cur, int len);
1151
1152 /*
1153 * Maximal mount counts between two filesystem checks
1154 */
1155 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1156 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1157
1158 /*
1159 * Behaviour when detecting errors
1160 */
1161 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1162 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1163 #define EXT4_ERRORS_PANIC 3 /* Panic */
1164 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1165
1166 /* Metadata checksum algorithm codes */
1167 #define EXT4_CRC32C_CHKSUM 1
1168
1169 /*
1170 * Structure of the super block
1171 */
1172 struct ext4_super_block {
1173 /*00*/ __le32 s_inodes_count; /* Inodes count */
1174 __le32 s_blocks_count_lo; /* Blocks count */
1175 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1176 __le32 s_free_blocks_count_lo; /* Free blocks count */
1177 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1178 __le32 s_first_data_block; /* First Data Block */
1179 __le32 s_log_block_size; /* Block size */
1180 __le32 s_log_cluster_size; /* Allocation cluster size */
1181 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1182 __le32 s_clusters_per_group; /* # Clusters per group */
1183 __le32 s_inodes_per_group; /* # Inodes per group */
1184 __le32 s_mtime; /* Mount time */
1185 /*30*/ __le32 s_wtime; /* Write time */
1186 __le16 s_mnt_count; /* Mount count */
1187 __le16 s_max_mnt_count; /* Maximal mount count */
1188 __le16 s_magic; /* Magic signature */
1189 __le16 s_state; /* File system state */
1190 __le16 s_errors; /* Behaviour when detecting errors */
1191 __le16 s_minor_rev_level; /* minor revision level */
1192 /*40*/ __le32 s_lastcheck; /* time of last check */
1193 __le32 s_checkinterval; /* max. time between checks */
1194 __le32 s_creator_os; /* OS */
1195 __le32 s_rev_level; /* Revision level */
1196 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1197 __le16 s_def_resgid; /* Default gid for reserved blocks */
1198 /*
1199 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1200 *
1201 * Note: the difference between the compatible feature set and
1202 * the incompatible feature set is that if there is a bit set
1203 * in the incompatible feature set that the kernel doesn't
1204 * know about, it should refuse to mount the filesystem.
1205 *
1206 * e2fsck's requirements are more strict; if it doesn't know
1207 * about a feature in either the compatible or incompatible
1208 * feature set, it must abort and not try to meddle with
1209 * things it doesn't understand...
1210 */
1211 __le32 s_first_ino; /* First non-reserved inode */
1212 __le16 s_inode_size; /* size of inode structure */
1213 __le16 s_block_group_nr; /* block group # of this superblock */
1214 __le32 s_feature_compat; /* compatible feature set */
1215 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1216 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1217 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1218 /*78*/ char s_volume_name[16]; /* volume name */
1219 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1220 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1221 /*
1222 * Performance hints. Directory preallocation should only
1223 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1224 */
1225 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1226 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1227 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1228 /*
1229 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1230 */
1231 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1232 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1233 __le32 s_journal_dev; /* device number of journal file */
1234 __le32 s_last_orphan; /* start of list of inodes to delete */
1235 __le32 s_hash_seed[4]; /* HTREE hash seed */
1236 __u8 s_def_hash_version; /* Default hash version to use */
1237 __u8 s_jnl_backup_type;
1238 __le16 s_desc_size; /* size of group descriptor */
1239 /*100*/ __le32 s_default_mount_opts;
1240 __le32 s_first_meta_bg; /* First metablock block group */
1241 __le32 s_mkfs_time; /* When the filesystem was created */
1242 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1243 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1244 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1245 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1246 __le32 s_free_blocks_count_hi; /* Free blocks count */
1247 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1248 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1249 __le32 s_flags; /* Miscellaneous flags */
1250 __le16 s_raid_stride; /* RAID stride */
1251 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1252 __le64 s_mmp_block; /* Block for multi-mount protection */
1253 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1254 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1255 __u8 s_checksum_type; /* metadata checksum algorithm used */
1256 __u8 s_encryption_level; /* versioning level for encryption */
1257 __u8 s_reserved_pad; /* Padding to next 32bits */
1258 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1259 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1260 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1261 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1262 snapshot's future use */
1263 __le32 s_snapshot_list; /* inode number of the head of the
1264 on-disk snapshot list */
1265 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1266 __le32 s_error_count; /* number of fs errors */
1267 __le32 s_first_error_time; /* first time an error happened */
1268 __le32 s_first_error_ino; /* inode involved in first error */
1269 __le64 s_first_error_block; /* block involved of first error */
1270 __u8 s_first_error_func[32]; /* function where the error happened */
1271 __le32 s_first_error_line; /* line number where error happened */
1272 __le32 s_last_error_time; /* most recent time of an error */
1273 __le32 s_last_error_ino; /* inode involved in last error */
1274 __le32 s_last_error_line; /* line number where error happened */
1275 __le64 s_last_error_block; /* block involved of last error */
1276 __u8 s_last_error_func[32]; /* function where the error happened */
1277 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1278 __u8 s_mount_opts[64];
1279 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1280 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1281 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1282 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1283 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1284 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1285 __le32 s_lpf_ino; /* Location of the lost+found inode */
1286 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1287 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1288 __le32 s_reserved[98]; /* Padding to the end of the block */
1289 __le32 s_checksum; /* crc32c(superblock) */
1290 };
1291
1292 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1293
1294 #ifdef __KERNEL__
1295
1296 /*
1297 * run-time mount flags
1298 */
1299 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1300 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1301 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1302
1303 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1304 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1305 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1306 #else
1307 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1308 #endif
1309
1310 /* Number of quota types we support */
1311 #define EXT4_MAXQUOTAS 3
1312
1313 /*
1314 * fourth extended-fs super-block data in memory
1315 */
1316 struct ext4_sb_info {
1317 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1318 unsigned long s_inodes_per_block;/* Number of inodes per block */
1319 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1320 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1321 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1322 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1323 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1324 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1325 ext4_group_t s_groups_count; /* Number of groups in the fs */
1326 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1327 unsigned long s_overhead; /* # of fs overhead clusters */
1328 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1329 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1330 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1331 struct buffer_head * s_sbh; /* Buffer containing the super block */
1332 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1333 struct buffer_head **s_group_desc;
1334 unsigned int s_mount_opt;
1335 unsigned int s_mount_opt2;
1336 unsigned int s_mount_flags;
1337 unsigned int s_def_mount_opt;
1338 ext4_fsblk_t s_sb_block;
1339 atomic64_t s_resv_clusters;
1340 kuid_t s_resuid;
1341 kgid_t s_resgid;
1342 unsigned short s_mount_state;
1343 unsigned short s_pad;
1344 int s_addr_per_block_bits;
1345 int s_desc_per_block_bits;
1346 int s_inode_size;
1347 int s_first_ino;
1348 unsigned int s_inode_readahead_blks;
1349 unsigned int s_inode_goal;
1350 spinlock_t s_next_gen_lock;
1351 u32 s_next_generation;
1352 u32 s_hash_seed[4];
1353 int s_def_hash_version;
1354 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1355 struct percpu_counter s_freeclusters_counter;
1356 struct percpu_counter s_freeinodes_counter;
1357 struct percpu_counter s_dirs_counter;
1358 struct percpu_counter s_dirtyclusters_counter;
1359 struct blockgroup_lock *s_blockgroup_lock;
1360 struct proc_dir_entry *s_proc;
1361 struct kobject s_kobj;
1362 struct completion s_kobj_unregister;
1363 struct super_block *s_sb;
1364
1365 /* Journaling */
1366 struct journal_s *s_journal;
1367 struct list_head s_orphan;
1368 struct mutex s_orphan_lock;
1369 unsigned long s_resize_flags; /* Flags indicating if there
1370 is a resizer */
1371 unsigned long s_commit_interval;
1372 u32 s_max_batch_time;
1373 u32 s_min_batch_time;
1374 struct block_device *journal_bdev;
1375 #ifdef CONFIG_QUOTA
1376 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1377 int s_jquota_fmt; /* Format of quota to use */
1378 #endif
1379 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1380 struct rb_root system_blks;
1381
1382 #ifdef EXTENTS_STATS
1383 /* ext4 extents stats */
1384 unsigned long s_ext_min;
1385 unsigned long s_ext_max;
1386 unsigned long s_depth_max;
1387 spinlock_t s_ext_stats_lock;
1388 unsigned long s_ext_blocks;
1389 unsigned long s_ext_extents;
1390 #endif
1391
1392 /* for buddy allocator */
1393 struct ext4_group_info ***s_group_info;
1394 struct inode *s_buddy_cache;
1395 spinlock_t s_md_lock;
1396 unsigned short *s_mb_offsets;
1397 unsigned int *s_mb_maxs;
1398 unsigned int s_group_info_size;
1399
1400 /* tunables */
1401 unsigned long s_stripe;
1402 unsigned int s_mb_stream_request;
1403 unsigned int s_mb_max_to_scan;
1404 unsigned int s_mb_min_to_scan;
1405 unsigned int s_mb_stats;
1406 unsigned int s_mb_order2_reqs;
1407 unsigned int s_mb_group_prealloc;
1408 unsigned int s_max_dir_size_kb;
1409 /* where last allocation was done - for stream allocation */
1410 unsigned long s_mb_last_group;
1411 unsigned long s_mb_last_start;
1412
1413 /* stats for buddy allocator */
1414 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1415 atomic_t s_bal_success; /* we found long enough chunks */
1416 atomic_t s_bal_allocated; /* in blocks */
1417 atomic_t s_bal_ex_scanned; /* total extents scanned */
1418 atomic_t s_bal_goals; /* goal hits */
1419 atomic_t s_bal_breaks; /* too long searches */
1420 atomic_t s_bal_2orders; /* 2^order hits */
1421 spinlock_t s_bal_lock;
1422 unsigned long s_mb_buddies_generated;
1423 unsigned long long s_mb_generation_time;
1424 atomic_t s_mb_lost_chunks;
1425 atomic_t s_mb_preallocated;
1426 atomic_t s_mb_discarded;
1427 atomic_t s_lock_busy;
1428
1429 /* locality groups */
1430 struct ext4_locality_group __percpu *s_locality_groups;
1431
1432 /* for write statistics */
1433 unsigned long s_sectors_written_start;
1434 u64 s_kbytes_written;
1435
1436 /* the size of zero-out chunk */
1437 unsigned int s_extent_max_zeroout_kb;
1438
1439 unsigned int s_log_groups_per_flex;
1440 struct flex_groups *s_flex_groups;
1441 ext4_group_t s_flex_groups_allocated;
1442
1443 /* workqueue for reserved extent conversions (buffered io) */
1444 struct workqueue_struct *rsv_conversion_wq;
1445
1446 /* timer for periodic error stats printing */
1447 struct timer_list s_err_report;
1448
1449 /* Lazy inode table initialization info */
1450 struct ext4_li_request *s_li_request;
1451 /* Wait multiplier for lazy initialization thread */
1452 unsigned int s_li_wait_mult;
1453
1454 /* Kernel thread for multiple mount protection */
1455 struct task_struct *s_mmp_tsk;
1456
1457 /* record the last minlen when FITRIM is called. */
1458 atomic_t s_last_trim_minblks;
1459
1460 /* Reference to checksum algorithm driver via cryptoapi */
1461 struct crypto_shash *s_chksum_driver;
1462
1463 /* Precomputed FS UUID checksum for seeding other checksums */
1464 __u32 s_csum_seed;
1465
1466 /* Reclaim extents from extent status tree */
1467 struct shrinker s_es_shrinker;
1468 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1469 long s_es_nr_inode;
1470 struct ext4_es_stats s_es_stats;
1471 struct mb_cache *s_mb_cache;
1472 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1473
1474 /* Ratelimit ext4 messages. */
1475 struct ratelimit_state s_err_ratelimit_state;
1476 struct ratelimit_state s_warning_ratelimit_state;
1477 struct ratelimit_state s_msg_ratelimit_state;
1478 };
1479
1480 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1481 {
1482 return sb->s_fs_info;
1483 }
1484 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1485 {
1486 return container_of(inode, struct ext4_inode_info, vfs_inode);
1487 }
1488
1489 static inline struct timespec ext4_current_time(struct inode *inode)
1490 {
1491 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1492 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1493 }
1494
1495 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1496 {
1497 return ino == EXT4_ROOT_INO ||
1498 ino == EXT4_USR_QUOTA_INO ||
1499 ino == EXT4_GRP_QUOTA_INO ||
1500 ino == EXT4_BOOT_LOADER_INO ||
1501 ino == EXT4_JOURNAL_INO ||
1502 ino == EXT4_RESIZE_INO ||
1503 (ino >= EXT4_FIRST_INO(sb) &&
1504 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1505 }
1506
1507 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1508 struct ext4_io_end *io_end)
1509 {
1510 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1511 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1512 atomic_inc(&EXT4_I(inode)->i_unwritten);
1513 }
1514 }
1515
1516 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1517 {
1518 return inode->i_private;
1519 }
1520
1521 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1522 {
1523 inode->i_private = io;
1524 }
1525
1526 /*
1527 * Inode dynamic state flags
1528 */
1529 enum {
1530 EXT4_STATE_JDATA, /* journaled data exists */
1531 EXT4_STATE_NEW, /* inode is newly created */
1532 EXT4_STATE_XATTR, /* has in-inode xattrs */
1533 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1534 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1535 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1536 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1537 EXT4_STATE_NEWENTRY, /* File just added to dir */
1538 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1539 nolocking */
1540 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1541 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1542 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1543 };
1544
1545 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1546 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1547 { \
1548 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1549 } \
1550 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1551 { \
1552 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1553 } \
1554 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1555 { \
1556 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1557 }
1558
1559 /* Add these declarations here only so that these functions can be
1560 * found by name. Otherwise, they are very hard to locate. */
1561 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1562 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1563 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1564 EXT4_INODE_BIT_FNS(flag, flags, 0)
1565
1566 /* Add these declarations here only so that these functions can be
1567 * found by name. Otherwise, they are very hard to locate. */
1568 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1569 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1570 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1571 #if (BITS_PER_LONG < 64)
1572 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1573
1574 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1575 {
1576 (ei)->i_state_flags = 0;
1577 }
1578 #else
1579 EXT4_INODE_BIT_FNS(state, flags, 32)
1580
1581 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1582 {
1583 /* We depend on the fact that callers will set i_flags */
1584 }
1585 #endif
1586 #else
1587 /* Assume that user mode programs are passing in an ext4fs superblock, not
1588 * a kernel struct super_block. This will allow us to call the feature-test
1589 * macros from user land. */
1590 #define EXT4_SB(sb) (sb)
1591 #endif
1592
1593 /*
1594 * Returns true if the inode is inode is encrypted
1595 */
1596 static inline int ext4_encrypted_inode(struct inode *inode)
1597 {
1598 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1599 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1600 #else
1601 return 0;
1602 #endif
1603 }
1604
1605 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1606
1607 /*
1608 * Codes for operating systems
1609 */
1610 #define EXT4_OS_LINUX 0
1611 #define EXT4_OS_HURD 1
1612 #define EXT4_OS_MASIX 2
1613 #define EXT4_OS_FREEBSD 3
1614 #define EXT4_OS_LITES 4
1615
1616 /*
1617 * Revision levels
1618 */
1619 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1620 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1621
1622 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1623 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1624
1625 #define EXT4_GOOD_OLD_INODE_SIZE 128
1626
1627 /*
1628 * Feature set definitions
1629 */
1630
1631 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1632 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1633 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1634 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1635 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1636 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1637 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1638 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1639 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1640 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1641 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1642 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1643 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1644 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1645 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1646 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1647 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1648 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1649 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1650
1651 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1652 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1653 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1654 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1655 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1656 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1657 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1658
1659 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1660 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1661 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1662 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1663 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1664 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1665 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1666 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1667 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1668 /*
1669 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1670 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1671 * all other data structures' checksums. However, the METADATA_CSUM and
1672 * GDT_CSUM bits are mutually exclusive.
1673 */
1674 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1675 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1676 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1677
1678 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1679 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1680 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1681 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1682 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1683 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1684 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1685 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1686 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1687 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1688 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1689 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1690 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1691 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1692 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1693
1694 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1695 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1696 { \
1697 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1698 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1699 } \
1700 static inline void ext4_set_feature_##name(struct super_block *sb) \
1701 { \
1702 EXT4_SB(sb)->s_es->s_feature_compat |= \
1703 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1704 } \
1705 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1706 { \
1707 EXT4_SB(sb)->s_es->s_feature_compat &= \
1708 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1709 }
1710
1711 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1712 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1713 { \
1714 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1715 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1716 } \
1717 static inline void ext4_set_feature_##name(struct super_block *sb) \
1718 { \
1719 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1720 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1721 } \
1722 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1723 { \
1724 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1725 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1726 }
1727
1728 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1729 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1730 { \
1731 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1732 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1733 } \
1734 static inline void ext4_set_feature_##name(struct super_block *sb) \
1735 { \
1736 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1737 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1738 } \
1739 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1740 { \
1741 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1742 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1743 }
1744
1745 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1746 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1747 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1748 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1749 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1750 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1751 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1752
1753 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1754 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1755 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1756 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1757 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1758 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1759 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1760 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1761 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1762 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1763 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1764 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1765
1766 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1769 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1770 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1771 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1772 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1773 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1774 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1775 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1776 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1777 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1778 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1779 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1780 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1781
1782 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1783 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1784 EXT4_FEATURE_INCOMPAT_META_BG)
1785 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1786 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1787 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1788
1789 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1790 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1791 EXT4_FEATURE_INCOMPAT_RECOVER| \
1792 EXT4_FEATURE_INCOMPAT_META_BG)
1793 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1794 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1795 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1796
1797 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1798 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1799 EXT4_FEATURE_INCOMPAT_RECOVER| \
1800 EXT4_FEATURE_INCOMPAT_META_BG| \
1801 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1802 EXT4_FEATURE_INCOMPAT_64BIT| \
1803 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1804 EXT4_FEATURE_INCOMPAT_MMP | \
1805 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1806 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1807 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1808 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1809 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1810 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1811 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1812 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1813 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1814 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1815 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1816 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1817 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1818 EXT4_FEATURE_RO_COMPAT_PROJECT)
1819
1820 #define EXTN_FEATURE_FUNCS(ver) \
1821 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1822 { \
1823 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1824 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1825 } \
1826 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1827 { \
1828 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1829 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1830 } \
1831 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1832 { \
1833 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1834 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1835 }
1836
1837 EXTN_FEATURE_FUNCS(2)
1838 EXTN_FEATURE_FUNCS(3)
1839 EXTN_FEATURE_FUNCS(4)
1840
1841 static inline bool ext4_has_compat_features(struct super_block *sb)
1842 {
1843 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1844 }
1845 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1846 {
1847 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1848 }
1849 static inline bool ext4_has_incompat_features(struct super_block *sb)
1850 {
1851 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1852 }
1853
1854 /*
1855 * Default values for user and/or group using reserved blocks
1856 */
1857 #define EXT4_DEF_RESUID 0
1858 #define EXT4_DEF_RESGID 0
1859
1860 /*
1861 * Default project ID
1862 */
1863 #define EXT4_DEF_PROJID 0
1864
1865 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1866
1867 /*
1868 * Default mount options
1869 */
1870 #define EXT4_DEFM_DEBUG 0x0001
1871 #define EXT4_DEFM_BSDGROUPS 0x0002
1872 #define EXT4_DEFM_XATTR_USER 0x0004
1873 #define EXT4_DEFM_ACL 0x0008
1874 #define EXT4_DEFM_UID16 0x0010
1875 #define EXT4_DEFM_JMODE 0x0060
1876 #define EXT4_DEFM_JMODE_DATA 0x0020
1877 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1878 #define EXT4_DEFM_JMODE_WBACK 0x0060
1879 #define EXT4_DEFM_NOBARRIER 0x0100
1880 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1881 #define EXT4_DEFM_DISCARD 0x0400
1882 #define EXT4_DEFM_NODELALLOC 0x0800
1883
1884 /*
1885 * Default journal batch times
1886 */
1887 #define EXT4_DEF_MIN_BATCH_TIME 0
1888 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1889
1890 /*
1891 * Minimum number of groups in a flexgroup before we separate out
1892 * directories into the first block group of a flexgroup
1893 */
1894 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1895
1896 /*
1897 * Structure of a directory entry
1898 */
1899 #define EXT4_NAME_LEN 255
1900
1901 struct ext4_dir_entry {
1902 __le32 inode; /* Inode number */
1903 __le16 rec_len; /* Directory entry length */
1904 __le16 name_len; /* Name length */
1905 char name[EXT4_NAME_LEN]; /* File name */
1906 };
1907
1908 /*
1909 * The new version of the directory entry. Since EXT4 structures are
1910 * stored in intel byte order, and the name_len field could never be
1911 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1912 * file_type field.
1913 */
1914 struct ext4_dir_entry_2 {
1915 __le32 inode; /* Inode number */
1916 __le16 rec_len; /* Directory entry length */
1917 __u8 name_len; /* Name length */
1918 __u8 file_type;
1919 char name[EXT4_NAME_LEN]; /* File name */
1920 };
1921
1922 /*
1923 * This is a bogus directory entry at the end of each leaf block that
1924 * records checksums.
1925 */
1926 struct ext4_dir_entry_tail {
1927 __le32 det_reserved_zero1; /* Pretend to be unused */
1928 __le16 det_rec_len; /* 12 */
1929 __u8 det_reserved_zero2; /* Zero name length */
1930 __u8 det_reserved_ft; /* 0xDE, fake file type */
1931 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1932 };
1933
1934 #define EXT4_DIRENT_TAIL(block, blocksize) \
1935 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1936 ((blocksize) - \
1937 sizeof(struct ext4_dir_entry_tail))))
1938
1939 /*
1940 * Ext4 directory file types. Only the low 3 bits are used. The
1941 * other bits are reserved for now.
1942 */
1943 #define EXT4_FT_UNKNOWN 0
1944 #define EXT4_FT_REG_FILE 1
1945 #define EXT4_FT_DIR 2
1946 #define EXT4_FT_CHRDEV 3
1947 #define EXT4_FT_BLKDEV 4
1948 #define EXT4_FT_FIFO 5
1949 #define EXT4_FT_SOCK 6
1950 #define EXT4_FT_SYMLINK 7
1951
1952 #define EXT4_FT_MAX 8
1953
1954 #define EXT4_FT_DIR_CSUM 0xDE
1955
1956 /*
1957 * EXT4_DIR_PAD defines the directory entries boundaries
1958 *
1959 * NOTE: It must be a multiple of 4
1960 */
1961 #define EXT4_DIR_PAD 4
1962 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1963 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1964 ~EXT4_DIR_ROUND)
1965 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1966
1967 /*
1968 * If we ever get support for fs block sizes > page_size, we'll need
1969 * to remove the #if statements in the next two functions...
1970 */
1971 static inline unsigned int
1972 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1973 {
1974 unsigned len = le16_to_cpu(dlen);
1975
1976 #if (PAGE_CACHE_SIZE >= 65536)
1977 if (len == EXT4_MAX_REC_LEN || len == 0)
1978 return blocksize;
1979 return (len & 65532) | ((len & 3) << 16);
1980 #else
1981 return len;
1982 #endif
1983 }
1984
1985 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1986 {
1987 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1988 BUG();
1989 #if (PAGE_CACHE_SIZE >= 65536)
1990 if (len < 65536)
1991 return cpu_to_le16(len);
1992 if (len == blocksize) {
1993 if (blocksize == 65536)
1994 return cpu_to_le16(EXT4_MAX_REC_LEN);
1995 else
1996 return cpu_to_le16(0);
1997 }
1998 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1999 #else
2000 return cpu_to_le16(len);
2001 #endif
2002 }
2003
2004 /*
2005 * Hash Tree Directory indexing
2006 * (c) Daniel Phillips, 2001
2007 */
2008
2009 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2010 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2011 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2012 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2013
2014 /* Legal values for the dx_root hash_version field: */
2015
2016 #define DX_HASH_LEGACY 0
2017 #define DX_HASH_HALF_MD4 1
2018 #define DX_HASH_TEA 2
2019 #define DX_HASH_LEGACY_UNSIGNED 3
2020 #define DX_HASH_HALF_MD4_UNSIGNED 4
2021 #define DX_HASH_TEA_UNSIGNED 5
2022
2023 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2024 const void *address, unsigned int length)
2025 {
2026 struct {
2027 struct shash_desc shash;
2028 char ctx[4];
2029 } desc;
2030 int err;
2031
2032 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2033
2034 desc.shash.tfm = sbi->s_chksum_driver;
2035 desc.shash.flags = 0;
2036 *(u32 *)desc.ctx = crc;
2037
2038 err = crypto_shash_update(&desc.shash, address, length);
2039 BUG_ON(err);
2040
2041 return *(u32 *)desc.ctx;
2042 }
2043
2044 #ifdef __KERNEL__
2045
2046 /* hash info structure used by the directory hash */
2047 struct dx_hash_info
2048 {
2049 u32 hash;
2050 u32 minor_hash;
2051 int hash_version;
2052 u32 *seed;
2053 };
2054
2055
2056 /* 32 and 64 bit signed EOF for dx directories */
2057 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2058 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2059
2060
2061 /*
2062 * Control parameters used by ext4_htree_next_block
2063 */
2064 #define HASH_NB_ALWAYS 1
2065
2066 struct ext4_filename {
2067 const struct qstr *usr_fname;
2068 struct ext4_str disk_name;
2069 struct dx_hash_info hinfo;
2070 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2071 struct ext4_str crypto_buf;
2072 #endif
2073 };
2074
2075 #define fname_name(p) ((p)->disk_name.name)
2076 #define fname_len(p) ((p)->disk_name.len)
2077
2078 /*
2079 * Describe an inode's exact location on disk and in memory
2080 */
2081 struct ext4_iloc
2082 {
2083 struct buffer_head *bh;
2084 unsigned long offset;
2085 ext4_group_t block_group;
2086 };
2087
2088 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2089 {
2090 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2091 }
2092
2093 /*
2094 * This structure is stuffed into the struct file's private_data field
2095 * for directories. It is where we put information so that we can do
2096 * readdir operations in hash tree order.
2097 */
2098 struct dir_private_info {
2099 struct rb_root root;
2100 struct rb_node *curr_node;
2101 struct fname *extra_fname;
2102 loff_t last_pos;
2103 __u32 curr_hash;
2104 __u32 curr_minor_hash;
2105 __u32 next_hash;
2106 };
2107
2108 /* calculate the first block number of the group */
2109 static inline ext4_fsblk_t
2110 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2111 {
2112 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2113 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2114 }
2115
2116 /*
2117 * Special error return code only used by dx_probe() and its callers.
2118 */
2119 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2120
2121 /*
2122 * Timeout and state flag for lazy initialization inode thread.
2123 */
2124 #define EXT4_DEF_LI_WAIT_MULT 10
2125 #define EXT4_DEF_LI_MAX_START_DELAY 5
2126 #define EXT4_LAZYINIT_QUIT 0x0001
2127 #define EXT4_LAZYINIT_RUNNING 0x0002
2128
2129 /*
2130 * Lazy inode table initialization info
2131 */
2132 struct ext4_lazy_init {
2133 unsigned long li_state;
2134 struct list_head li_request_list;
2135 struct mutex li_list_mtx;
2136 };
2137
2138 struct ext4_li_request {
2139 struct super_block *lr_super;
2140 struct ext4_sb_info *lr_sbi;
2141 ext4_group_t lr_next_group;
2142 struct list_head lr_request;
2143 unsigned long lr_next_sched;
2144 unsigned long lr_timeout;
2145 };
2146
2147 struct ext4_features {
2148 struct kobject f_kobj;
2149 struct completion f_kobj_unregister;
2150 };
2151
2152 /*
2153 * This structure will be used for multiple mount protection. It will be
2154 * written into the block number saved in the s_mmp_block field in the
2155 * superblock. Programs that check MMP should assume that if
2156 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2157 * to use the filesystem, regardless of how old the timestamp is.
2158 */
2159 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2160 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2161 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2162 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2163
2164 struct mmp_struct {
2165 __le32 mmp_magic; /* Magic number for MMP */
2166 __le32 mmp_seq; /* Sequence no. updated periodically */
2167
2168 /*
2169 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2170 * purposes and do not affect the correctness of the algorithm
2171 */
2172 __le64 mmp_time; /* Time last updated */
2173 char mmp_nodename[64]; /* Node which last updated MMP block */
2174 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2175
2176 /*
2177 * mmp_check_interval is used to verify if the MMP block has been
2178 * updated on the block device. The value is updated based on the
2179 * maximum time to write the MMP block during an update cycle.
2180 */
2181 __le16 mmp_check_interval;
2182
2183 __le16 mmp_pad1;
2184 __le32 mmp_pad2[226];
2185 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2186 };
2187
2188 /* arguments passed to the mmp thread */
2189 struct mmpd_data {
2190 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2191 struct super_block *sb; /* super block of the fs */
2192 };
2193
2194 /*
2195 * Check interval multiplier
2196 * The MMP block is written every update interval and initially checked every
2197 * update interval x the multiplier (the value is then adapted based on the
2198 * write latency). The reason is that writes can be delayed under load and we
2199 * don't want readers to incorrectly assume that the filesystem is no longer
2200 * in use.
2201 */
2202 #define EXT4_MMP_CHECK_MULT 2UL
2203
2204 /*
2205 * Minimum interval for MMP checking in seconds.
2206 */
2207 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2208
2209 /*
2210 * Maximum interval for MMP checking in seconds.
2211 */
2212 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2213
2214 /*
2215 * Function prototypes
2216 */
2217
2218 /*
2219 * Ok, these declarations are also in <linux/kernel.h> but none of the
2220 * ext4 source programs needs to include it so they are duplicated here.
2221 */
2222 # define NORET_TYPE /**/
2223 # define ATTRIB_NORET __attribute__((noreturn))
2224 # define NORET_AND noreturn,
2225
2226 /* bitmap.c */
2227 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2228 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2229 struct ext4_group_desc *gdp,
2230 struct buffer_head *bh, int sz);
2231 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2232 struct ext4_group_desc *gdp,
2233 struct buffer_head *bh, int sz);
2234 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2235 struct ext4_group_desc *gdp,
2236 struct buffer_head *bh);
2237 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2238 struct ext4_group_desc *gdp,
2239 struct buffer_head *bh);
2240
2241 /* balloc.c */
2242 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2243 ext4_fsblk_t blocknr,
2244 ext4_group_t *blockgrpp,
2245 ext4_grpblk_t *offsetp);
2246 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2247 ext4_fsblk_t block);
2248
2249 extern unsigned int ext4_block_group(struct super_block *sb,
2250 ext4_fsblk_t blocknr);
2251 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2252 ext4_fsblk_t blocknr);
2253 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2254 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2255 ext4_group_t group);
2256 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2257 ext4_fsblk_t goal,
2258 unsigned int flags,
2259 unsigned long *count,
2260 int *errp);
2261 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2262 s64 nclusters, unsigned int flags);
2263 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2264 extern void ext4_check_blocks_bitmap(struct super_block *);
2265 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2266 ext4_group_t block_group,
2267 struct buffer_head ** bh);
2268 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2269
2270 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2271 ext4_group_t block_group);
2272 extern int ext4_wait_block_bitmap(struct super_block *sb,
2273 ext4_group_t block_group,
2274 struct buffer_head *bh);
2275 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2276 ext4_group_t block_group);
2277 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2278 ext4_group_t block_group,
2279 struct ext4_group_desc *gdp);
2280 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2281
2282 /* crypto_policy.c */
2283 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2284 struct inode *child);
2285 int ext4_inherit_context(struct inode *parent, struct inode *child);
2286 void ext4_to_hex(char *dst, char *src, size_t src_size);
2287 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2288 struct inode *inode);
2289 int ext4_get_policy(struct inode *inode,
2290 struct ext4_encryption_policy *policy);
2291
2292 /* crypto.c */
2293 extern struct kmem_cache *ext4_crypt_info_cachep;
2294 bool ext4_valid_contents_enc_mode(uint32_t mode);
2295 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2296 extern struct workqueue_struct *ext4_read_workqueue;
2297 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2298 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2299 void ext4_restore_control_page(struct page *data_page);
2300 struct page *ext4_encrypt(struct inode *inode,
2301 struct page *plaintext_page);
2302 int ext4_decrypt(struct page *page);
2303 int ext4_encrypted_zeroout(struct inode *inode, ext4_lblk_t lblk,
2304 ext4_fsblk_t pblk, ext4_lblk_t len);
2305 extern const struct dentry_operations ext4_encrypted_d_ops;
2306
2307 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2308 int ext4_init_crypto(void);
2309 void ext4_exit_crypto(void);
2310 static inline int ext4_sb_has_crypto(struct super_block *sb)
2311 {
2312 return ext4_has_feature_encrypt(sb);
2313 }
2314 #else
2315 static inline int ext4_init_crypto(void) { return 0; }
2316 static inline void ext4_exit_crypto(void) { }
2317 static inline int ext4_sb_has_crypto(struct super_block *sb)
2318 {
2319 return 0;
2320 }
2321 #endif
2322
2323 /* crypto_fname.c */
2324 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2325 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2326 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2327 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2328 u32 ilen, struct ext4_str *crypto_str);
2329 int _ext4_fname_disk_to_usr(struct inode *inode,
2330 struct dx_hash_info *hinfo,
2331 const struct ext4_str *iname,
2332 struct ext4_str *oname);
2333 int ext4_fname_disk_to_usr(struct inode *inode,
2334 struct dx_hash_info *hinfo,
2335 const struct ext4_dir_entry_2 *de,
2336 struct ext4_str *oname);
2337 int ext4_fname_usr_to_disk(struct inode *inode,
2338 const struct qstr *iname,
2339 struct ext4_str *oname);
2340 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2341 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2342 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2343 int lookup, struct ext4_filename *fname);
2344 void ext4_fname_free_filename(struct ext4_filename *fname);
2345 #else
2346 static inline
2347 int ext4_setup_fname_crypto(struct inode *inode)
2348 {
2349 return 0;
2350 }
2351 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2352 static inline int ext4_fname_setup_filename(struct inode *dir,
2353 const struct qstr *iname,
2354 int lookup, struct ext4_filename *fname)
2355 {
2356 fname->usr_fname = iname;
2357 fname->disk_name.name = (unsigned char *) iname->name;
2358 fname->disk_name.len = iname->len;
2359 return 0;
2360 }
2361 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2362 #endif
2363
2364
2365 /* crypto_key.c */
2366 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2367 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2368 int _ext4_get_encryption_info(struct inode *inode);
2369
2370 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2371 int ext4_has_encryption_key(struct inode *inode);
2372
2373 static inline int ext4_get_encryption_info(struct inode *inode)
2374 {
2375 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2376
2377 if (!ci ||
2378 (ci->ci_keyring_key &&
2379 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2380 (1 << KEY_FLAG_REVOKED) |
2381 (1 << KEY_FLAG_DEAD)))))
2382 return _ext4_get_encryption_info(inode);
2383 return 0;
2384 }
2385
2386 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2387 {
2388 return EXT4_I(inode)->i_crypt_info;
2389 }
2390
2391 #else
2392 static inline int ext4_has_encryption_key(struct inode *inode)
2393 {
2394 return 0;
2395 }
2396 static inline int ext4_get_encryption_info(struct inode *inode)
2397 {
2398 return 0;
2399 }
2400 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2401 {
2402 return NULL;
2403 }
2404 #endif
2405
2406
2407 /* dir.c */
2408 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2409 struct file *,
2410 struct ext4_dir_entry_2 *,
2411 struct buffer_head *, char *, int,
2412 unsigned int);
2413 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2414 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2415 (de), (bh), (buf), (size), (offset)))
2416 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2417 __u32 minor_hash,
2418 struct ext4_dir_entry_2 *dirent,
2419 struct ext4_str *ent_name);
2420 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2421 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2422 struct buffer_head *bh,
2423 void *buf, int buf_size,
2424 struct ext4_filename *fname,
2425 struct ext4_dir_entry_2 **dest_de);
2426 int ext4_insert_dentry(struct inode *dir,
2427 struct inode *inode,
2428 struct ext4_dir_entry_2 *de,
2429 int buf_size,
2430 struct ext4_filename *fname);
2431 static inline void ext4_update_dx_flag(struct inode *inode)
2432 {
2433 if (!ext4_has_feature_dir_index(inode->i_sb))
2434 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2435 }
2436 static unsigned char ext4_filetype_table[] = {
2437 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2438 };
2439
2440 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2441 {
2442 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2443 return DT_UNKNOWN;
2444
2445 return ext4_filetype_table[filetype];
2446 }
2447 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2448 void *buf, int buf_size);
2449
2450 /* fsync.c */
2451 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2452
2453 /* hash.c */
2454 extern int ext4fs_dirhash(const char *name, int len, struct
2455 dx_hash_info *hinfo);
2456
2457 /* ialloc.c */
2458 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2459 const struct qstr *qstr, __u32 goal,
2460 uid_t *owner, int handle_type,
2461 unsigned int line_no, int nblocks);
2462
2463 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2464 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2465 0, 0, 0)
2466 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2467 type, nblocks) \
2468 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2469 (type), __LINE__, (nblocks))
2470
2471
2472 extern void ext4_free_inode(handle_t *, struct inode *);
2473 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2474 extern unsigned long ext4_count_free_inodes(struct super_block *);
2475 extern unsigned long ext4_count_dirs(struct super_block *);
2476 extern void ext4_check_inodes_bitmap(struct super_block *);
2477 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2478 extern int ext4_init_inode_table(struct super_block *sb,
2479 ext4_group_t group, int barrier);
2480 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2481
2482 /* mballoc.c */
2483 extern const struct file_operations ext4_seq_mb_groups_fops;
2484 extern long ext4_mb_stats;
2485 extern long ext4_mb_max_to_scan;
2486 extern int ext4_mb_init(struct super_block *);
2487 extern int ext4_mb_release(struct super_block *);
2488 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2489 struct ext4_allocation_request *, int *);
2490 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2491 extern void ext4_discard_preallocations(struct inode *);
2492 extern int __init ext4_init_mballoc(void);
2493 extern void ext4_exit_mballoc(void);
2494 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2495 struct buffer_head *bh, ext4_fsblk_t block,
2496 unsigned long count, int flags);
2497 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2498 ext4_group_t ngroups);
2499 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2500 ext4_group_t i, struct ext4_group_desc *desc);
2501 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2502 ext4_fsblk_t block, unsigned long count);
2503 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2504
2505 /* inode.c */
2506 int ext4_inode_is_fast_symlink(struct inode *inode);
2507 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2508 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2509 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2510 struct buffer_head *bh_result, int create);
2511 int ext4_dax_mmap_get_block(struct inode *inode, sector_t iblock,
2512 struct buffer_head *bh_result, int create);
2513 int ext4_get_block(struct inode *inode, sector_t iblock,
2514 struct buffer_head *bh_result, int create);
2515 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2516 struct buffer_head *bh, int create);
2517 int ext4_walk_page_buffers(handle_t *handle,
2518 struct buffer_head *head,
2519 unsigned from,
2520 unsigned to,
2521 int *partial,
2522 int (*fn)(handle_t *handle,
2523 struct buffer_head *bh));
2524 int do_journal_get_write_access(handle_t *handle,
2525 struct buffer_head *bh);
2526 #define FALL_BACK_TO_NONDELALLOC 1
2527 #define CONVERT_INLINE_DATA 2
2528
2529 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2530 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2531 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2532 extern int ext4_setattr(struct dentry *, struct iattr *);
2533 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2534 struct kstat *stat);
2535 extern void ext4_evict_inode(struct inode *);
2536 extern void ext4_clear_inode(struct inode *);
2537 extern int ext4_sync_inode(handle_t *, struct inode *);
2538 extern void ext4_dirty_inode(struct inode *, int);
2539 extern int ext4_change_inode_journal_flag(struct inode *, int);
2540 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2541 extern int ext4_inode_attach_jinode(struct inode *inode);
2542 extern int ext4_can_truncate(struct inode *inode);
2543 extern void ext4_truncate(struct inode *);
2544 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2545 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2546 extern void ext4_set_inode_flags(struct inode *);
2547 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2548 extern int ext4_alloc_da_blocks(struct inode *inode);
2549 extern void ext4_set_aops(struct inode *inode);
2550 extern int ext4_writepage_trans_blocks(struct inode *);
2551 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2552 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2553 loff_t lstart, loff_t lend);
2554 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2555 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2556 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2557 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2558 extern void ext4_da_update_reserve_space(struct inode *inode,
2559 int used, int quota_claim);
2560 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2561 ext4_fsblk_t pblk, ext4_lblk_t len);
2562
2563 /* indirect.c */
2564 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2565 struct ext4_map_blocks *map, int flags);
2566 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2567 loff_t offset);
2568 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2569 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2570 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2571 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2572 ext4_lblk_t start, ext4_lblk_t end);
2573
2574 /* ioctl.c */
2575 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2576 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2577
2578 /* migrate.c */
2579 extern int ext4_ext_migrate(struct inode *);
2580 extern int ext4_ind_migrate(struct inode *inode);
2581
2582 /* namei.c */
2583 extern int ext4_dirent_csum_verify(struct inode *inode,
2584 struct ext4_dir_entry *dirent);
2585 extern int ext4_orphan_add(handle_t *, struct inode *);
2586 extern int ext4_orphan_del(handle_t *, struct inode *);
2587 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2588 __u32 start_minor_hash, __u32 *next_hash);
2589 extern int ext4_search_dir(struct buffer_head *bh,
2590 char *search_buf,
2591 int buf_size,
2592 struct inode *dir,
2593 struct ext4_filename *fname,
2594 const struct qstr *d_name,
2595 unsigned int offset,
2596 struct ext4_dir_entry_2 **res_dir);
2597 extern int ext4_generic_delete_entry(handle_t *handle,
2598 struct inode *dir,
2599 struct ext4_dir_entry_2 *de_del,
2600 struct buffer_head *bh,
2601 void *entry_buf,
2602 int buf_size,
2603 int csum_size);
2604 extern int ext4_empty_dir(struct inode *inode);
2605
2606 /* resize.c */
2607 extern int ext4_group_add(struct super_block *sb,
2608 struct ext4_new_group_data *input);
2609 extern int ext4_group_extend(struct super_block *sb,
2610 struct ext4_super_block *es,
2611 ext4_fsblk_t n_blocks_count);
2612 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2613
2614 /* super.c */
2615 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2616 extern int ext4_calculate_overhead(struct super_block *sb);
2617 extern void ext4_superblock_csum_set(struct super_block *sb);
2618 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2619 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2620 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2621 ext4_group_t ngroup);
2622 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2623 char nbuf[16]);
2624
2625 extern __printf(4, 5)
2626 void __ext4_error(struct super_block *, const char *, unsigned int,
2627 const char *, ...);
2628 extern __printf(5, 6)
2629 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2630 const char *, ...);
2631 extern __printf(5, 6)
2632 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2633 const char *, ...);
2634 extern void __ext4_std_error(struct super_block *, const char *,
2635 unsigned int, int);
2636 extern __printf(4, 5)
2637 void __ext4_abort(struct super_block *, const char *, unsigned int,
2638 const char *, ...);
2639 extern __printf(4, 5)
2640 void __ext4_warning(struct super_block *, const char *, unsigned int,
2641 const char *, ...);
2642 extern __printf(4, 5)
2643 void __ext4_warning_inode(const struct inode *inode, const char *function,
2644 unsigned int line, const char *fmt, ...);
2645 extern __printf(3, 4)
2646 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2647 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2648 const char *, unsigned int, const char *);
2649 extern __printf(7, 8)
2650 void __ext4_grp_locked_error(const char *, unsigned int,
2651 struct super_block *, ext4_group_t,
2652 unsigned long, ext4_fsblk_t,
2653 const char *, ...);
2654
2655 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2656 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2657
2658 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2659 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2660
2661 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2662 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2663
2664 #ifdef CONFIG_PRINTK
2665
2666 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2667 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2668 #define ext4_error_file(file, func, line, block, fmt, ...) \
2669 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2670 #define ext4_error(sb, fmt, ...) \
2671 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2672 #define ext4_abort(sb, fmt, ...) \
2673 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2674 #define ext4_warning(sb, fmt, ...) \
2675 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2676 #define ext4_warning_inode(inode, fmt, ...) \
2677 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2678 #define ext4_msg(sb, level, fmt, ...) \
2679 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2680 #define dump_mmp_msg(sb, mmp, msg) \
2681 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2682 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2683 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2684 fmt, ##__VA_ARGS__)
2685
2686 #else
2687
2688 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2689 do { \
2690 no_printk(fmt, ##__VA_ARGS__); \
2691 __ext4_error_inode(inode, "", 0, block, " "); \
2692 } while (0)
2693 #define ext4_error_file(file, func, line, block, fmt, ...) \
2694 do { \
2695 no_printk(fmt, ##__VA_ARGS__); \
2696 __ext4_error_file(file, "", 0, block, " "); \
2697 } while (0)
2698 #define ext4_error(sb, fmt, ...) \
2699 do { \
2700 no_printk(fmt, ##__VA_ARGS__); \
2701 __ext4_error(sb, "", 0, " "); \
2702 } while (0)
2703 #define ext4_abort(sb, fmt, ...) \
2704 do { \
2705 no_printk(fmt, ##__VA_ARGS__); \
2706 __ext4_abort(sb, "", 0, " "); \
2707 } while (0)
2708 #define ext4_warning(sb, fmt, ...) \
2709 do { \
2710 no_printk(fmt, ##__VA_ARGS__); \
2711 __ext4_warning(sb, "", 0, " "); \
2712 } while (0)
2713 #define ext4_warning_inode(inode, fmt, ...) \
2714 do { \
2715 no_printk(fmt, ##__VA_ARGS__); \
2716 __ext4_warning_inode(inode, "", 0, " "); \
2717 } while (0)
2718 #define ext4_msg(sb, level, fmt, ...) \
2719 do { \
2720 no_printk(fmt, ##__VA_ARGS__); \
2721 __ext4_msg(sb, "", " "); \
2722 } while (0)
2723 #define dump_mmp_msg(sb, mmp, msg) \
2724 __dump_mmp_msg(sb, mmp, "", 0, "")
2725 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2726 do { \
2727 no_printk(fmt, ##__VA_ARGS__); \
2728 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2729 } while (0)
2730
2731 #endif
2732
2733 extern void ext4_update_dynamic_rev(struct super_block *sb);
2734 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2735 __u32 compat);
2736 extern int ext4_update_rocompat_feature(handle_t *handle,
2737 struct super_block *sb, __u32 rocompat);
2738 extern int ext4_update_incompat_feature(handle_t *handle,
2739 struct super_block *sb, __u32 incompat);
2740 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2741 struct ext4_group_desc *bg);
2742 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2743 struct ext4_group_desc *bg);
2744 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2745 struct ext4_group_desc *bg);
2746 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2747 struct ext4_group_desc *bg);
2748 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2749 struct ext4_group_desc *bg);
2750 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2751 struct ext4_group_desc *bg);
2752 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2753 struct ext4_group_desc *bg);
2754 extern void ext4_block_bitmap_set(struct super_block *sb,
2755 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2756 extern void ext4_inode_bitmap_set(struct super_block *sb,
2757 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2758 extern void ext4_inode_table_set(struct super_block *sb,
2759 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2760 extern void ext4_free_group_clusters_set(struct super_block *sb,
2761 struct ext4_group_desc *bg,
2762 __u32 count);
2763 extern void ext4_free_inodes_set(struct super_block *sb,
2764 struct ext4_group_desc *bg, __u32 count);
2765 extern void ext4_used_dirs_set(struct super_block *sb,
2766 struct ext4_group_desc *bg, __u32 count);
2767 extern void ext4_itable_unused_set(struct super_block *sb,
2768 struct ext4_group_desc *bg, __u32 count);
2769 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2770 struct ext4_group_desc *gdp);
2771 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2772 struct ext4_group_desc *gdp);
2773 extern int ext4_register_li_request(struct super_block *sb,
2774 ext4_group_t first_not_zeroed);
2775
2776 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2777 {
2778 return ext4_has_feature_gdt_csum(sb) ||
2779 EXT4_SB(sb)->s_chksum_driver != NULL;
2780 }
2781
2782 static inline int ext4_has_metadata_csum(struct super_block *sb)
2783 {
2784 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2785 !EXT4_SB(sb)->s_chksum_driver);
2786
2787 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2788 }
2789 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2790 {
2791 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2792 le32_to_cpu(es->s_blocks_count_lo);
2793 }
2794
2795 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2796 {
2797 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2798 le32_to_cpu(es->s_r_blocks_count_lo);
2799 }
2800
2801 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2802 {
2803 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2804 le32_to_cpu(es->s_free_blocks_count_lo);
2805 }
2806
2807 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2808 ext4_fsblk_t blk)
2809 {
2810 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2811 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2812 }
2813
2814 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2815 ext4_fsblk_t blk)
2816 {
2817 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2818 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2819 }
2820
2821 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2822 ext4_fsblk_t blk)
2823 {
2824 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2825 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2826 }
2827
2828 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2829 {
2830 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2831 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2832 le32_to_cpu(raw_inode->i_size_lo);
2833 else
2834 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2835 }
2836
2837 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2838 {
2839 raw_inode->i_size_lo = cpu_to_le32(i_size);
2840 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2841 }
2842
2843 static inline
2844 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2845 ext4_group_t group)
2846 {
2847 struct ext4_group_info ***grp_info;
2848 long indexv, indexh;
2849 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2850 grp_info = EXT4_SB(sb)->s_group_info;
2851 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2852 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2853 return grp_info[indexv][indexh];
2854 }
2855
2856 /*
2857 * Reading s_groups_count requires using smp_rmb() afterwards. See
2858 * the locking protocol documented in the comments of ext4_group_add()
2859 * in resize.c
2860 */
2861 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2862 {
2863 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2864
2865 smp_rmb();
2866 return ngroups;
2867 }
2868
2869 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2870 ext4_group_t block_group)
2871 {
2872 return block_group >> sbi->s_log_groups_per_flex;
2873 }
2874
2875 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2876 {
2877 return 1 << sbi->s_log_groups_per_flex;
2878 }
2879
2880 #define ext4_std_error(sb, errno) \
2881 do { \
2882 if ((errno)) \
2883 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2884 } while (0)
2885
2886 #ifdef CONFIG_SMP
2887 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2888 * counters. So we need to make sure we have free clusters more
2889 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2890 */
2891 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2892 #else
2893 #define EXT4_FREECLUSTERS_WATERMARK 0
2894 #endif
2895
2896 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2897 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2898 {
2899 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2900 !inode_is_locked(inode));
2901 down_write(&EXT4_I(inode)->i_data_sem);
2902 if (newsize > EXT4_I(inode)->i_disksize)
2903 EXT4_I(inode)->i_disksize = newsize;
2904 up_write(&EXT4_I(inode)->i_data_sem);
2905 }
2906
2907 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2908 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2909 {
2910 int changed = 0;
2911
2912 if (newsize > inode->i_size) {
2913 i_size_write(inode, newsize);
2914 changed = 1;
2915 }
2916 if (newsize > EXT4_I(inode)->i_disksize) {
2917 ext4_update_i_disksize(inode, newsize);
2918 changed |= 2;
2919 }
2920 return changed;
2921 }
2922
2923 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2924 loff_t len);
2925
2926 struct ext4_group_info {
2927 unsigned long bb_state;
2928 struct rb_root bb_free_root;
2929 ext4_grpblk_t bb_first_free; /* first free block */
2930 ext4_grpblk_t bb_free; /* total free blocks */
2931 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2932 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2933 struct list_head bb_prealloc_list;
2934 #ifdef DOUBLE_CHECK
2935 void *bb_bitmap;
2936 #endif
2937 struct rw_semaphore alloc_sem;
2938 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2939 * regions, index is order.
2940 * bb_counters[3] = 5 means
2941 * 5 free 8-block regions. */
2942 };
2943
2944 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2945 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2946 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2947 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2948
2949 #define EXT4_MB_GRP_NEED_INIT(grp) \
2950 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2951 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2952 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2953 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2954 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2955
2956 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2957 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2958 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2959 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2960 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2961 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2962
2963 #define EXT4_MAX_CONTENTION 8
2964 #define EXT4_CONTENTION_THRESHOLD 2
2965
2966 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2967 ext4_group_t group)
2968 {
2969 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2970 }
2971
2972 /*
2973 * Returns true if the filesystem is busy enough that attempts to
2974 * access the block group locks has run into contention.
2975 */
2976 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2977 {
2978 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2979 }
2980
2981 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2982 {
2983 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2984 if (spin_trylock(lock))
2985 /*
2986 * We're able to grab the lock right away, so drop the
2987 * lock contention counter.
2988 */
2989 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2990 else {
2991 /*
2992 * The lock is busy, so bump the contention counter,
2993 * and then wait on the spin lock.
2994 */
2995 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2996 EXT4_MAX_CONTENTION);
2997 spin_lock(lock);
2998 }
2999 }
3000
3001 static inline void ext4_unlock_group(struct super_block *sb,
3002 ext4_group_t group)
3003 {
3004 spin_unlock(ext4_group_lock_ptr(sb, group));
3005 }
3006
3007 /*
3008 * Block validity checking
3009 */
3010 #define ext4_check_indirect_blockref(inode, bh) \
3011 ext4_check_blockref(__func__, __LINE__, inode, \
3012 (__le32 *)(bh)->b_data, \
3013 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3014
3015 #define ext4_ind_check_inode(inode) \
3016 ext4_check_blockref(__func__, __LINE__, inode, \
3017 EXT4_I(inode)->i_data, \
3018 EXT4_NDIR_BLOCKS)
3019
3020 /*
3021 * Inodes and files operations
3022 */
3023
3024 /* dir.c */
3025 extern const struct file_operations ext4_dir_operations;
3026
3027 /* file.c */
3028 extern const struct inode_operations ext4_file_inode_operations;
3029 extern const struct file_operations ext4_file_operations;
3030 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3031
3032 /* inline.c */
3033 extern int ext4_get_max_inline_size(struct inode *inode);
3034 extern int ext4_find_inline_data_nolock(struct inode *inode);
3035 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3036 unsigned int len);
3037 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3038
3039 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3040 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3041 struct inode *inode,
3042 loff_t pos, unsigned len,
3043 unsigned flags,
3044 struct page **pagep);
3045 extern int ext4_write_inline_data_end(struct inode *inode,
3046 loff_t pos, unsigned len,
3047 unsigned copied,
3048 struct page *page);
3049 extern struct buffer_head *
3050 ext4_journalled_write_inline_data(struct inode *inode,
3051 unsigned len,
3052 struct page *page);
3053 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3054 struct inode *inode,
3055 loff_t pos, unsigned len,
3056 unsigned flags,
3057 struct page **pagep,
3058 void **fsdata);
3059 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3060 unsigned len, unsigned copied,
3061 struct page *page);
3062 extern int ext4_try_add_inline_entry(handle_t *handle,
3063 struct ext4_filename *fname,
3064 struct inode *dir, struct inode *inode);
3065 extern int ext4_try_create_inline_dir(handle_t *handle,
3066 struct inode *parent,
3067 struct inode *inode);
3068 extern int ext4_read_inline_dir(struct file *filp,
3069 struct dir_context *ctx,
3070 int *has_inline_data);
3071 extern int htree_inlinedir_to_tree(struct file *dir_file,
3072 struct inode *dir, ext4_lblk_t block,
3073 struct dx_hash_info *hinfo,
3074 __u32 start_hash, __u32 start_minor_hash,
3075 int *has_inline_data);
3076 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3077 struct ext4_filename *fname,
3078 const struct qstr *d_name,
3079 struct ext4_dir_entry_2 **res_dir,
3080 int *has_inline_data);
3081 extern int ext4_delete_inline_entry(handle_t *handle,
3082 struct inode *dir,
3083 struct ext4_dir_entry_2 *de_del,
3084 struct buffer_head *bh,
3085 int *has_inline_data);
3086 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3087 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3088 struct ext4_dir_entry_2 **parent_de,
3089 int *retval);
3090 extern int ext4_inline_data_fiemap(struct inode *inode,
3091 struct fiemap_extent_info *fieinfo,
3092 int *has_inline, __u64 start, __u64 len);
3093 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3094 struct inode *inode,
3095 int needed);
3096 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3097
3098 extern int ext4_convert_inline_data(struct inode *inode);
3099
3100 static inline int ext4_has_inline_data(struct inode *inode)
3101 {
3102 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3103 EXT4_I(inode)->i_inline_off;
3104 }
3105
3106 /* namei.c */
3107 extern const struct inode_operations ext4_dir_inode_operations;
3108 extern const struct inode_operations ext4_special_inode_operations;
3109 extern struct dentry *ext4_get_parent(struct dentry *child);
3110 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3111 struct ext4_dir_entry_2 *de,
3112 int blocksize, int csum_size,
3113 unsigned int parent_ino, int dotdot_real_len);
3114 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3115 unsigned int blocksize);
3116 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3117 struct inode *inode,
3118 struct buffer_head *bh);
3119 #define S_SHIFT 12
3120 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3121 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3122 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3123 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3124 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3125 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3126 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3127 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3128 };
3129
3130 static inline void ext4_set_de_type(struct super_block *sb,
3131 struct ext4_dir_entry_2 *de,
3132 umode_t mode) {
3133 if (ext4_has_feature_filetype(sb))
3134 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3135 }
3136
3137 /* readpages.c */
3138 extern int ext4_mpage_readpages(struct address_space *mapping,
3139 struct list_head *pages, struct page *page,
3140 unsigned nr_pages);
3141
3142 /* symlink.c */
3143 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3144 extern const struct inode_operations ext4_symlink_inode_operations;
3145 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3146
3147 /* sysfs.c */
3148 extern int ext4_register_sysfs(struct super_block *sb);
3149 extern void ext4_unregister_sysfs(struct super_block *sb);
3150 extern int __init ext4_init_sysfs(void);
3151 extern void ext4_exit_sysfs(void);
3152
3153 /* block_validity */
3154 extern void ext4_release_system_zone(struct super_block *sb);
3155 extern int ext4_setup_system_zone(struct super_block *sb);
3156 extern int __init ext4_init_system_zone(void);
3157 extern void ext4_exit_system_zone(void);
3158 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3159 ext4_fsblk_t start_blk,
3160 unsigned int count);
3161 extern int ext4_check_blockref(const char *, unsigned int,
3162 struct inode *, __le32 *, unsigned int);
3163
3164 /* extents.c */
3165 struct ext4_ext_path;
3166 struct ext4_extent;
3167
3168 /*
3169 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3170 * __le32.
3171 */
3172 #define EXT_MAX_BLOCKS 0xffffffff
3173
3174 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3175 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3176 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3177 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3178 struct ext4_map_blocks *map, int flags);
3179 extern void ext4_ext_truncate(handle_t *, struct inode *);
3180 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3181 ext4_lblk_t end);
3182 extern void ext4_ext_init(struct super_block *);
3183 extern void ext4_ext_release(struct super_block *);
3184 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3185 loff_t len);
3186 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3187 loff_t offset, ssize_t len);
3188 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3189 struct ext4_map_blocks *map, int flags);
3190 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3191 ext4_lblk_t lblocks);
3192 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3193 int num,
3194 struct ext4_ext_path *path);
3195 extern int ext4_can_extents_be_merged(struct inode *inode,
3196 struct ext4_extent *ex1,
3197 struct ext4_extent *ex2);
3198 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3199 struct ext4_ext_path **,
3200 struct ext4_extent *, int);
3201 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3202 struct ext4_ext_path **,
3203 int flags);
3204 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3205 extern int ext4_ext_check_inode(struct inode *inode);
3206 extern int ext4_find_delalloc_range(struct inode *inode,
3207 ext4_lblk_t lblk_start,
3208 ext4_lblk_t lblk_end);
3209 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3210 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3211 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3212 __u64 start, __u64 len);
3213 extern int ext4_ext_precache(struct inode *inode);
3214 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3215 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3216 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3217 struct inode *inode2, ext4_lblk_t lblk1,
3218 ext4_lblk_t lblk2, ext4_lblk_t count,
3219 int mark_unwritten,int *err);
3220
3221 /* move_extent.c */
3222 extern void ext4_double_down_write_data_sem(struct inode *first,
3223 struct inode *second);
3224 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3225 struct inode *donor_inode);
3226 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3227 __u64 start_orig, __u64 start_donor,
3228 __u64 len, __u64 *moved_len);
3229
3230 /* page-io.c */
3231 extern int __init ext4_init_pageio(void);
3232 extern void ext4_exit_pageio(void);
3233 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3234 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3235 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3236 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3237 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3238 struct writeback_control *wbc);
3239 extern void ext4_end_io_rsv_work(struct work_struct *work);
3240 extern void ext4_io_submit(struct ext4_io_submit *io);
3241 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3242 struct page *page,
3243 int len,
3244 struct writeback_control *wbc,
3245 bool keep_towrite);
3246
3247 /* mmp.c */
3248 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3249
3250 /*
3251 * Add new method to test whether block and inode bitmaps are properly
3252 * initialized. With uninit_bg reading the block from disk is not enough
3253 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3254 */
3255 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3256
3257 static inline int bitmap_uptodate(struct buffer_head *bh)
3258 {
3259 return (buffer_uptodate(bh) &&
3260 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3261 }
3262 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3263 {
3264 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3265 }
3266
3267 /*
3268 * Disable DIO read nolock optimization, so new dioreaders will be forced
3269 * to grab i_mutex
3270 */
3271 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3272 {
3273 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3274 smp_mb();
3275 }
3276 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3277 {
3278 smp_mb();
3279 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3280 }
3281
3282 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3283
3284 /* For ioend & aio unwritten conversion wait queues */
3285 #define EXT4_WQ_HASH_SZ 37
3286 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3287 EXT4_WQ_HASH_SZ])
3288 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
3289 EXT4_WQ_HASH_SZ])
3290 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3291 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3292
3293 #define EXT4_RESIZING 0
3294 extern int ext4_resize_begin(struct super_block *sb);
3295 extern void ext4_resize_end(struct super_block *sb);
3296
3297 #endif /* __KERNEL__ */
3298
3299 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3300 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3301
3302 #endif /* _EXT4_H */
Linux with added FPC-III support