?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
[linux-2.6.git] / fs / ext4 / ext4.h
blobaf815ea9d7cc4b6e209e28eea6f9bc97a6f247ce
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/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #include <crypto/hash.h>
33 #ifdef __KERNEL__
34 #include <linux/compat.h>
35 #endif
36
37 /*
38 * The fourth extended filesystem constants/structures
39 */
40
41 /*
42 * Define EXT4FS_DEBUG to produce debug messages
43 */
44 #undef EXT4FS_DEBUG
45
46 /*
47 * Debug code
48 */
49 #ifdef EXT4FS_DEBUG
50 #define ext4_debug(f, a...) \
51 do { \
52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
53 __FILE__, __LINE__, __func__); \
54 printk(KERN_DEBUG f, ## a); \
55 } while (0)
56 #else
57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
58 #endif
59
60 /*
61 * Turn on EXT_DEBUG to get lots of info about extents operations.
62 */
63 #define EXT_DEBUG__
64 #ifdef EXT_DEBUG
65 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
66 #else
67 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
68 #endif
69
70 #define EXT4_ERROR_INODE(inode, fmt, a...) \
71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
72
73 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
75
76 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
78
79 /* data type for block offset of block group */
80 typedef int ext4_grpblk_t;
81
82 /* data type for filesystem-wide blocks number */
83 typedef unsigned long long ext4_fsblk_t;
84
85 /* data type for file logical block number */
86 typedef __u32 ext4_lblk_t;
87
88 /* data type for block group number */
89 typedef unsigned int ext4_group_t;
90
91 /*
92 * Flags used in mballoc's allocation_context flags field.
93 *
94 * Also used to show what's going on for debugging purposes when the
95 * flag field is exported via the traceport interface
96 */
97
98 /* prefer goal again. length */
99 #define EXT4_MB_HINT_MERGE 0x0001
100 /* blocks already reserved */
101 #define EXT4_MB_HINT_RESERVED 0x0002
102 /* metadata is being allocated */
103 #define EXT4_MB_HINT_METADATA 0x0004
104 /* first blocks in the file */
105 #define EXT4_MB_HINT_FIRST 0x0008
106 /* search for the best chunk */
107 #define EXT4_MB_HINT_BEST 0x0010
108 /* data is being allocated */
109 #define EXT4_MB_HINT_DATA 0x0020
110 /* don't preallocate (for tails) */
111 #define EXT4_MB_HINT_NOPREALLOC 0x0040
112 /* allocate for locality group */
113 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
114 /* allocate goal blocks or none */
115 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
116 /* goal is meaningful */
117 #define EXT4_MB_HINT_TRY_GOAL 0x0200
118 /* blocks already pre-reserved by delayed allocation */
119 #define EXT4_MB_DELALLOC_RESERVED 0x0400
120 /* We are doing stream allocation */
121 #define EXT4_MB_STREAM_ALLOC 0x0800
122 /* Use reserved root blocks if needed */
123 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
124 /* Use blocks from reserved pool */
125 #define EXT4_MB_USE_RESERVED 0x2000
126
127 struct ext4_allocation_request {
128 /* target inode for block we're allocating */
129 struct inode *inode;
130 /* how many blocks we want to allocate */
131 unsigned int len;
132 /* logical block in target inode */
133 ext4_lblk_t logical;
134 /* the closest logical allocated block to the left */
135 ext4_lblk_t lleft;
136 /* the closest logical allocated block to the right */
137 ext4_lblk_t lright;
138 /* phys. target (a hint) */
139 ext4_fsblk_t goal;
140 /* phys. block for the closest logical allocated block to the left */
141 ext4_fsblk_t pleft;
142 /* phys. block for the closest logical allocated block to the right */
143 ext4_fsblk_t pright;
144 /* flags. see above EXT4_MB_HINT_* */
145 unsigned int flags;
146 };
147
148 /*
149 * Logical to physical block mapping, used by ext4_map_blocks()
150 *
151 * This structure is used to pass requests into ext4_map_blocks() as
152 * well as to store the information returned by ext4_map_blocks(). It
153 * takes less room on the stack than a struct buffer_head.
154 */
155 #define EXT4_MAP_NEW (1 << BH_New)
156 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
157 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
158 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
159 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
160 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
161 * ext4_map_blocks wants to know whether or not the underlying cluster has
162 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
163 * the requested mapping was from previously mapped (or delayed allocated)
164 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
165 * should never appear on buffer_head's state flags.
166 */
167 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
168 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
169 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
170 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
171
172 struct ext4_map_blocks {
173 ext4_fsblk_t m_pblk;
174 ext4_lblk_t m_lblk;
175 unsigned int m_len;
176 unsigned int m_flags;
177 };
178
179 /*
180 * Flags for ext4_io_end->flags
181 */
182 #define EXT4_IO_END_UNWRITTEN 0x0001
183
184 /*
185 * For converting uninitialized extents on a work queue. 'handle' is used for
186 * buffered writeback.
187 */
188 typedef struct ext4_io_end {
189 struct list_head list; /* per-file finished IO list */
190 handle_t *handle; /* handle reserved for extent
191 * conversion */
192 struct inode *inode; /* file being written to */
193 struct bio *bio; /* Linked list of completed
194 * bios covering the extent */
195 unsigned int flag; /* unwritten or not */
196 loff_t offset; /* offset in the file */
197 ssize_t size; /* size of the extent */
198 atomic_t count; /* reference counter */
199 } ext4_io_end_t;
200
201 struct ext4_io_submit {
202 int io_op;
203 struct bio *io_bio;
204 ext4_io_end_t *io_end;
205 sector_t io_next_block;
206 };
207
208 /*
209 * Special inodes numbers
210 */
211 #define EXT4_BAD_INO 1 /* Bad blocks inode */
212 #define EXT4_ROOT_INO 2 /* Root inode */
213 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
214 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
215 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
216 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
217 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
218 #define EXT4_JOURNAL_INO 8 /* Journal inode */
219
220 /* First non-reserved inode for old ext4 filesystems */
221 #define EXT4_GOOD_OLD_FIRST_INO 11
222
223 /*
224 * Maximal count of links to a file
225 */
226 #define EXT4_LINK_MAX 65000
227
228 /*
229 * Macro-instructions used to manage several block sizes
230 */
231 #define EXT4_MIN_BLOCK_SIZE 1024
232 #define EXT4_MAX_BLOCK_SIZE 65536
233 #define EXT4_MIN_BLOCK_LOG_SIZE 10
234 #define EXT4_MAX_BLOCK_LOG_SIZE 16
235 #ifdef __KERNEL__
236 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
237 #else
238 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
239 #endif
240 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
241 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
242 EXT4_SB(s)->s_cluster_bits)
243 #ifdef __KERNEL__
244 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
245 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
246 #else
247 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
248 #endif
249 #ifdef __KERNEL__
250 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
251 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
252 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
253 #else
254 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
255 EXT4_GOOD_OLD_INODE_SIZE : \
256 (s)->s_inode_size)
257 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
258 EXT4_GOOD_OLD_FIRST_INO : \
259 (s)->s_first_ino)
260 #endif
261 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
262
263 /* Translate a block number to a cluster number */
264 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
265 /* Translate a cluster number to a block number */
266 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
267 /* Translate # of blks to # of clusters */
268 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
269 (sbi)->s_cluster_bits)
270
271 /*
272 * Structure of a blocks group descriptor
273 */
274 struct ext4_group_desc
275 {
276 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
277 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
278 __le32 bg_inode_table_lo; /* Inodes table block */
279 __le16 bg_free_blocks_count_lo;/* Free blocks count */
280 __le16 bg_free_inodes_count_lo;/* Free inodes count */
281 __le16 bg_used_dirs_count_lo; /* Directories count */
282 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
283 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
284 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
285 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
286 __le16 bg_itable_unused_lo; /* Unused inodes count */
287 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
288 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
289 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
290 __le32 bg_inode_table_hi; /* Inodes table block MSB */
291 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
292 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
293 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
294 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
295 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
296 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
297 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
298 __u32 bg_reserved;
299 };
300
301 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
302 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
303 sizeof(__le16))
304 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
305 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
306 sizeof(__le16))
307
308 /*
309 * Structure of a flex block group info
310 */
311
312 struct flex_groups {
313 atomic64_t free_clusters;
314 atomic_t free_inodes;
315 atomic_t used_dirs;
316 };
317
318 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
319 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
320 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
321
322 /*
323 * Macro-instructions used to manage group descriptors
324 */
325 #define EXT4_MIN_DESC_SIZE 32
326 #define EXT4_MIN_DESC_SIZE_64BIT 64
327 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
328 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
329 #ifdef __KERNEL__
330 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
331 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
332 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
333 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
334 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
335 #else
336 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
337 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
338 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
339 #endif
340
341 /*
342 * Constants relative to the data blocks
343 */
344 #define EXT4_NDIR_BLOCKS 12
345 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
346 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
347 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
348 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
349
350 /*
351 * Inode flags
352 */
353 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
354 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
355 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
356 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
357 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
358 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
359 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
360 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
361 /* Reserved for compression usage... */
362 #define EXT4_DIRTY_FL 0x00000100
363 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
364 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
365 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
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_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
379
380 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
381 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
382
383 /* Flags that should be inherited by new inodes from their parent. */
384 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
385 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
386 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
387 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
388
389 /* Flags that are appropriate for regular files (all but dir-specific ones). */
390 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
391
392 /* Flags that are appropriate for non-directories/regular files. */
393 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
394
395 /* Mask out flags that are inappropriate for the given type of inode. */
396 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
397 {
398 if (S_ISDIR(mode))
399 return flags;
400 else if (S_ISREG(mode))
401 return flags & EXT4_REG_FLMASK;
402 else
403 return flags & EXT4_OTHER_FLMASK;
404 }
405
406 /*
407 * Inode flags used for atomic set/get
408 */
409 enum {
410 EXT4_INODE_SECRM = 0, /* Secure deletion */
411 EXT4_INODE_UNRM = 1, /* Undelete */
412 EXT4_INODE_COMPR = 2, /* Compress file */
413 EXT4_INODE_SYNC = 3, /* Synchronous updates */
414 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
415 EXT4_INODE_APPEND = 5, /* writes to file may only append */
416 EXT4_INODE_NODUMP = 6, /* do not dump file */
417 EXT4_INODE_NOATIME = 7, /* do not update atime */
418 /* Reserved for compression usage... */
419 EXT4_INODE_DIRTY = 8,
420 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
421 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
422 EXT4_INODE_ECOMPR = 11, /* Compression error */
423 /* End compression flags --- maybe not all used */
424 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
425 EXT4_INODE_IMAGIC = 13, /* AFS directory */
426 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
427 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
428 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
429 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
430 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
431 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
432 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
433 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
434 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
435 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
436 };
437
438 /*
439 * Since it's pretty easy to mix up bit numbers and hex values, we use a
440 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
441 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
442 * any extra space in the compiled kernel image, otherwise, the build will fail.
443 * It's important that these values are the same, since we are using
444 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
445 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
446 * values found in ext2, ext3 and ext4 filesystems, and of course the values
447 * defined in e2fsprogs.
448 *
449 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
450 */
451 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
452 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
453
454 static inline void ext4_check_flag_values(void)
455 {
456 CHECK_FLAG_VALUE(SECRM);
457 CHECK_FLAG_VALUE(UNRM);
458 CHECK_FLAG_VALUE(COMPR);
459 CHECK_FLAG_VALUE(SYNC);
460 CHECK_FLAG_VALUE(IMMUTABLE);
461 CHECK_FLAG_VALUE(APPEND);
462 CHECK_FLAG_VALUE(NODUMP);
463 CHECK_FLAG_VALUE(NOATIME);
464 CHECK_FLAG_VALUE(DIRTY);
465 CHECK_FLAG_VALUE(COMPRBLK);
466 CHECK_FLAG_VALUE(NOCOMPR);
467 CHECK_FLAG_VALUE(ECOMPR);
468 CHECK_FLAG_VALUE(INDEX);
469 CHECK_FLAG_VALUE(IMAGIC);
470 CHECK_FLAG_VALUE(JOURNAL_DATA);
471 CHECK_FLAG_VALUE(NOTAIL);
472 CHECK_FLAG_VALUE(DIRSYNC);
473 CHECK_FLAG_VALUE(TOPDIR);
474 CHECK_FLAG_VALUE(HUGE_FILE);
475 CHECK_FLAG_VALUE(EXTENTS);
476 CHECK_FLAG_VALUE(EA_INODE);
477 CHECK_FLAG_VALUE(EOFBLOCKS);
478 CHECK_FLAG_VALUE(INLINE_DATA);
479 CHECK_FLAG_VALUE(RESERVED);
480 }
481
482 /* Used to pass group descriptor data when online resize is done */
483 struct ext4_new_group_input {
484 __u32 group; /* Group number for this data */
485 __u64 block_bitmap; /* Absolute block number of block bitmap */
486 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
487 __u64 inode_table; /* Absolute block number of inode table start */
488 __u32 blocks_count; /* Total number of blocks in this group */
489 __u16 reserved_blocks; /* Number of reserved blocks in this group */
490 __u16 unused;
491 };
492
493 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
494 struct compat_ext4_new_group_input {
495 u32 group;
496 compat_u64 block_bitmap;
497 compat_u64 inode_bitmap;
498 compat_u64 inode_table;
499 u32 blocks_count;
500 u16 reserved_blocks;
501 u16 unused;
502 };
503 #endif
504
505 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
506 struct ext4_new_group_data {
507 __u32 group;
508 __u64 block_bitmap;
509 __u64 inode_bitmap;
510 __u64 inode_table;
511 __u32 blocks_count;
512 __u16 reserved_blocks;
513 __u16 unused;
514 __u32 free_blocks_count;
515 };
516
517 /* Indexes used to index group tables in ext4_new_group_data */
518 enum {
519 BLOCK_BITMAP = 0, /* block bitmap */
520 INODE_BITMAP, /* inode bitmap */
521 INODE_TABLE, /* inode tables */
522 GROUP_TABLE_COUNT,
523 };
524
525 /*
526 * Flags used by ext4_map_blocks()
527 */
528 /* Allocate any needed blocks and/or convert an unitialized
529 extent to be an initialized ext4 */
530 #define EXT4_GET_BLOCKS_CREATE 0x0001
531 /* Request the creation of an unitialized extent */
532 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
533 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
534 EXT4_GET_BLOCKS_CREATE)
535 /* Caller is from the delayed allocation writeout path
536 * finally doing the actual allocation of delayed blocks */
537 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
538 /* caller is from the direct IO path, request to creation of an
539 unitialized extents if not allocated, split the uninitialized
540 extent if blocks has been preallocated already*/
541 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
542 #define EXT4_GET_BLOCKS_CONVERT 0x0010
543 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
544 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
545 /* Convert extent to initialized after IO complete */
546 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
547 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
548 /* Eventual metadata allocation (due to growing extent tree)
549 * should not fail, so try to use reserved blocks for that.*/
550 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
551 /* Don't normalize allocation size (used for fallocate) */
552 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
553 /* Request will not result in inode size update (user for fallocate) */
554 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
555 /* Do not take i_data_sem locking in ext4_map_blocks */
556 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
557 /* Do not put hole in extent cache */
558 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
559
560 /*
561 * The bit position of these flags must not overlap with any of the
562 * EXT4_GET_BLOCKS_*. They are used by ext4_ext_find_extent(),
563 * read_extent_tree_block(), ext4_split_extent_at(),
564 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
565 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
566 * caching the extents when reading from the extent tree while a
567 * truncate or punch hole operation is in progress.
568 */
569 #define EXT4_EX_NOCACHE 0x0400
570 #define EXT4_EX_FORCE_CACHE 0x0800
571
572 /*
573 * Flags used by ext4_free_blocks
574 */
575 #define EXT4_FREE_BLOCKS_METADATA 0x0001
576 #define EXT4_FREE_BLOCKS_FORGET 0x0002
577 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
578 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
579 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
580 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
581 #define EXT4_FREE_BLOCKS_RESERVE 0x0040
582
583 /*
584 * ioctl commands
585 */
586 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
587 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
588 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
589 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
590 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
591 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
592 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
593 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
594 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
595 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
596 #define EXT4_IOC_MIGRATE _IO('f', 9)
597 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
598 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
599 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
600 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
601 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
602 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
603 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
604
605 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
606 /*
607 * ioctl commands in 32 bit emulation
608 */
609 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
610 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
611 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
612 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
613 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
614 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
615 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
616 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
618 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
619 #endif
620
621 /* Max physical block we can address w/o extents */
622 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
623
624 /*
625 * Structure of an inode on the disk
626 */
627 struct ext4_inode {
628 __le16 i_mode; /* File mode */
629 __le16 i_uid; /* Low 16 bits of Owner Uid */
630 __le32 i_size_lo; /* Size in bytes */
631 __le32 i_atime; /* Access time */
632 __le32 i_ctime; /* Inode Change time */
633 __le32 i_mtime; /* Modification time */
634 __le32 i_dtime; /* Deletion Time */
635 __le16 i_gid; /* Low 16 bits of Group Id */
636 __le16 i_links_count; /* Links count */
637 __le32 i_blocks_lo; /* Blocks count */
638 __le32 i_flags; /* File flags */
639 union {
640 struct {
641 __le32 l_i_version;
642 } linux1;
643 struct {
644 __u32 h_i_translator;
645 } hurd1;
646 struct {
647 __u32 m_i_reserved1;
648 } masix1;
649 } osd1; /* OS dependent 1 */
650 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
651 __le32 i_generation; /* File version (for NFS) */
652 __le32 i_file_acl_lo; /* File ACL */
653 __le32 i_size_high;
654 __le32 i_obso_faddr; /* Obsoleted fragment address */
655 union {
656 struct {
657 __le16 l_i_blocks_high; /* were l_i_reserved1 */
658 __le16 l_i_file_acl_high;
659 __le16 l_i_uid_high; /* these 2 fields */
660 __le16 l_i_gid_high; /* were reserved2[0] */
661 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
662 __le16 l_i_reserved;
663 } linux2;
664 struct {
665 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
666 __u16 h_i_mode_high;
667 __u16 h_i_uid_high;
668 __u16 h_i_gid_high;
669 __u32 h_i_author;
670 } hurd2;
671 struct {
672 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
673 __le16 m_i_file_acl_high;
674 __u32 m_i_reserved2[2];
675 } masix2;
676 } osd2; /* OS dependent 2 */
677 __le16 i_extra_isize;
678 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
679 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
680 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
681 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
682 __le32 i_crtime; /* File Creation time */
683 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
684 __le32 i_version_hi; /* high 32 bits for 64-bit version */
685 };
686
687 struct move_extent {
688 __u32 reserved; /* should be zero */
689 __u32 donor_fd; /* donor file descriptor */
690 __u64 orig_start; /* logical start offset in block for orig */
691 __u64 donor_start; /* logical start offset in block for donor */
692 __u64 len; /* block length to be moved */
693 __u64 moved_len; /* moved block length */
694 };
695
696 #define EXT4_EPOCH_BITS 2
697 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
698 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
699
700 /*
701 * Extended fields will fit into an inode if the filesystem was formatted
702 * with large inodes (-I 256 or larger) and there are not currently any EAs
703 * consuming all of the available space. For new inodes we always reserve
704 * enough space for the kernel's known extended fields, but for inodes
705 * created with an old kernel this might not have been the case. None of
706 * the extended inode fields is critical for correct filesystem operation.
707 * This macro checks if a certain field fits in the inode. Note that
708 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
709 */
710 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
711 ((offsetof(typeof(*ext4_inode), field) + \
712 sizeof((ext4_inode)->field)) \
713 <= (EXT4_GOOD_OLD_INODE_SIZE + \
714 (einode)->i_extra_isize)) \
715
716 static inline __le32 ext4_encode_extra_time(struct timespec *time)
717 {
718 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
719 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
720 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
721 }
722
723 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
724 {
725 if (sizeof(time->tv_sec) > 4)
726 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
727 << 32;
728 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
729 }
730
731 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
732 do { \
733 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
734 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
735 (raw_inode)->xtime ## _extra = \
736 ext4_encode_extra_time(&(inode)->xtime); \
737 } while (0)
738
739 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
740 do { \
741 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
742 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
743 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
744 (raw_inode)->xtime ## _extra = \
745 ext4_encode_extra_time(&(einode)->xtime); \
746 } while (0)
747
748 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
749 do { \
750 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
751 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
752 ext4_decode_extra_time(&(inode)->xtime, \
753 raw_inode->xtime ## _extra); \
754 else \
755 (inode)->xtime.tv_nsec = 0; \
756 } while (0)
757
758 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
759 do { \
760 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
761 (einode)->xtime.tv_sec = \
762 (signed)le32_to_cpu((raw_inode)->xtime); \
763 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
764 ext4_decode_extra_time(&(einode)->xtime, \
765 raw_inode->xtime ## _extra); \
766 else \
767 (einode)->xtime.tv_nsec = 0; \
768 } while (0)
769
770 #define i_disk_version osd1.linux1.l_i_version
771
772 #if defined(__KERNEL__) || defined(__linux__)
773 #define i_reserved1 osd1.linux1.l_i_reserved1
774 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
775 #define i_blocks_high osd2.linux2.l_i_blocks_high
776 #define i_uid_low i_uid
777 #define i_gid_low i_gid
778 #define i_uid_high osd2.linux2.l_i_uid_high
779 #define i_gid_high osd2.linux2.l_i_gid_high
780 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
781
782 #elif defined(__GNU__)
783
784 #define i_translator osd1.hurd1.h_i_translator
785 #define i_uid_high osd2.hurd2.h_i_uid_high
786 #define i_gid_high osd2.hurd2.h_i_gid_high
787 #define i_author osd2.hurd2.h_i_author
788
789 #elif defined(__masix__)
790
791 #define i_reserved1 osd1.masix1.m_i_reserved1
792 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
793 #define i_reserved2 osd2.masix2.m_i_reserved2
794
795 #endif /* defined(__KERNEL__) || defined(__linux__) */
796
797 #include "extents_status.h"
798
799 /*
800 * fourth extended file system inode data in memory
801 */
802 struct ext4_inode_info {
803 __le32 i_data[15]; /* unconverted */
804 __u32 i_dtime;
805 ext4_fsblk_t i_file_acl;
806
807 /*
808 * i_block_group is the number of the block group which contains
809 * this file's inode. Constant across the lifetime of the inode,
810 * it is ued for making block allocation decisions - we try to
811 * place a file's data blocks near its inode block, and new inodes
812 * near to their parent directory's inode.
813 */
814 ext4_group_t i_block_group;
815 ext4_lblk_t i_dir_start_lookup;
816 #if (BITS_PER_LONG < 64)
817 unsigned long i_state_flags; /* Dynamic state flags */
818 #endif
819 unsigned long i_flags;
820
821 /*
822 * Extended attributes can be read independently of the main file
823 * data. Taking i_mutex even when reading would cause contention
824 * between readers of EAs and writers of regular file data, so
825 * instead we synchronize on xattr_sem when reading or changing
826 * EAs.
827 */
828 struct rw_semaphore xattr_sem;
829
830 struct list_head i_orphan; /* unlinked but open inodes */
831
832 /*
833 * i_disksize keeps track of what the inode size is ON DISK, not
834 * in memory. During truncate, i_size is set to the new size by
835 * the VFS prior to calling ext4_truncate(), but the filesystem won't
836 * set i_disksize to 0 until the truncate is actually under way.
837 *
838 * The intent is that i_disksize always represents the blocks which
839 * are used by this file. This allows recovery to restart truncate
840 * on orphans if we crash during truncate. We actually write i_disksize
841 * into the on-disk inode when writing inodes out, instead of i_size.
842 *
843 * The only time when i_disksize and i_size may be different is when
844 * a truncate is in progress. The only things which change i_disksize
845 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
846 */
847 loff_t i_disksize;
848
849 /*
850 * i_data_sem is for serialising ext4_truncate() against
851 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
852 * data tree are chopped off during truncate. We can't do that in
853 * ext4 because whenever we perform intermediate commits during
854 * truncate, the inode and all the metadata blocks *must* be in a
855 * consistent state which allows truncation of the orphans to restart
856 * during recovery. Hence we must fix the get_block-vs-truncate race
857 * by other means, so we have i_data_sem.
858 */
859 struct rw_semaphore i_data_sem;
860 struct inode vfs_inode;
861 struct jbd2_inode *jinode;
862
863 /*
864 * File creation time. Its function is same as that of
865 * struct timespec i_{a,c,m}time in the generic inode.
866 */
867 struct timespec i_crtime;
868
869 /* mballoc */
870 struct list_head i_prealloc_list;
871 spinlock_t i_prealloc_lock;
872
873 /* extents status tree */
874 struct ext4_es_tree i_es_tree;
875 rwlock_t i_es_lock;
876 struct list_head i_es_lru;
877 unsigned int i_es_lru_nr; /* protected by i_es_lock */
878 unsigned long i_touch_when; /* jiffies of last accessing */
879
880 /* ialloc */
881 ext4_group_t i_last_alloc_group;
882
883 /* allocation reservation info for delalloc */
884 /* In case of bigalloc, these refer to clusters rather than blocks */
885 unsigned int i_reserved_data_blocks;
886 unsigned int i_reserved_meta_blocks;
887 unsigned int i_allocated_meta_blocks;
888 ext4_lblk_t i_da_metadata_calc_last_lblock;
889 int i_da_metadata_calc_len;
890
891 /* on-disk additional length */
892 __u16 i_extra_isize;
893
894 /* Indicate the inline data space. */
895 u16 i_inline_off;
896 u16 i_inline_size;
897
898 #ifdef CONFIG_QUOTA
899 /* quota space reservation, managed internally by quota code */
900 qsize_t i_reserved_quota;
901 #endif
902
903 /* Lock protecting lists below */
904 spinlock_t i_completed_io_lock;
905 /*
906 * Completed IOs that need unwritten extents handling and have
907 * transaction reserved
908 */
909 struct list_head i_rsv_conversion_list;
910 /*
911 * Completed IOs that need unwritten extents handling and don't have
912 * transaction reserved
913 */
914 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
915 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
916 struct work_struct i_rsv_conversion_work;
917
918 spinlock_t i_block_reservation_lock;
919
920 /*
921 * Transactions that contain inode's metadata needed to complete
922 * fsync and fdatasync, respectively.
923 */
924 tid_t i_sync_tid;
925 tid_t i_datasync_tid;
926
927 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
928 __u32 i_csum_seed;
929 };
930
931 /*
932 * File system states
933 */
934 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
935 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
936 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
937
938 /*
939 * Misc. filesystem flags
940 */
941 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
942 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
943 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
944
945 /*
946 * Mount flags set via mount options or defaults
947 */
948 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
949 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
950 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
951 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
952 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
953 #define EXT4_MOUNT_ERRORS_MASK 0x00070
954 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
955 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
956 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
957 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
958 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
959 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
960 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
961 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
962 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
963 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
964 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
965 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
966 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
967 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
968 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
969 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
970 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
971 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
972 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
973 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
974 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
975 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
976 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
977
978 /*
979 * Mount flags set either automatically (could not be set by mount option)
980 * based on per file system feature or property or in special cases such as
981 * distinguishing between explicit mount option definition and default.
982 */
983 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
984 specified delalloc */
985 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
986 size of blocksize * 8
987 blocks */
988
989 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
990 ~EXT4_MOUNT_##opt
991 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
992 EXT4_MOUNT_##opt
993 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
994 EXT4_MOUNT_##opt)
995
996 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
997 ~EXT4_MOUNT2_##opt
998 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
999 EXT4_MOUNT2_##opt
1000 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1001 EXT4_MOUNT2_##opt)
1002
1003 #define ext4_test_and_set_bit __test_and_set_bit_le
1004 #define ext4_set_bit __set_bit_le
1005 #define ext4_set_bit_atomic ext2_set_bit_atomic
1006 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1007 #define ext4_clear_bit __clear_bit_le
1008 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1009 #define ext4_test_bit test_bit_le
1010 #define ext4_find_next_zero_bit find_next_zero_bit_le
1011 #define ext4_find_next_bit find_next_bit_le
1012
1013 extern void ext4_set_bits(void *bm, int cur, int len);
1014
1015 /*
1016 * Maximal mount counts between two filesystem checks
1017 */
1018 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1019 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1020
1021 /*
1022 * Behaviour when detecting errors
1023 */
1024 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1025 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1026 #define EXT4_ERRORS_PANIC 3 /* Panic */
1027 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1028
1029 /* Metadata checksum algorithm codes */
1030 #define EXT4_CRC32C_CHKSUM 1
1031
1032 /*
1033 * Structure of the super block
1034 */
1035 struct ext4_super_block {
1036 /*00*/ __le32 s_inodes_count; /* Inodes count */
1037 __le32 s_blocks_count_lo; /* Blocks count */
1038 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1039 __le32 s_free_blocks_count_lo; /* Free blocks count */
1040 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1041 __le32 s_first_data_block; /* First Data Block */
1042 __le32 s_log_block_size; /* Block size */
1043 __le32 s_log_cluster_size; /* Allocation cluster size */
1044 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1045 __le32 s_clusters_per_group; /* # Clusters per group */
1046 __le32 s_inodes_per_group; /* # Inodes per group */
1047 __le32 s_mtime; /* Mount time */
1048 /*30*/ __le32 s_wtime; /* Write time */
1049 __le16 s_mnt_count; /* Mount count */
1050 __le16 s_max_mnt_count; /* Maximal mount count */
1051 __le16 s_magic; /* Magic signature */
1052 __le16 s_state; /* File system state */
1053 __le16 s_errors; /* Behaviour when detecting errors */
1054 __le16 s_minor_rev_level; /* minor revision level */
1055 /*40*/ __le32 s_lastcheck; /* time of last check */
1056 __le32 s_checkinterval; /* max. time between checks */
1057 __le32 s_creator_os; /* OS */
1058 __le32 s_rev_level; /* Revision level */
1059 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1060 __le16 s_def_resgid; /* Default gid for reserved blocks */
1061 /*
1062 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1063 *
1064 * Note: the difference between the compatible feature set and
1065 * the incompatible feature set is that if there is a bit set
1066 * in the incompatible feature set that the kernel doesn't
1067 * know about, it should refuse to mount the filesystem.
1068 *
1069 * e2fsck's requirements are more strict; if it doesn't know
1070 * about a feature in either the compatible or incompatible
1071 * feature set, it must abort and not try to meddle with
1072 * things it doesn't understand...
1073 */
1074 __le32 s_first_ino; /* First non-reserved inode */
1075 __le16 s_inode_size; /* size of inode structure */
1076 __le16 s_block_group_nr; /* block group # of this superblock */
1077 __le32 s_feature_compat; /* compatible feature set */
1078 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1079 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1080 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1081 /*78*/ char s_volume_name[16]; /* volume name */
1082 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1083 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1084 /*
1085 * Performance hints. Directory preallocation should only
1086 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1087 */
1088 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1089 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1090 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1091 /*
1092 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1093 */
1094 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1095 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1096 __le32 s_journal_dev; /* device number of journal file */
1097 __le32 s_last_orphan; /* start of list of inodes to delete */
1098 __le32 s_hash_seed[4]; /* HTREE hash seed */
1099 __u8 s_def_hash_version; /* Default hash version to use */
1100 __u8 s_jnl_backup_type;
1101 __le16 s_desc_size; /* size of group descriptor */
1102 /*100*/ __le32 s_default_mount_opts;
1103 __le32 s_first_meta_bg; /* First metablock block group */
1104 __le32 s_mkfs_time; /* When the filesystem was created */
1105 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1106 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1107 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1108 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1109 __le32 s_free_blocks_count_hi; /* Free blocks count */
1110 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1111 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1112 __le32 s_flags; /* Miscellaneous flags */
1113 __le16 s_raid_stride; /* RAID stride */
1114 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1115 __le64 s_mmp_block; /* Block for multi-mount protection */
1116 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1117 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1118 __u8 s_checksum_type; /* metadata checksum algorithm used */
1119 __le16 s_reserved_pad;
1120 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1121 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1122 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1123 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1124 snapshot's future use */
1125 __le32 s_snapshot_list; /* inode number of the head of the
1126 on-disk snapshot list */
1127 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1128 __le32 s_error_count; /* number of fs errors */
1129 __le32 s_first_error_time; /* first time an error happened */
1130 __le32 s_first_error_ino; /* inode involved in first error */
1131 __le64 s_first_error_block; /* block involved of first error */
1132 __u8 s_first_error_func[32]; /* function where the error happened */
1133 __le32 s_first_error_line; /* line number where error happened */
1134 __le32 s_last_error_time; /* most recent time of an error */
1135 __le32 s_last_error_ino; /* inode involved in last error */
1136 __le32 s_last_error_line; /* line number where error happened */
1137 __le64 s_last_error_block; /* block involved of last error */
1138 __u8 s_last_error_func[32]; /* function where the error happened */
1139 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1140 __u8 s_mount_opts[64];
1141 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1142 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1143 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1144 __le32 s_reserved[108]; /* Padding to the end of the block */
1145 __le32 s_checksum; /* crc32c(superblock) */
1146 };
1147
1148 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1149
1150 #ifdef __KERNEL__
1151
1152 /*
1153 * run-time mount flags
1154 */
1155 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1156 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1157
1158 /*
1159 * fourth extended-fs super-block data in memory
1160 */
1161 struct ext4_sb_info {
1162 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1163 unsigned long s_inodes_per_block;/* Number of inodes per block */
1164 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1165 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1166 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1167 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1168 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1169 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1170 ext4_group_t s_groups_count; /* Number of groups in the fs */
1171 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1172 unsigned long s_overhead; /* # of fs overhead clusters */
1173 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1174 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1175 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1176 struct buffer_head * s_sbh; /* Buffer containing the super block */
1177 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1178 struct buffer_head **s_group_desc;
1179 unsigned int s_mount_opt;
1180 unsigned int s_mount_opt2;
1181 unsigned int s_mount_flags;
1182 unsigned int s_def_mount_opt;
1183 ext4_fsblk_t s_sb_block;
1184 atomic64_t s_resv_clusters;
1185 kuid_t s_resuid;
1186 kgid_t s_resgid;
1187 unsigned short s_mount_state;
1188 unsigned short s_pad;
1189 int s_addr_per_block_bits;
1190 int s_desc_per_block_bits;
1191 int s_inode_size;
1192 int s_first_ino;
1193 unsigned int s_inode_readahead_blks;
1194 unsigned int s_inode_goal;
1195 spinlock_t s_next_gen_lock;
1196 u32 s_next_generation;
1197 u32 s_hash_seed[4];
1198 int s_def_hash_version;
1199 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1200 struct percpu_counter s_freeclusters_counter;
1201 struct percpu_counter s_freeinodes_counter;
1202 struct percpu_counter s_dirs_counter;
1203 struct percpu_counter s_dirtyclusters_counter;
1204 struct blockgroup_lock *s_blockgroup_lock;
1205 struct proc_dir_entry *s_proc;
1206 struct kobject s_kobj;
1207 struct completion s_kobj_unregister;
1208 struct super_block *s_sb;
1209
1210 /* Journaling */
1211 struct journal_s *s_journal;
1212 struct list_head s_orphan;
1213 struct mutex s_orphan_lock;
1214 unsigned long s_resize_flags; /* Flags indicating if there
1215 is a resizer */
1216 unsigned long s_commit_interval;
1217 u32 s_max_batch_time;
1218 u32 s_min_batch_time;
1219 struct block_device *journal_bdev;
1220 #ifdef CONFIG_QUOTA
1221 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1222 int s_jquota_fmt; /* Format of quota to use */
1223 #endif
1224 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1225 struct rb_root system_blks;
1226
1227 #ifdef EXTENTS_STATS
1228 /* ext4 extents stats */
1229 unsigned long s_ext_min;
1230 unsigned long s_ext_max;
1231 unsigned long s_depth_max;
1232 spinlock_t s_ext_stats_lock;
1233 unsigned long s_ext_blocks;
1234 unsigned long s_ext_extents;
1235 #endif
1236
1237 /* for buddy allocator */
1238 struct ext4_group_info ***s_group_info;
1239 struct inode *s_buddy_cache;
1240 spinlock_t s_md_lock;
1241 unsigned short *s_mb_offsets;
1242 unsigned int *s_mb_maxs;
1243 unsigned int s_group_info_size;
1244
1245 /* tunables */
1246 unsigned long s_stripe;
1247 unsigned int s_mb_stream_request;
1248 unsigned int s_mb_max_to_scan;
1249 unsigned int s_mb_min_to_scan;
1250 unsigned int s_mb_stats;
1251 unsigned int s_mb_order2_reqs;
1252 unsigned int s_mb_group_prealloc;
1253 unsigned int s_max_dir_size_kb;
1254 /* where last allocation was done - for stream allocation */
1255 unsigned long s_mb_last_group;
1256 unsigned long s_mb_last_start;
1257
1258 /* stats for buddy allocator */
1259 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1260 atomic_t s_bal_success; /* we found long enough chunks */
1261 atomic_t s_bal_allocated; /* in blocks */
1262 atomic_t s_bal_ex_scanned; /* total extents scanned */
1263 atomic_t s_bal_goals; /* goal hits */
1264 atomic_t s_bal_breaks; /* too long searches */
1265 atomic_t s_bal_2orders; /* 2^order hits */
1266 spinlock_t s_bal_lock;
1267 unsigned long s_mb_buddies_generated;
1268 unsigned long long s_mb_generation_time;
1269 atomic_t s_mb_lost_chunks;
1270 atomic_t s_mb_preallocated;
1271 atomic_t s_mb_discarded;
1272 atomic_t s_lock_busy;
1273
1274 /* locality groups */
1275 struct ext4_locality_group __percpu *s_locality_groups;
1276
1277 /* for write statistics */
1278 unsigned long s_sectors_written_start;
1279 u64 s_kbytes_written;
1280
1281 /* the size of zero-out chunk */
1282 unsigned int s_extent_max_zeroout_kb;
1283
1284 unsigned int s_log_groups_per_flex;
1285 struct flex_groups *s_flex_groups;
1286 ext4_group_t s_flex_groups_allocated;
1287
1288 /* workqueue for reserved extent conversions (buffered io) */
1289 struct workqueue_struct *rsv_conversion_wq;
1290
1291 /* timer for periodic error stats printing */
1292 struct timer_list s_err_report;
1293
1294 /* Lazy inode table initialization info */
1295 struct ext4_li_request *s_li_request;
1296 /* Wait multiplier for lazy initialization thread */
1297 unsigned int s_li_wait_mult;
1298
1299 /* Kernel thread for multiple mount protection */
1300 struct task_struct *s_mmp_tsk;
1301
1302 /* record the last minlen when FITRIM is called. */
1303 atomic_t s_last_trim_minblks;
1304
1305 /* Reference to checksum algorithm driver via cryptoapi */
1306 struct crypto_shash *s_chksum_driver;
1307
1308 /* Precomputed FS UUID checksum for seeding other checksums */
1309 __u32 s_csum_seed;
1310
1311 /* Reclaim extents from extent status tree */
1312 struct shrinker s_es_shrinker;
1313 struct list_head s_es_lru;
1314 unsigned long s_es_last_sorted;
1315 struct percpu_counter s_extent_cache_cnt;
1316 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1317 };
1318
1319 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1320 {
1321 return sb->s_fs_info;
1322 }
1323 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1324 {
1325 return container_of(inode, struct ext4_inode_info, vfs_inode);
1326 }
1327
1328 static inline struct timespec ext4_current_time(struct inode *inode)
1329 {
1330 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1331 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1332 }
1333
1334 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1335 {
1336 return ino == EXT4_ROOT_INO ||
1337 ino == EXT4_USR_QUOTA_INO ||
1338 ino == EXT4_GRP_QUOTA_INO ||
1339 ino == EXT4_BOOT_LOADER_INO ||
1340 ino == EXT4_JOURNAL_INO ||
1341 ino == EXT4_RESIZE_INO ||
1342 (ino >= EXT4_FIRST_INO(sb) &&
1343 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1344 }
1345
1346 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1347 struct ext4_io_end *io_end)
1348 {
1349 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1350 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1351 atomic_inc(&EXT4_I(inode)->i_unwritten);
1352 }
1353 }
1354
1355 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1356 {
1357 return inode->i_private;
1358 }
1359
1360 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1361 {
1362 inode->i_private = io;
1363 }
1364
1365 /*
1366 * Inode dynamic state flags
1367 */
1368 enum {
1369 EXT4_STATE_JDATA, /* journaled data exists */
1370 EXT4_STATE_NEW, /* inode is newly created */
1371 EXT4_STATE_XATTR, /* has in-inode xattrs */
1372 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1373 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1374 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1375 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1376 EXT4_STATE_NEWENTRY, /* File just added to dir */
1377 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1378 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1379 nolocking */
1380 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1381 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1382 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1383 };
1384
1385 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1386 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1387 { \
1388 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1389 } \
1390 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1391 { \
1392 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1393 } \
1394 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1395 { \
1396 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1397 }
1398
1399 EXT4_INODE_BIT_FNS(flag, flags, 0)
1400 #if (BITS_PER_LONG < 64)
1401 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1402
1403 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1404 {
1405 (ei)->i_state_flags = 0;
1406 }
1407 #else
1408 EXT4_INODE_BIT_FNS(state, flags, 32)
1409
1410 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1411 {
1412 /* We depend on the fact that callers will set i_flags */
1413 }
1414 #endif
1415 #else
1416 /* Assume that user mode programs are passing in an ext4fs superblock, not
1417 * a kernel struct super_block. This will allow us to call the feature-test
1418 * macros from user land. */
1419 #define EXT4_SB(sb) (sb)
1420 #endif
1421
1422 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1423
1424 /*
1425 * Codes for operating systems
1426 */
1427 #define EXT4_OS_LINUX 0
1428 #define EXT4_OS_HURD 1
1429 #define EXT4_OS_MASIX 2
1430 #define EXT4_OS_FREEBSD 3
1431 #define EXT4_OS_LITES 4
1432
1433 /*
1434 * Revision levels
1435 */
1436 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1437 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1438
1439 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1440 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1441
1442 #define EXT4_GOOD_OLD_INODE_SIZE 128
1443
1444 /*
1445 * Feature set definitions
1446 */
1447
1448 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1449 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1450 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1451 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1452 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1453 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1454 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1455 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1456 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1457 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1458 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1459 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1460 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1461 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1462 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1463 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1464 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1465 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1466
1467 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1468 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1469 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1470 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1471 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1472 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1473
1474 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1475 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1476 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1477 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1478 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1479 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1480 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1481 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1482 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1483 /*
1484 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1485 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1486 * all other data structures' checksums. However, the METADATA_CSUM and
1487 * GDT_CSUM bits are mutually exclusive.
1488 */
1489 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1490
1491 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1492 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1493 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1494 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1495 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1496 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1497 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1498 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1499 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1500 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1501 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1502 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1503 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1504 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1505
1506 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1507 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1508 EXT4_FEATURE_INCOMPAT_META_BG)
1509 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1510 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1511 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1512
1513 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1514 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1515 EXT4_FEATURE_INCOMPAT_RECOVER| \
1516 EXT4_FEATURE_INCOMPAT_META_BG)
1517 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1518 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1519 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1520
1521 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1522 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1523 EXT4_FEATURE_INCOMPAT_RECOVER| \
1524 EXT4_FEATURE_INCOMPAT_META_BG| \
1525 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1526 EXT4_FEATURE_INCOMPAT_64BIT| \
1527 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1528 EXT4_FEATURE_INCOMPAT_MMP | \
1529 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1530 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1531 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1532 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1533 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1534 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1535 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1536 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1537 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1538 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1539 EXT4_FEATURE_RO_COMPAT_QUOTA)
1540
1541 /*
1542 * Default values for user and/or group using reserved blocks
1543 */
1544 #define EXT4_DEF_RESUID 0
1545 #define EXT4_DEF_RESGID 0
1546
1547 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1548
1549 /*
1550 * Default mount options
1551 */
1552 #define EXT4_DEFM_DEBUG 0x0001
1553 #define EXT4_DEFM_BSDGROUPS 0x0002
1554 #define EXT4_DEFM_XATTR_USER 0x0004
1555 #define EXT4_DEFM_ACL 0x0008
1556 #define EXT4_DEFM_UID16 0x0010
1557 #define EXT4_DEFM_JMODE 0x0060
1558 #define EXT4_DEFM_JMODE_DATA 0x0020
1559 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1560 #define EXT4_DEFM_JMODE_WBACK 0x0060
1561 #define EXT4_DEFM_NOBARRIER 0x0100
1562 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1563 #define EXT4_DEFM_DISCARD 0x0400
1564 #define EXT4_DEFM_NODELALLOC 0x0800
1565
1566 /*
1567 * Default journal batch times
1568 */
1569 #define EXT4_DEF_MIN_BATCH_TIME 0
1570 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1571
1572 /*
1573 * Minimum number of groups in a flexgroup before we separate out
1574 * directories into the first block group of a flexgroup
1575 */
1576 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1577
1578 /*
1579 * Structure of a directory entry
1580 */
1581 #define EXT4_NAME_LEN 255
1582
1583 struct ext4_dir_entry {
1584 __le32 inode; /* Inode number */
1585 __le16 rec_len; /* Directory entry length */
1586 __le16 name_len; /* Name length */
1587 char name[EXT4_NAME_LEN]; /* File name */
1588 };
1589
1590 /*
1591 * The new version of the directory entry. Since EXT4 structures are
1592 * stored in intel byte order, and the name_len field could never be
1593 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1594 * file_type field.
1595 */
1596 struct ext4_dir_entry_2 {
1597 __le32 inode; /* Inode number */
1598 __le16 rec_len; /* Directory entry length */
1599 __u8 name_len; /* Name length */
1600 __u8 file_type;
1601 char name[EXT4_NAME_LEN]; /* File name */
1602 };
1603
1604 /*
1605 * This is a bogus directory entry at the end of each leaf block that
1606 * records checksums.
1607 */
1608 struct ext4_dir_entry_tail {
1609 __le32 det_reserved_zero1; /* Pretend to be unused */
1610 __le16 det_rec_len; /* 12 */
1611 __u8 det_reserved_zero2; /* Zero name length */
1612 __u8 det_reserved_ft; /* 0xDE, fake file type */
1613 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1614 };
1615
1616 #define EXT4_DIRENT_TAIL(block, blocksize) \
1617 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1618 ((blocksize) - \
1619 sizeof(struct ext4_dir_entry_tail))))
1620
1621 /*
1622 * Ext4 directory file types. Only the low 3 bits are used. The
1623 * other bits are reserved for now.
1624 */
1625 #define EXT4_FT_UNKNOWN 0
1626 #define EXT4_FT_REG_FILE 1
1627 #define EXT4_FT_DIR 2
1628 #define EXT4_FT_CHRDEV 3
1629 #define EXT4_FT_BLKDEV 4
1630 #define EXT4_FT_FIFO 5
1631 #define EXT4_FT_SOCK 6
1632 #define EXT4_FT_SYMLINK 7
1633
1634 #define EXT4_FT_MAX 8
1635
1636 #define EXT4_FT_DIR_CSUM 0xDE
1637
1638 /*
1639 * EXT4_DIR_PAD defines the directory entries boundaries
1640 *
1641 * NOTE: It must be a multiple of 4
1642 */
1643 #define EXT4_DIR_PAD 4
1644 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1645 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1646 ~EXT4_DIR_ROUND)
1647 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1648
1649 /*
1650 * If we ever get support for fs block sizes > page_size, we'll need
1651 * to remove the #if statements in the next two functions...
1652 */
1653 static inline unsigned int
1654 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1655 {
1656 unsigned len = le16_to_cpu(dlen);
1657
1658 #if (PAGE_CACHE_SIZE >= 65536)
1659 if (len == EXT4_MAX_REC_LEN || len == 0)
1660 return blocksize;
1661 return (len & 65532) | ((len & 3) << 16);
1662 #else
1663 return len;
1664 #endif
1665 }
1666
1667 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1668 {
1669 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1670 BUG();
1671 #if (PAGE_CACHE_SIZE >= 65536)
1672 if (len < 65536)
1673 return cpu_to_le16(len);
1674 if (len == blocksize) {
1675 if (blocksize == 65536)
1676 return cpu_to_le16(EXT4_MAX_REC_LEN);
1677 else
1678 return cpu_to_le16(0);
1679 }
1680 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1681 #else
1682 return cpu_to_le16(len);
1683 #endif
1684 }
1685
1686 /*
1687 * Hash Tree Directory indexing
1688 * (c) Daniel Phillips, 2001
1689 */
1690
1691 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1692 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1693 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1694 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1695 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1696
1697 /* Legal values for the dx_root hash_version field: */
1698
1699 #define DX_HASH_LEGACY 0
1700 #define DX_HASH_HALF_MD4 1
1701 #define DX_HASH_TEA 2
1702 #define DX_HASH_LEGACY_UNSIGNED 3
1703 #define DX_HASH_HALF_MD4_UNSIGNED 4
1704 #define DX_HASH_TEA_UNSIGNED 5
1705
1706 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1707 const void *address, unsigned int length)
1708 {
1709 struct {
1710 struct shash_desc shash;
1711 char ctx[4];
1712 } desc;
1713 int err;
1714
1715 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1716
1717 desc.shash.tfm = sbi->s_chksum_driver;
1718 desc.shash.flags = 0;
1719 *(u32 *)desc.ctx = crc;
1720
1721 err = crypto_shash_update(&desc.shash, address, length);
1722 BUG_ON(err);
1723
1724 return *(u32 *)desc.ctx;
1725 }
1726
1727 #ifdef __KERNEL__
1728
1729 /* hash info structure used by the directory hash */
1730 struct dx_hash_info
1731 {
1732 u32 hash;
1733 u32 minor_hash;
1734 int hash_version;
1735 u32 *seed;
1736 };
1737
1738
1739 /* 32 and 64 bit signed EOF for dx directories */
1740 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1741 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1742
1743
1744 /*
1745 * Control parameters used by ext4_htree_next_block
1746 */
1747 #define HASH_NB_ALWAYS 1
1748
1749
1750 /*
1751 * Describe an inode's exact location on disk and in memory
1752 */
1753 struct ext4_iloc
1754 {
1755 struct buffer_head *bh;
1756 unsigned long offset;
1757 ext4_group_t block_group;
1758 };
1759
1760 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1761 {
1762 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1763 }
1764
1765 /*
1766 * This structure is stuffed into the struct file's private_data field
1767 * for directories. It is where we put information so that we can do
1768 * readdir operations in hash tree order.
1769 */
1770 struct dir_private_info {
1771 struct rb_root root;
1772 struct rb_node *curr_node;
1773 struct fname *extra_fname;
1774 loff_t last_pos;
1775 __u32 curr_hash;
1776 __u32 curr_minor_hash;
1777 __u32 next_hash;
1778 };
1779
1780 /* calculate the first block number of the group */
1781 static inline ext4_fsblk_t
1782 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1783 {
1784 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1785 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1786 }
1787
1788 /*
1789 * Special error return code only used by dx_probe() and its callers.
1790 */
1791 #define ERR_BAD_DX_DIR -75000
1792
1793 /*
1794 * Timeout and state flag for lazy initialization inode thread.
1795 */
1796 #define EXT4_DEF_LI_WAIT_MULT 10
1797 #define EXT4_DEF_LI_MAX_START_DELAY 5
1798 #define EXT4_LAZYINIT_QUIT 0x0001
1799 #define EXT4_LAZYINIT_RUNNING 0x0002
1800
1801 /*
1802 * Lazy inode table initialization info
1803 */
1804 struct ext4_lazy_init {
1805 unsigned long li_state;
1806 struct list_head li_request_list;
1807 struct mutex li_list_mtx;
1808 };
1809
1810 struct ext4_li_request {
1811 struct super_block *lr_super;
1812 struct ext4_sb_info *lr_sbi;
1813 ext4_group_t lr_next_group;
1814 struct list_head lr_request;
1815 unsigned long lr_next_sched;
1816 unsigned long lr_timeout;
1817 };
1818
1819 struct ext4_features {
1820 struct kobject f_kobj;
1821 struct completion f_kobj_unregister;
1822 };
1823
1824 /*
1825 * This structure will be used for multiple mount protection. It will be
1826 * written into the block number saved in the s_mmp_block field in the
1827 * superblock. Programs that check MMP should assume that if
1828 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1829 * to use the filesystem, regardless of how old the timestamp is.
1830 */
1831 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1832 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1833 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1834 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1835
1836 struct mmp_struct {
1837 __le32 mmp_magic; /* Magic number for MMP */
1838 __le32 mmp_seq; /* Sequence no. updated periodically */
1839
1840 /*
1841 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1842 * purposes and do not affect the correctness of the algorithm
1843 */
1844 __le64 mmp_time; /* Time last updated */
1845 char mmp_nodename[64]; /* Node which last updated MMP block */
1846 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1847
1848 /*
1849 * mmp_check_interval is used to verify if the MMP block has been
1850 * updated on the block device. The value is updated based on the
1851 * maximum time to write the MMP block during an update cycle.
1852 */
1853 __le16 mmp_check_interval;
1854
1855 __le16 mmp_pad1;
1856 __le32 mmp_pad2[226];
1857 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1858 };
1859
1860 /* arguments passed to the mmp thread */
1861 struct mmpd_data {
1862 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1863 struct super_block *sb; /* super block of the fs */
1864 };
1865
1866 /*
1867 * Check interval multiplier
1868 * The MMP block is written every update interval and initially checked every
1869 * update interval x the multiplier (the value is then adapted based on the
1870 * write latency). The reason is that writes can be delayed under load and we
1871 * don't want readers to incorrectly assume that the filesystem is no longer
1872 * in use.
1873 */
1874 #define EXT4_MMP_CHECK_MULT 2UL
1875
1876 /*
1877 * Minimum interval for MMP checking in seconds.
1878 */
1879 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1880
1881 /*
1882 * Maximum interval for MMP checking in seconds.
1883 */
1884 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1885
1886 /*
1887 * Function prototypes
1888 */
1889
1890 /*
1891 * Ok, these declarations are also in <linux/kernel.h> but none of the
1892 * ext4 source programs needs to include it so they are duplicated here.
1893 */
1894 # define NORET_TYPE /**/
1895 # define ATTRIB_NORET __attribute__((noreturn))
1896 # define NORET_AND noreturn,
1897
1898 /* bitmap.c */
1899 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1900 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1901 struct ext4_group_desc *gdp,
1902 struct buffer_head *bh, int sz);
1903 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1904 struct ext4_group_desc *gdp,
1905 struct buffer_head *bh, int sz);
1906 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1907 struct ext4_group_desc *gdp,
1908 struct buffer_head *bh);
1909 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1910 struct ext4_group_desc *gdp,
1911 struct buffer_head *bh);
1912
1913 /* balloc.c */
1914 extern void ext4_get_group_no_and_offset(struct super_block *sb,
1915 ext4_fsblk_t blocknr,
1916 ext4_group_t *blockgrpp,
1917 ext4_grpblk_t *offsetp);
1918 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1919 ext4_fsblk_t block);
1920
1921 extern void ext4_validate_block_bitmap(struct super_block *sb,
1922 struct ext4_group_desc *desc,
1923 ext4_group_t block_group,
1924 struct buffer_head *bh);
1925 extern unsigned int ext4_block_group(struct super_block *sb,
1926 ext4_fsblk_t blocknr);
1927 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1928 ext4_fsblk_t blocknr);
1929 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1930 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1931 ext4_group_t group);
1932 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1933 ext4_fsblk_t goal,
1934 unsigned int flags,
1935 unsigned long *count,
1936 int *errp);
1937 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1938 s64 nclusters, unsigned int flags);
1939 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1940 extern void ext4_check_blocks_bitmap(struct super_block *);
1941 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1942 ext4_group_t block_group,
1943 struct buffer_head ** bh);
1944 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1945
1946 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1947 ext4_group_t block_group);
1948 extern int ext4_wait_block_bitmap(struct super_block *sb,
1949 ext4_group_t block_group,
1950 struct buffer_head *bh);
1951 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1952 ext4_group_t block_group);
1953 extern void ext4_init_block_bitmap(struct super_block *sb,
1954 struct buffer_head *bh,
1955 ext4_group_t group,
1956 struct ext4_group_desc *desc);
1957 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1958 ext4_group_t block_group,
1959 struct ext4_group_desc *gdp);
1960 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1961 ext4_group_t block_group,
1962 struct ext4_group_desc *gdp);
1963 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1964
1965 /* dir.c */
1966 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1967 struct file *,
1968 struct ext4_dir_entry_2 *,
1969 struct buffer_head *, char *, int,
1970 unsigned int);
1971 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1972 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1973 (de), (bh), (buf), (size), (offset)))
1974 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1975 __u32 minor_hash,
1976 struct ext4_dir_entry_2 *dirent);
1977 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1978 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1979 struct buffer_head *bh,
1980 void *buf, int buf_size,
1981 const char *name, int namelen,
1982 struct ext4_dir_entry_2 **dest_de);
1983 void ext4_insert_dentry(struct inode *inode,
1984 struct ext4_dir_entry_2 *de,
1985 int buf_size,
1986 const char *name, int namelen);
1987 static inline void ext4_update_dx_flag(struct inode *inode)
1988 {
1989 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1990 EXT4_FEATURE_COMPAT_DIR_INDEX))
1991 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1992 }
1993 static unsigned char ext4_filetype_table[] = {
1994 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1995 };
1996
1997 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
1998 {
1999 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2000 (filetype >= EXT4_FT_MAX))
2001 return DT_UNKNOWN;
2002
2003 return ext4_filetype_table[filetype];
2004 }
2005
2006 /* fsync.c */
2007 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2008
2009 /* hash.c */
2010 extern int ext4fs_dirhash(const char *name, int len, struct
2011 dx_hash_info *hinfo);
2012
2013 /* ialloc.c */
2014 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2015 const struct qstr *qstr, __u32 goal,
2016 uid_t *owner, int handle_type,
2017 unsigned int line_no, int nblocks);
2018
2019 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2020 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2021 0, 0, 0)
2022 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2023 type, nblocks) \
2024 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2025 (type), __LINE__, (nblocks))
2026
2027
2028 extern void ext4_free_inode(handle_t *, struct inode *);
2029 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2030 extern unsigned long ext4_count_free_inodes(struct super_block *);
2031 extern unsigned long ext4_count_dirs(struct super_block *);
2032 extern void ext4_check_inodes_bitmap(struct super_block *);
2033 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2034 extern int ext4_init_inode_table(struct super_block *sb,
2035 ext4_group_t group, int barrier);
2036 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2037
2038 /* mballoc.c */
2039 extern long ext4_mb_stats;
2040 extern long ext4_mb_max_to_scan;
2041 extern int ext4_mb_init(struct super_block *);
2042 extern int ext4_mb_release(struct super_block *);
2043 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2044 struct ext4_allocation_request *, int *);
2045 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2046 extern void ext4_discard_preallocations(struct inode *);
2047 extern int __init ext4_init_mballoc(void);
2048 extern void ext4_exit_mballoc(void);
2049 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2050 struct buffer_head *bh, ext4_fsblk_t block,
2051 unsigned long count, int flags);
2052 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2053 ext4_group_t ngroups);
2054 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2055 ext4_group_t i, struct ext4_group_desc *desc);
2056 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2057 ext4_fsblk_t block, unsigned long count);
2058 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2059
2060 /* inode.c */
2061 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2062 ext4_lblk_t, int, int *);
2063 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2064 ext4_lblk_t, int, int *);
2065 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2066 struct buffer_head *bh_result, int create);
2067 int ext4_get_block(struct inode *inode, sector_t iblock,
2068 struct buffer_head *bh_result, int create);
2069 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2070 struct buffer_head *bh, int create);
2071 int ext4_walk_page_buffers(handle_t *handle,
2072 struct buffer_head *head,
2073 unsigned from,
2074 unsigned to,
2075 int *partial,
2076 int (*fn)(handle_t *handle,
2077 struct buffer_head *bh));
2078 int do_journal_get_write_access(handle_t *handle,
2079 struct buffer_head *bh);
2080 #define FALL_BACK_TO_NONDELALLOC 1
2081 #define CONVERT_INLINE_DATA 2
2082
2083 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2084 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2085 extern int ext4_setattr(struct dentry *, struct iattr *);
2086 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2087 struct kstat *stat);
2088 extern void ext4_evict_inode(struct inode *);
2089 extern void ext4_clear_inode(struct inode *);
2090 extern int ext4_sync_inode(handle_t *, struct inode *);
2091 extern void ext4_dirty_inode(struct inode *, int);
2092 extern int ext4_change_inode_journal_flag(struct inode *, int);
2093 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2094 extern int ext4_inode_attach_jinode(struct inode *inode);
2095 extern int ext4_can_truncate(struct inode *inode);
2096 extern void ext4_truncate(struct inode *);
2097 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2098 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2099 extern void ext4_set_inode_flags(struct inode *);
2100 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2101 extern int ext4_alloc_da_blocks(struct inode *inode);
2102 extern void ext4_set_aops(struct inode *inode);
2103 extern int ext4_writepage_trans_blocks(struct inode *);
2104 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2105 extern int ext4_block_truncate_page(handle_t *handle,
2106 struct address_space *mapping, loff_t from);
2107 extern int ext4_block_zero_page_range(handle_t *handle,
2108 struct address_space *mapping, loff_t from, loff_t length);
2109 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2110 loff_t lstart, loff_t lend);
2111 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2112 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2113 extern void ext4_da_update_reserve_space(struct inode *inode,
2114 int used, int quota_claim);
2115
2116 /* indirect.c */
2117 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2118 struct ext4_map_blocks *map, int flags);
2119 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2120 const struct iovec *iov, loff_t offset,
2121 unsigned long nr_segs);
2122 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2123 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2124 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2125 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
2126 ext4_lblk_t first, ext4_lblk_t stop);
2127
2128 /* ioctl.c */
2129 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2130 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2131
2132 /* migrate.c */
2133 extern int ext4_ext_migrate(struct inode *);
2134 extern int ext4_ind_migrate(struct inode *inode);
2135
2136 /* namei.c */
2137 extern int ext4_dirent_csum_verify(struct inode *inode,
2138 struct ext4_dir_entry *dirent);
2139 extern int ext4_orphan_add(handle_t *, struct inode *);
2140 extern int ext4_orphan_del(handle_t *, struct inode *);
2141 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2142 __u32 start_minor_hash, __u32 *next_hash);
2143 extern int search_dir(struct buffer_head *bh,
2144 char *search_buf,
2145 int buf_size,
2146 struct inode *dir,
2147 const struct qstr *d_name,
2148 unsigned int offset,
2149 struct ext4_dir_entry_2 **res_dir);
2150 extern int ext4_generic_delete_entry(handle_t *handle,
2151 struct inode *dir,
2152 struct ext4_dir_entry_2 *de_del,
2153 struct buffer_head *bh,
2154 void *entry_buf,
2155 int buf_size,
2156 int csum_size);
2157
2158 /* resize.c */
2159 extern int ext4_group_add(struct super_block *sb,
2160 struct ext4_new_group_data *input);
2161 extern int ext4_group_extend(struct super_block *sb,
2162 struct ext4_super_block *es,
2163 ext4_fsblk_t n_blocks_count);
2164 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2165
2166 /* super.c */
2167 extern int ext4_calculate_overhead(struct super_block *sb);
2168 extern int ext4_superblock_csum_verify(struct super_block *sb,
2169 struct ext4_super_block *es);
2170 extern void ext4_superblock_csum_set(struct super_block *sb);
2171 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2172 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2173 extern void ext4_kvfree(void *ptr);
2174 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2175 ext4_group_t ngroup);
2176 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2177 char nbuf[16]);
2178
2179 extern __printf(4, 5)
2180 void __ext4_error(struct super_block *, const char *, unsigned int,
2181 const char *, ...);
2182 extern __printf(5, 6)
2183 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2184 const char *, ...);
2185 extern __printf(5, 6)
2186 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2187 const char *, ...);
2188 extern void __ext4_std_error(struct super_block *, const char *,
2189 unsigned int, int);
2190 extern __printf(4, 5)
2191 void __ext4_abort(struct super_block *, const char *, unsigned int,
2192 const char *, ...);
2193 extern __printf(4, 5)
2194 void __ext4_warning(struct super_block *, const char *, unsigned int,
2195 const char *, ...);
2196 extern __printf(3, 4)
2197 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2198 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2199 const char *, unsigned int, const char *);
2200 extern __printf(7, 8)
2201 void __ext4_grp_locked_error(const char *, unsigned int,
2202 struct super_block *, ext4_group_t,
2203 unsigned long, ext4_fsblk_t,
2204 const char *, ...);
2205
2206 #ifdef CONFIG_PRINTK
2207
2208 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2209 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2210 #define ext4_error_file(file, func, line, block, fmt, ...) \
2211 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2212 #define ext4_error(sb, fmt, ...) \
2213 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2214 #define ext4_abort(sb, fmt, ...) \
2215 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2216 #define ext4_warning(sb, fmt, ...) \
2217 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2218 #define ext4_msg(sb, level, fmt, ...) \
2219 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2220 #define dump_mmp_msg(sb, mmp, msg) \
2221 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2222 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2223 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2224 fmt, ##__VA_ARGS__)
2225
2226 #else
2227
2228 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2229 do { \
2230 no_printk(fmt, ##__VA_ARGS__); \
2231 __ext4_error_inode(inode, "", 0, block, " "); \
2232 } while (0)
2233 #define ext4_error_file(file, func, line, block, fmt, ...) \
2234 do { \
2235 no_printk(fmt, ##__VA_ARGS__); \
2236 __ext4_error_file(file, "", 0, block, " "); \
2237 } while (0)
2238 #define ext4_error(sb, fmt, ...) \
2239 do { \
2240 no_printk(fmt, ##__VA_ARGS__); \
2241 __ext4_error(sb, "", 0, " "); \
2242 } while (0)
2243 #define ext4_abort(sb, fmt, ...) \
2244 do { \
2245 no_printk(fmt, ##__VA_ARGS__); \
2246 __ext4_abort(sb, "", 0, " "); \
2247 } while (0)
2248 #define ext4_warning(sb, fmt, ...) \
2249 do { \
2250 no_printk(fmt, ##__VA_ARGS__); \
2251 __ext4_warning(sb, "", 0, " "); \
2252 } while (0)
2253 #define ext4_msg(sb, level, fmt, ...) \
2254 do { \
2255 no_printk(fmt, ##__VA_ARGS__); \
2256 __ext4_msg(sb, "", " "); \
2257 } while (0)
2258 #define dump_mmp_msg(sb, mmp, msg) \
2259 __dump_mmp_msg(sb, mmp, "", 0, "")
2260 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2261 do { \
2262 no_printk(fmt, ##__VA_ARGS__); \
2263 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2264 } while (0)
2265
2266 #endif
2267
2268 extern void ext4_update_dynamic_rev(struct super_block *sb);
2269 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2270 __u32 compat);
2271 extern int ext4_update_rocompat_feature(handle_t *handle,
2272 struct super_block *sb, __u32 rocompat);
2273 extern int ext4_update_incompat_feature(handle_t *handle,
2274 struct super_block *sb, __u32 incompat);
2275 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2276 struct ext4_group_desc *bg);
2277 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2278 struct ext4_group_desc *bg);
2279 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2280 struct ext4_group_desc *bg);
2281 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2282 struct ext4_group_desc *bg);
2283 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2284 struct ext4_group_desc *bg);
2285 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2286 struct ext4_group_desc *bg);
2287 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2288 struct ext4_group_desc *bg);
2289 extern void ext4_block_bitmap_set(struct super_block *sb,
2290 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2291 extern void ext4_inode_bitmap_set(struct super_block *sb,
2292 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2293 extern void ext4_inode_table_set(struct super_block *sb,
2294 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2295 extern void ext4_free_group_clusters_set(struct super_block *sb,
2296 struct ext4_group_desc *bg,
2297 __u32 count);
2298 extern void ext4_free_inodes_set(struct super_block *sb,
2299 struct ext4_group_desc *bg, __u32 count);
2300 extern void ext4_used_dirs_set(struct super_block *sb,
2301 struct ext4_group_desc *bg, __u32 count);
2302 extern void ext4_itable_unused_set(struct super_block *sb,
2303 struct ext4_group_desc *bg, __u32 count);
2304 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2305 struct ext4_group_desc *gdp);
2306 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2307 struct ext4_group_desc *gdp);
2308 extern int ext4_register_li_request(struct super_block *sb,
2309 ext4_group_t first_not_zeroed);
2310
2311 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2312 {
2313 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2314 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2315 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2316 }
2317
2318 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2319 {
2320 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2321 le32_to_cpu(es->s_blocks_count_lo);
2322 }
2323
2324 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2325 {
2326 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2327 le32_to_cpu(es->s_r_blocks_count_lo);
2328 }
2329
2330 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2331 {
2332 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2333 le32_to_cpu(es->s_free_blocks_count_lo);
2334 }
2335
2336 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2337 ext4_fsblk_t blk)
2338 {
2339 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2340 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2341 }
2342
2343 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2344 ext4_fsblk_t blk)
2345 {
2346 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2347 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2348 }
2349
2350 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2351 ext4_fsblk_t blk)
2352 {
2353 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2354 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2355 }
2356
2357 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2358 {
2359 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2360 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2361 le32_to_cpu(raw_inode->i_size_lo);
2362 else
2363 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2364 }
2365
2366 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2367 {
2368 raw_inode->i_size_lo = cpu_to_le32(i_size);
2369 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2370 }
2371
2372 static inline
2373 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2374 ext4_group_t group)
2375 {
2376 struct ext4_group_info ***grp_info;
2377 long indexv, indexh;
2378 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2379 grp_info = EXT4_SB(sb)->s_group_info;
2380 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2381 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2382 return grp_info[indexv][indexh];
2383 }
2384
2385 /*
2386 * Reading s_groups_count requires using smp_rmb() afterwards. See
2387 * the locking protocol documented in the comments of ext4_group_add()
2388 * in resize.c
2389 */
2390 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2391 {
2392 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2393
2394 smp_rmb();
2395 return ngroups;
2396 }
2397
2398 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2399 ext4_group_t block_group)
2400 {
2401 return block_group >> sbi->s_log_groups_per_flex;
2402 }
2403
2404 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2405 {
2406 return 1 << sbi->s_log_groups_per_flex;
2407 }
2408
2409 #define ext4_std_error(sb, errno) \
2410 do { \
2411 if ((errno)) \
2412 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2413 } while (0)
2414
2415 #ifdef CONFIG_SMP
2416 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2417 * counters. So we need to make sure we have free clusters more
2418 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2419 */
2420 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2421 #else
2422 #define EXT4_FREECLUSTERS_WATERMARK 0
2423 #endif
2424
2425 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2426 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2427 {
2428 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2429 !mutex_is_locked(&inode->i_mutex));
2430 down_write(&EXT4_I(inode)->i_data_sem);
2431 if (newsize > EXT4_I(inode)->i_disksize)
2432 EXT4_I(inode)->i_disksize = newsize;
2433 up_write(&EXT4_I(inode)->i_data_sem);
2434 }
2435
2436 /*
2437 * Update i_disksize after writeback has been started. Races with truncate
2438 * are avoided by checking i_size under i_data_sem.
2439 */
2440 static inline void ext4_wb_update_i_disksize(struct inode *inode, loff_t newsize)
2441 {
2442 loff_t i_size;
2443
2444 down_write(&EXT4_I(inode)->i_data_sem);
2445 i_size = i_size_read(inode);
2446 if (newsize > i_size)
2447 newsize = i_size;
2448 if (newsize > EXT4_I(inode)->i_disksize)
2449 EXT4_I(inode)->i_disksize = newsize;
2450 up_write(&EXT4_I(inode)->i_data_sem);
2451 }
2452
2453 struct ext4_group_info {
2454 unsigned long bb_state;
2455 struct rb_root bb_free_root;
2456 ext4_grpblk_t bb_first_free; /* first free block */
2457 ext4_grpblk_t bb_free; /* total free blocks */
2458 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2459 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2460 struct list_head bb_prealloc_list;
2461 #ifdef DOUBLE_CHECK
2462 void *bb_bitmap;
2463 #endif
2464 struct rw_semaphore alloc_sem;
2465 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2466 * regions, index is order.
2467 * bb_counters[3] = 5 means
2468 * 5 free 8-block regions. */
2469 };
2470
2471 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2472 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2473 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2474 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2475
2476 #define EXT4_MB_GRP_NEED_INIT(grp) \
2477 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2478 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2479 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2480 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2481 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2482
2483 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2484 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2485 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2486 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2487 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2488 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2489
2490 #define EXT4_MAX_CONTENTION 8
2491 #define EXT4_CONTENTION_THRESHOLD 2
2492
2493 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2494 ext4_group_t group)
2495 {
2496 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2497 }
2498
2499 /*
2500 * Returns true if the filesystem is busy enough that attempts to
2501 * access the block group locks has run into contention.
2502 */
2503 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2504 {
2505 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2506 }
2507
2508 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2509 {
2510 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2511 if (spin_trylock(lock))
2512 /*
2513 * We're able to grab the lock right away, so drop the
2514 * lock contention counter.
2515 */
2516 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2517 else {
2518 /*
2519 * The lock is busy, so bump the contention counter,
2520 * and then wait on the spin lock.
2521 */
2522 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2523 EXT4_MAX_CONTENTION);
2524 spin_lock(lock);
2525 }
2526 }
2527
2528 static inline void ext4_unlock_group(struct super_block *sb,
2529 ext4_group_t group)
2530 {
2531 spin_unlock(ext4_group_lock_ptr(sb, group));
2532 }
2533
2534 /*
2535 * Block validity checking
2536 */
2537 #define ext4_check_indirect_blockref(inode, bh) \
2538 ext4_check_blockref(__func__, __LINE__, inode, \
2539 (__le32 *)(bh)->b_data, \
2540 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2541
2542 #define ext4_ind_check_inode(inode) \
2543 ext4_check_blockref(__func__, __LINE__, inode, \
2544 EXT4_I(inode)->i_data, \
2545 EXT4_NDIR_BLOCKS)
2546
2547 /*
2548 * Inodes and files operations
2549 */
2550
2551 /* dir.c */
2552 extern const struct file_operations ext4_dir_operations;
2553
2554 /* file.c */
2555 extern const struct inode_operations ext4_file_inode_operations;
2556 extern const struct file_operations ext4_file_operations;
2557 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2558 extern void ext4_unwritten_wait(struct inode *inode);
2559
2560 /* inline.c */
2561 extern int ext4_has_inline_data(struct inode *inode);
2562 extern int ext4_get_inline_size(struct inode *inode);
2563 extern int ext4_get_max_inline_size(struct inode *inode);
2564 extern int ext4_find_inline_data_nolock(struct inode *inode);
2565 extern void ext4_write_inline_data(struct inode *inode,
2566 struct ext4_iloc *iloc,
2567 void *buffer, loff_t pos,
2568 unsigned int len);
2569 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2570 unsigned int len);
2571 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2572 unsigned int len);
2573 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2574
2575 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2576 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2577 struct inode *inode,
2578 loff_t pos, unsigned len,
2579 unsigned flags,
2580 struct page **pagep);
2581 extern int ext4_write_inline_data_end(struct inode *inode,
2582 loff_t pos, unsigned len,
2583 unsigned copied,
2584 struct page *page);
2585 extern struct buffer_head *
2586 ext4_journalled_write_inline_data(struct inode *inode,
2587 unsigned len,
2588 struct page *page);
2589 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2590 struct inode *inode,
2591 loff_t pos, unsigned len,
2592 unsigned flags,
2593 struct page **pagep,
2594 void **fsdata);
2595 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2596 unsigned len, unsigned copied,
2597 struct page *page);
2598 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2599 struct inode *inode);
2600 extern int ext4_try_create_inline_dir(handle_t *handle,
2601 struct inode *parent,
2602 struct inode *inode);
2603 extern int ext4_read_inline_dir(struct file *filp,
2604 struct dir_context *ctx,
2605 int *has_inline_data);
2606 extern int htree_inlinedir_to_tree(struct file *dir_file,
2607 struct inode *dir, ext4_lblk_t block,
2608 struct dx_hash_info *hinfo,
2609 __u32 start_hash, __u32 start_minor_hash,
2610 int *has_inline_data);
2611 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2612 const struct qstr *d_name,
2613 struct ext4_dir_entry_2 **res_dir,
2614 int *has_inline_data);
2615 extern int ext4_delete_inline_entry(handle_t *handle,
2616 struct inode *dir,
2617 struct ext4_dir_entry_2 *de_del,
2618 struct buffer_head *bh,
2619 int *has_inline_data);
2620 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2621 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2622 struct ext4_dir_entry_2 **parent_de,
2623 int *retval);
2624 extern int ext4_inline_data_fiemap(struct inode *inode,
2625 struct fiemap_extent_info *fieinfo,
2626 int *has_inline);
2627 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2628 struct inode *inode,
2629 int needed);
2630 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2631
2632 extern int ext4_convert_inline_data(struct inode *inode);
2633
2634 /* namei.c */
2635 extern const struct inode_operations ext4_dir_inode_operations;
2636 extern const struct inode_operations ext4_special_inode_operations;
2637 extern struct dentry *ext4_get_parent(struct dentry *child);
2638 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2639 struct ext4_dir_entry_2 *de,
2640 int blocksize, int csum_size,
2641 unsigned int parent_ino, int dotdot_real_len);
2642 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2643 unsigned int blocksize);
2644 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2645 struct inode *inode,
2646 struct buffer_head *bh);
2647 #define S_SHIFT 12
2648 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2649 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2650 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2651 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2652 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2653 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2654 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2655 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2656 };
2657
2658 static inline void ext4_set_de_type(struct super_block *sb,
2659 struct ext4_dir_entry_2 *de,
2660 umode_t mode) {
2661 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2662 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2663 }
2664
2665
2666 /* symlink.c */
2667 extern const struct inode_operations ext4_symlink_inode_operations;
2668 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2669
2670 /* block_validity */
2671 extern void ext4_release_system_zone(struct super_block *sb);
2672 extern int ext4_setup_system_zone(struct super_block *sb);
2673 extern int __init ext4_init_system_zone(void);
2674 extern void ext4_exit_system_zone(void);
2675 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2676 ext4_fsblk_t start_blk,
2677 unsigned int count);
2678 extern int ext4_check_blockref(const char *, unsigned int,
2679 struct inode *, __le32 *, unsigned int);
2680
2681 /* extents.c */
2682 struct ext4_ext_path;
2683 struct ext4_extent;
2684
2685 /*
2686 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2687 * __le32.
2688 */
2689 #define EXT_MAX_BLOCKS 0xffffffff
2690
2691 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2692 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2693 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2694 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2695 struct ext4_map_blocks *map, int flags);
2696 extern void ext4_ext_truncate(handle_t *, struct inode *);
2697 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2698 ext4_lblk_t end);
2699 extern void ext4_ext_init(struct super_block *);
2700 extern void ext4_ext_release(struct super_block *);
2701 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2702 loff_t len);
2703 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2704 loff_t offset, ssize_t len);
2705 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2706 struct ext4_map_blocks *map, int flags);
2707 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2708 ext4_lblk_t lblocks);
2709 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2710 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2711 int num,
2712 struct ext4_ext_path *path);
2713 extern int ext4_can_extents_be_merged(struct inode *inode,
2714 struct ext4_extent *ex1,
2715 struct ext4_extent *ex2);
2716 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2717 struct ext4_ext_path *,
2718 struct ext4_extent *, int);
2719 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2720 struct ext4_ext_path *,
2721 int flags);
2722 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2723 extern int ext4_ext_check_inode(struct inode *inode);
2724 extern int ext4_find_delalloc_range(struct inode *inode,
2725 ext4_lblk_t lblk_start,
2726 ext4_lblk_t lblk_end);
2727 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2728 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2729 __u64 start, __u64 len);
2730 extern int ext4_ext_precache(struct inode *inode);
2731
2732 /* move_extent.c */
2733 extern void ext4_double_down_write_data_sem(struct inode *first,
2734 struct inode *second);
2735 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2736 struct inode *donor_inode);
2737 void ext4_inode_double_lock(struct inode *inode1, struct inode *inode2);
2738 void ext4_inode_double_unlock(struct inode *inode1, struct inode *inode2);
2739 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2740 __u64 start_orig, __u64 start_donor,
2741 __u64 len, __u64 *moved_len);
2742
2743 /* page-io.c */
2744 extern int __init ext4_init_pageio(void);
2745 extern void ext4_exit_pageio(void);
2746 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2747 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2748 extern int ext4_put_io_end(ext4_io_end_t *io_end);
2749 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2750 extern void ext4_io_submit_init(struct ext4_io_submit *io,
2751 struct writeback_control *wbc);
2752 extern void ext4_end_io_rsv_work(struct work_struct *work);
2753 extern void ext4_io_submit(struct ext4_io_submit *io);
2754 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2755 struct page *page,
2756 int len,
2757 struct writeback_control *wbc);
2758
2759 /* mmp.c */
2760 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2761 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2762 extern int ext4_mmp_csum_verify(struct super_block *sb,
2763 struct mmp_struct *mmp);
2764
2765 /*
2766 * Note that these flags will never ever appear in a buffer_head's state flag.
2767 * See EXT4_MAP_... to see where this is used.
2768 */
2769 enum ext4_state_bits {
2770 BH_Uninit /* blocks are allocated but uninitialized on disk */
2771 = BH_JBDPrivateStart,
2772 BH_AllocFromCluster, /* allocated blocks were part of already
2773 * allocated cluster. */
2774 };
2775
2776 /*
2777 * Add new method to test whether block and inode bitmaps are properly
2778 * initialized. With uninit_bg reading the block from disk is not enough
2779 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2780 */
2781 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2782
2783 static inline int bitmap_uptodate(struct buffer_head *bh)
2784 {
2785 return (buffer_uptodate(bh) &&
2786 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2787 }
2788 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2789 {
2790 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2791 }
2792
2793 /*
2794 * Disable DIO read nolock optimization, so new dioreaders will be forced
2795 * to grab i_mutex
2796 */
2797 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2798 {
2799 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2800 smp_mb();
2801 }
2802 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2803 {
2804 smp_mb();
2805 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2806 }
2807
2808 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2809
2810 /* For ioend & aio unwritten conversion wait queues */
2811 #define EXT4_WQ_HASH_SZ 37
2812 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2813 EXT4_WQ_HASH_SZ])
2814 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2815 EXT4_WQ_HASH_SZ])
2816 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2817 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2818
2819 #define EXT4_RESIZING 0
2820 extern int ext4_resize_begin(struct super_block *sb);
2821 extern void ext4_resize_end(struct super_block *sb);
2822
2823 #endif /* __KERNEL__ */
2824
2825 #endif /* _EXT4_H */
Linus' kernel tree