?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
[linux-2.6.git] / fs / ext4 / ext4.h
blobe6185031c1ccb5d7ef6c35185d77e5b06d648a5d
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 <linux/ratelimit.h>
33 #include <crypto/hash.h>
34 #ifdef __KERNEL__
35 #include <linux/compat.h>
36 #endif
37
38 /*
39 * The fourth extended filesystem constants/structures
40 */
41
42 /*
43 * Define EXT4FS_DEBUG to produce debug messages
44 */
45 #undef EXT4FS_DEBUG
46
47 /*
48 * Debug code
49 */
50 #ifdef EXT4FS_DEBUG
51 #define ext4_debug(f, a...) \
52 do { \
53 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
54 __FILE__, __LINE__, __func__); \
55 printk(KERN_DEBUG f, ## a); \
56 } while (0)
57 #else
58 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
59 #endif
60
61 /*
62 * Turn on EXT_DEBUG to get lots of info about extents operations.
63 */
64 #define EXT_DEBUG__
65 #ifdef EXT_DEBUG
66 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
67 #else
68 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
69 #endif
70
71 #define EXT4_ERROR_INODE(inode, fmt, a...) \
72 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
73
74 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
75 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
76
77 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
78 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
79
80 /* data type for block offset of block group */
81 typedef int ext4_grpblk_t;
82
83 /* data type for filesystem-wide blocks number */
84 typedef unsigned long long ext4_fsblk_t;
85
86 /* data type for file logical block number */
87 typedef __u32 ext4_lblk_t;
88
89 /* data type for block group number */
90 typedef unsigned int ext4_group_t;
91
92 /*
93 * Flags used in mballoc's allocation_context flags field.
94 *
95 * Also used to show what's going on for debugging purposes when the
96 * flag field is exported via the traceport interface
97 */
98
99 /* prefer goal again. length */
100 #define EXT4_MB_HINT_MERGE 0x0001
101 /* blocks already reserved */
102 #define EXT4_MB_HINT_RESERVED 0x0002
103 /* metadata is being allocated */
104 #define EXT4_MB_HINT_METADATA 0x0004
105 /* first blocks in the file */
106 #define EXT4_MB_HINT_FIRST 0x0008
107 /* search for the best chunk */
108 #define EXT4_MB_HINT_BEST 0x0010
109 /* data is being allocated */
110 #define EXT4_MB_HINT_DATA 0x0020
111 /* don't preallocate (for tails) */
112 #define EXT4_MB_HINT_NOPREALLOC 0x0040
113 /* allocate for locality group */
114 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
115 /* allocate goal blocks or none */
116 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
117 /* goal is meaningful */
118 #define EXT4_MB_HINT_TRY_GOAL 0x0200
119 /* blocks already pre-reserved by delayed allocation */
120 #define EXT4_MB_DELALLOC_RESERVED 0x0400
121 /* We are doing stream allocation */
122 #define EXT4_MB_STREAM_ALLOC 0x0800
123 /* Use reserved root blocks if needed */
124 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
125 /* Use blocks from reserved pool */
126 #define EXT4_MB_USE_RESERVED 0x2000
127
128 struct ext4_allocation_request {
129 /* target inode for block we're allocating */
130 struct inode *inode;
131 /* how many blocks we want to allocate */
132 unsigned int len;
133 /* logical block in target inode */
134 ext4_lblk_t logical;
135 /* the closest logical allocated block to the left */
136 ext4_lblk_t lleft;
137 /* the closest logical allocated block to the right */
138 ext4_lblk_t lright;
139 /* phys. target (a hint) */
140 ext4_fsblk_t goal;
141 /* phys. block for the closest logical allocated block to the left */
142 ext4_fsblk_t pleft;
143 /* phys. block for the closest logical allocated block to the right */
144 ext4_fsblk_t pright;
145 /* flags. see above EXT4_MB_HINT_* */
146 unsigned int flags;
147 };
148
149 /*
150 * Logical to physical block mapping, used by ext4_map_blocks()
151 *
152 * This structure is used to pass requests into ext4_map_blocks() as
153 * well as to store the information returned by ext4_map_blocks(). It
154 * takes less room on the stack than a struct buffer_head.
155 */
156 #define EXT4_MAP_NEW (1 << BH_New)
157 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
158 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
159 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
160 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
161 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
162 * ext4_map_blocks wants to know whether or not the underlying cluster has
163 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
164 * the requested mapping was from previously mapped (or delayed allocated)
165 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
166 * should never appear on buffer_head's state flags.
167 */
168 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
169 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
170 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
171 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
172
173 struct ext4_map_blocks {
174 ext4_fsblk_t m_pblk;
175 ext4_lblk_t m_lblk;
176 unsigned int m_len;
177 unsigned int m_flags;
178 };
179
180 /*
181 * Flags for ext4_io_end->flags
182 */
183 #define EXT4_IO_END_UNWRITTEN 0x0001
184
185 /*
186 * For converting uninitialized extents on a work queue. 'handle' is used for
187 * buffered writeback.
188 */
189 typedef struct ext4_io_end {
190 struct list_head list; /* per-file finished IO list */
191 handle_t *handle; /* handle reserved for extent
192 * conversion */
193 struct inode *inode; /* file being written to */
194 struct bio *bio; /* Linked list of completed
195 * bios covering the extent */
196 unsigned int flag; /* unwritten or not */
197 loff_t offset; /* offset in the file */
198 ssize_t size; /* size of the extent */
199 atomic_t count; /* reference counter */
200 } ext4_io_end_t;
201
202 struct ext4_io_submit {
203 int io_op;
204 struct bio *io_bio;
205 ext4_io_end_t *io_end;
206 sector_t io_next_block;
207 };
208
209 /*
210 * Special inodes numbers
211 */
212 #define EXT4_BAD_INO 1 /* Bad blocks inode */
213 #define EXT4_ROOT_INO 2 /* Root inode */
214 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
218 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO 8 /* Journal inode */
220
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
223
224 /*
225 * Maximal count of links to a file
226 */
227 #define EXT4_LINK_MAX 65000
228
229 /*
230 * Macro-instructions used to manage several block sizes
231 */
232 #define EXT4_MIN_BLOCK_SIZE 1024
233 #define EXT4_MAX_BLOCK_SIZE 65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE 10
235 #define EXT4_MAX_BLOCK_LOG_SIZE 16
236 #ifdef __KERNEL__
237 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
238 #else
239 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
240 #endif
241 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
243 EXT4_SB(s)->s_cluster_bits)
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
247 #else
248 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
249 #endif
250 #ifdef __KERNEL__
251 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
254 #else
255 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 EXT4_GOOD_OLD_INODE_SIZE : \
257 (s)->s_inode_size)
258 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259 EXT4_GOOD_OLD_FIRST_INO : \
260 (s)->s_first_ino)
261 #endif
262 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
263
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270 (sbi)->s_cluster_bits)
271
272 /*
273 * Structure of a blocks group descriptor
274 */
275 struct ext4_group_desc
276 {
277 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
278 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
279 __le32 bg_inode_table_lo; /* Inodes table block */
280 __le16 bg_free_blocks_count_lo;/* Free blocks count */
281 __le16 bg_free_inodes_count_lo;/* Free inodes count */
282 __le16 bg_used_dirs_count_lo; /* Directories count */
283 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
284 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
285 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
286 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
287 __le16 bg_itable_unused_lo; /* Unused inodes count */
288 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
289 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
290 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
291 __le32 bg_inode_table_hi; /* Inodes table block MSB */
292 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
293 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
294 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
295 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
296 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
297 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
298 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
299 __u32 bg_reserved;
300 };
301
302 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
303 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
304 sizeof(__le16))
305 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
306 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
307 sizeof(__le16))
308
309 /*
310 * Structure of a flex block group info
311 */
312
313 struct flex_groups {
314 atomic64_t free_clusters;
315 atomic_t free_inodes;
316 atomic_t used_dirs;
317 };
318
319 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
320 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
321 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
322
323 /*
324 * Macro-instructions used to manage group descriptors
325 */
326 #define EXT4_MIN_DESC_SIZE 32
327 #define EXT4_MIN_DESC_SIZE_64BIT 64
328 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
329 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
330 #ifdef __KERNEL__
331 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
332 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
333 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
334 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
335 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
336 #else
337 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
338 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
339 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
340 #endif
341
342 /*
343 * Constants relative to the data blocks
344 */
345 #define EXT4_NDIR_BLOCKS 12
346 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
347 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
348 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
349 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
350
351 /*
352 * Inode flags
353 */
354 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
355 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
356 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
357 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
358 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
359 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
360 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
361 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
362 /* Reserved for compression usage... */
363 #define EXT4_DIRTY_FL 0x00000100
364 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
365 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
366 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
367 /* End compression flags --- maybe not all used */
368 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
369 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
370 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
371 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
372 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
373 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
374 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
375 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
376 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
377 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
378 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
379 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
380
381 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
382 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
383
384 /* Flags that should be inherited by new inodes from their parent. */
385 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
386 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
387 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
388 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
389
390 /* Flags that are appropriate for regular files (all but dir-specific ones). */
391 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
392
393 /* Flags that are appropriate for non-directories/regular files. */
394 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
395
396 /* Mask out flags that are inappropriate for the given type of inode. */
397 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
398 {
399 if (S_ISDIR(mode))
400 return flags;
401 else if (S_ISREG(mode))
402 return flags & EXT4_REG_FLMASK;
403 else
404 return flags & EXT4_OTHER_FLMASK;
405 }
406
407 /*
408 * Inode flags used for atomic set/get
409 */
410 enum {
411 EXT4_INODE_SECRM = 0, /* Secure deletion */
412 EXT4_INODE_UNRM = 1, /* Undelete */
413 EXT4_INODE_COMPR = 2, /* Compress file */
414 EXT4_INODE_SYNC = 3, /* Synchronous updates */
415 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
416 EXT4_INODE_APPEND = 5, /* writes to file may only append */
417 EXT4_INODE_NODUMP = 6, /* do not dump file */
418 EXT4_INODE_NOATIME = 7, /* do not update atime */
419 /* Reserved for compression usage... */
420 EXT4_INODE_DIRTY = 8,
421 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
422 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
423 EXT4_INODE_ECOMPR = 11, /* Compression error */
424 /* End compression flags --- maybe not all used */
425 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
426 EXT4_INODE_IMAGIC = 13, /* AFS directory */
427 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
428 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
429 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
430 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
431 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
432 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
433 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
434 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
435 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
436 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
437 };
438
439 /*
440 * Since it's pretty easy to mix up bit numbers and hex values, we use a
441 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
442 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
443 * any extra space in the compiled kernel image, otherwise, the build will fail.
444 * It's important that these values are the same, since we are using
445 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
446 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
447 * values found in ext2, ext3 and ext4 filesystems, and of course the values
448 * defined in e2fsprogs.
449 *
450 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
451 */
452 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
453 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
454
455 static inline void ext4_check_flag_values(void)
456 {
457 CHECK_FLAG_VALUE(SECRM);
458 CHECK_FLAG_VALUE(UNRM);
459 CHECK_FLAG_VALUE(COMPR);
460 CHECK_FLAG_VALUE(SYNC);
461 CHECK_FLAG_VALUE(IMMUTABLE);
462 CHECK_FLAG_VALUE(APPEND);
463 CHECK_FLAG_VALUE(NODUMP);
464 CHECK_FLAG_VALUE(NOATIME);
465 CHECK_FLAG_VALUE(DIRTY);
466 CHECK_FLAG_VALUE(COMPRBLK);
467 CHECK_FLAG_VALUE(NOCOMPR);
468 CHECK_FLAG_VALUE(ECOMPR);
469 CHECK_FLAG_VALUE(INDEX);
470 CHECK_FLAG_VALUE(IMAGIC);
471 CHECK_FLAG_VALUE(JOURNAL_DATA);
472 CHECK_FLAG_VALUE(NOTAIL);
473 CHECK_FLAG_VALUE(DIRSYNC);
474 CHECK_FLAG_VALUE(TOPDIR);
475 CHECK_FLAG_VALUE(HUGE_FILE);
476 CHECK_FLAG_VALUE(EXTENTS);
477 CHECK_FLAG_VALUE(EA_INODE);
478 CHECK_FLAG_VALUE(EOFBLOCKS);
479 CHECK_FLAG_VALUE(INLINE_DATA);
480 CHECK_FLAG_VALUE(RESERVED);
481 }
482
483 /* Used to pass group descriptor data when online resize is done */
484 struct ext4_new_group_input {
485 __u32 group; /* Group number for this data */
486 __u64 block_bitmap; /* Absolute block number of block bitmap */
487 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
488 __u64 inode_table; /* Absolute block number of inode table start */
489 __u32 blocks_count; /* Total number of blocks in this group */
490 __u16 reserved_blocks; /* Number of reserved blocks in this group */
491 __u16 unused;
492 };
493
494 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
495 struct compat_ext4_new_group_input {
496 u32 group;
497 compat_u64 block_bitmap;
498 compat_u64 inode_bitmap;
499 compat_u64 inode_table;
500 u32 blocks_count;
501 u16 reserved_blocks;
502 u16 unused;
503 };
504 #endif
505
506 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
507 struct ext4_new_group_data {
508 __u32 group;
509 __u64 block_bitmap;
510 __u64 inode_bitmap;
511 __u64 inode_table;
512 __u32 blocks_count;
513 __u16 reserved_blocks;
514 __u16 unused;
515 __u32 free_blocks_count;
516 };
517
518 /* Indexes used to index group tables in ext4_new_group_data */
519 enum {
520 BLOCK_BITMAP = 0, /* block bitmap */
521 INODE_BITMAP, /* inode bitmap */
522 INODE_TABLE, /* inode tables */
523 GROUP_TABLE_COUNT,
524 };
525
526 /*
527 * Flags used by ext4_map_blocks()
528 */
529 /* Allocate any needed blocks and/or convert an unitialized
530 extent to be an initialized ext4 */
531 #define EXT4_GET_BLOCKS_CREATE 0x0001
532 /* Request the creation of an unitialized extent */
533 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
534 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
535 EXT4_GET_BLOCKS_CREATE)
536 /* Caller is from the delayed allocation writeout path
537 * finally doing the actual allocation of delayed blocks */
538 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
539 /* caller is from the direct IO path, request to creation of an
540 unitialized extents if not allocated, split the uninitialized
541 extent if blocks has been preallocated already*/
542 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
543 #define EXT4_GET_BLOCKS_CONVERT 0x0010
544 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
545 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
546 /* Convert extent to initialized after IO complete */
547 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
548 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
549 /* Eventual metadata allocation (due to growing extent tree)
550 * should not fail, so try to use reserved blocks for that.*/
551 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
552 /* Don't normalize allocation size (used for fallocate) */
553 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
554 /* Request will not result in inode size update (user for fallocate) */
555 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
556 /* Do not take i_data_sem locking in ext4_map_blocks */
557 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
558 /* Do not put hole in extent cache */
559 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
560
561 /*
562 * The bit position of these flags must not overlap with any of the
563 * EXT4_GET_BLOCKS_*. They are used by ext4_ext_find_extent(),
564 * read_extent_tree_block(), ext4_split_extent_at(),
565 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
566 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
567 * caching the extents when reading from the extent tree while a
568 * truncate or punch hole operation is in progress.
569 */
570 #define EXT4_EX_NOCACHE 0x0400
571 #define EXT4_EX_FORCE_CACHE 0x0800
572
573 /*
574 * Flags used by ext4_free_blocks
575 */
576 #define EXT4_FREE_BLOCKS_METADATA 0x0001
577 #define EXT4_FREE_BLOCKS_FORGET 0x0002
578 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
579 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
580 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
581 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
582 #define EXT4_FREE_BLOCKS_RESERVE 0x0040
583
584 /*
585 * ioctl commands
586 */
587 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
588 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
589 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
590 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
591 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
592 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
593 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
594 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
595 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
596 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
597 #define EXT4_IOC_MIGRATE _IO('f', 9)
598 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
599 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
600 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
601 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
602 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
603 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
604 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
605
606 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
607 /*
608 * ioctl commands in 32 bit emulation
609 */
610 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
611 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
612 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
613 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
614 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
615 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
616 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
617 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
618 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
619 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
620 #endif
621
622 /* Max physical block we can address w/o extents */
623 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
624
625 /*
626 * Structure of an inode on the disk
627 */
628 struct ext4_inode {
629 __le16 i_mode; /* File mode */
630 __le16 i_uid; /* Low 16 bits of Owner Uid */
631 __le32 i_size_lo; /* Size in bytes */
632 __le32 i_atime; /* Access time */
633 __le32 i_ctime; /* Inode Change time */
634 __le32 i_mtime; /* Modification time */
635 __le32 i_dtime; /* Deletion Time */
636 __le16 i_gid; /* Low 16 bits of Group Id */
637 __le16 i_links_count; /* Links count */
638 __le32 i_blocks_lo; /* Blocks count */
639 __le32 i_flags; /* File flags */
640 union {
641 struct {
642 __le32 l_i_version;
643 } linux1;
644 struct {
645 __u32 h_i_translator;
646 } hurd1;
647 struct {
648 __u32 m_i_reserved1;
649 } masix1;
650 } osd1; /* OS dependent 1 */
651 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
652 __le32 i_generation; /* File version (for NFS) */
653 __le32 i_file_acl_lo; /* File ACL */
654 __le32 i_size_high;
655 __le32 i_obso_faddr; /* Obsoleted fragment address */
656 union {
657 struct {
658 __le16 l_i_blocks_high; /* were l_i_reserved1 */
659 __le16 l_i_file_acl_high;
660 __le16 l_i_uid_high; /* these 2 fields */
661 __le16 l_i_gid_high; /* were reserved2[0] */
662 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
663 __le16 l_i_reserved;
664 } linux2;
665 struct {
666 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
667 __u16 h_i_mode_high;
668 __u16 h_i_uid_high;
669 __u16 h_i_gid_high;
670 __u32 h_i_author;
671 } hurd2;
672 struct {
673 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
674 __le16 m_i_file_acl_high;
675 __u32 m_i_reserved2[2];
676 } masix2;
677 } osd2; /* OS dependent 2 */
678 __le16 i_extra_isize;
679 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
680 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
681 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
682 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
683 __le32 i_crtime; /* File Creation time */
684 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
685 __le32 i_version_hi; /* high 32 bits for 64-bit version */
686 };
687
688 struct move_extent {
689 __u32 reserved; /* should be zero */
690 __u32 donor_fd; /* donor file descriptor */
691 __u64 orig_start; /* logical start offset in block for orig */
692 __u64 donor_start; /* logical start offset in block for donor */
693 __u64 len; /* block length to be moved */
694 __u64 moved_len; /* moved block length */
695 };
696
697 #define EXT4_EPOCH_BITS 2
698 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
699 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
700
701 /*
702 * Extended fields will fit into an inode if the filesystem was formatted
703 * with large inodes (-I 256 or larger) and there are not currently any EAs
704 * consuming all of the available space. For new inodes we always reserve
705 * enough space for the kernel's known extended fields, but for inodes
706 * created with an old kernel this might not have been the case. None of
707 * the extended inode fields is critical for correct filesystem operation.
708 * This macro checks if a certain field fits in the inode. Note that
709 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
710 */
711 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
712 ((offsetof(typeof(*ext4_inode), field) + \
713 sizeof((ext4_inode)->field)) \
714 <= (EXT4_GOOD_OLD_INODE_SIZE + \
715 (einode)->i_extra_isize)) \
716
717 static inline __le32 ext4_encode_extra_time(struct timespec *time)
718 {
719 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
720 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
721 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
722 }
723
724 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
725 {
726 if (sizeof(time->tv_sec) > 4)
727 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
728 << 32;
729 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
730 }
731
732 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
733 do { \
734 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
735 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
736 (raw_inode)->xtime ## _extra = \
737 ext4_encode_extra_time(&(inode)->xtime); \
738 } while (0)
739
740 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
741 do { \
742 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
743 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
744 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
745 (raw_inode)->xtime ## _extra = \
746 ext4_encode_extra_time(&(einode)->xtime); \
747 } while (0)
748
749 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
750 do { \
751 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
752 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
753 ext4_decode_extra_time(&(inode)->xtime, \
754 raw_inode->xtime ## _extra); \
755 else \
756 (inode)->xtime.tv_nsec = 0; \
757 } while (0)
758
759 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
760 do { \
761 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
762 (einode)->xtime.tv_sec = \
763 (signed)le32_to_cpu((raw_inode)->xtime); \
764 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
765 ext4_decode_extra_time(&(einode)->xtime, \
766 raw_inode->xtime ## _extra); \
767 else \
768 (einode)->xtime.tv_nsec = 0; \
769 } while (0)
770
771 #define i_disk_version osd1.linux1.l_i_version
772
773 #if defined(__KERNEL__) || defined(__linux__)
774 #define i_reserved1 osd1.linux1.l_i_reserved1
775 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
776 #define i_blocks_high osd2.linux2.l_i_blocks_high
777 #define i_uid_low i_uid
778 #define i_gid_low i_gid
779 #define i_uid_high osd2.linux2.l_i_uid_high
780 #define i_gid_high osd2.linux2.l_i_gid_high
781 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
782
783 #elif defined(__GNU__)
784
785 #define i_translator osd1.hurd1.h_i_translator
786 #define i_uid_high osd2.hurd2.h_i_uid_high
787 #define i_gid_high osd2.hurd2.h_i_gid_high
788 #define i_author osd2.hurd2.h_i_author
789
790 #elif defined(__masix__)
791
792 #define i_reserved1 osd1.masix1.m_i_reserved1
793 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
794 #define i_reserved2 osd2.masix2.m_i_reserved2
795
796 #endif /* defined(__KERNEL__) || defined(__linux__) */
797
798 #include "extents_status.h"
799
800 /*
801 * fourth extended file system inode data in memory
802 */
803 struct ext4_inode_info {
804 __le32 i_data[15]; /* unconverted */
805 __u32 i_dtime;
806 ext4_fsblk_t i_file_acl;
807
808 /*
809 * i_block_group is the number of the block group which contains
810 * this file's inode. Constant across the lifetime of the inode,
811 * it is ued for making block allocation decisions - we try to
812 * place a file's data blocks near its inode block, and new inodes
813 * near to their parent directory's inode.
814 */
815 ext4_group_t i_block_group;
816 ext4_lblk_t i_dir_start_lookup;
817 #if (BITS_PER_LONG < 64)
818 unsigned long i_state_flags; /* Dynamic state flags */
819 #endif
820 unsigned long i_flags;
821
822 /*
823 * Extended attributes can be read independently of the main file
824 * data. Taking i_mutex even when reading would cause contention
825 * between readers of EAs and writers of regular file data, so
826 * instead we synchronize on xattr_sem when reading or changing
827 * EAs.
828 */
829 struct rw_semaphore xattr_sem;
830
831 struct list_head i_orphan; /* unlinked but open inodes */
832
833 /*
834 * i_disksize keeps track of what the inode size is ON DISK, not
835 * in memory. During truncate, i_size is set to the new size by
836 * the VFS prior to calling ext4_truncate(), but the filesystem won't
837 * set i_disksize to 0 until the truncate is actually under way.
838 *
839 * The intent is that i_disksize always represents the blocks which
840 * are used by this file. This allows recovery to restart truncate
841 * on orphans if we crash during truncate. We actually write i_disksize
842 * into the on-disk inode when writing inodes out, instead of i_size.
843 *
844 * The only time when i_disksize and i_size may be different is when
845 * a truncate is in progress. The only things which change i_disksize
846 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
847 */
848 loff_t i_disksize;
849
850 /*
851 * i_data_sem is for serialising ext4_truncate() against
852 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
853 * data tree are chopped off during truncate. We can't do that in
854 * ext4 because whenever we perform intermediate commits during
855 * truncate, the inode and all the metadata blocks *must* be in a
856 * consistent state which allows truncation of the orphans to restart
857 * during recovery. Hence we must fix the get_block-vs-truncate race
858 * by other means, so we have i_data_sem.
859 */
860 struct rw_semaphore i_data_sem;
861 struct inode vfs_inode;
862 struct jbd2_inode *jinode;
863
864 /*
865 * File creation time. Its function is same as that of
866 * struct timespec i_{a,c,m}time in the generic inode.
867 */
868 struct timespec i_crtime;
869
870 /* mballoc */
871 struct list_head i_prealloc_list;
872 spinlock_t i_prealloc_lock;
873
874 /* extents status tree */
875 struct ext4_es_tree i_es_tree;
876 rwlock_t i_es_lock;
877 struct list_head i_es_lru;
878 unsigned int i_es_lru_nr; /* protected by i_es_lock */
879 unsigned long i_touch_when; /* jiffies of last accessing */
880
881 /* ialloc */
882 ext4_group_t i_last_alloc_group;
883
884 /* allocation reservation info for delalloc */
885 /* In case of bigalloc, these refer to clusters rather than blocks */
886 unsigned int i_reserved_data_blocks;
887 unsigned int i_reserved_meta_blocks;
888 unsigned int i_allocated_meta_blocks;
889 ext4_lblk_t i_da_metadata_calc_last_lblock;
890 int i_da_metadata_calc_len;
891
892 /* on-disk additional length */
893 __u16 i_extra_isize;
894
895 /* Indicate the inline data space. */
896 u16 i_inline_off;
897 u16 i_inline_size;
898
899 #ifdef CONFIG_QUOTA
900 /* quota space reservation, managed internally by quota code */
901 qsize_t i_reserved_quota;
902 #endif
903
904 /* Lock protecting lists below */
905 spinlock_t i_completed_io_lock;
906 /*
907 * Completed IOs that need unwritten extents handling and have
908 * transaction reserved
909 */
910 struct list_head i_rsv_conversion_list;
911 /*
912 * Completed IOs that need unwritten extents handling and don't have
913 * transaction reserved
914 */
915 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
916 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
917 struct work_struct i_rsv_conversion_work;
918
919 spinlock_t i_block_reservation_lock;
920
921 /*
922 * Transactions that contain inode's metadata needed to complete
923 * fsync and fdatasync, respectively.
924 */
925 tid_t i_sync_tid;
926 tid_t i_datasync_tid;
927
928 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
929 __u32 i_csum_seed;
930 };
931
932 /*
933 * File system states
934 */
935 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
936 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
937 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
938
939 /*
940 * Misc. filesystem flags
941 */
942 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
943 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
944 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
945
946 /*
947 * Mount flags set via mount options or defaults
948 */
949 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
950 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
951 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
952 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
953 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
954 #define EXT4_MOUNT_ERRORS_MASK 0x00070
955 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
956 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
957 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
958 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
959 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
960 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
961 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
962 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
963 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
964 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
965 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
966 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
967 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
968 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
969 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
970 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
971 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
972 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
973 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
974 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
975 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
976 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
977 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
978
979 /*
980 * Mount flags set either automatically (could not be set by mount option)
981 * based on per file system feature or property or in special cases such as
982 * distinguishing between explicit mount option definition and default.
983 */
984 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
985 specified delalloc */
986 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
987 size of blocksize * 8
988 blocks */
989
990 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
991 ~EXT4_MOUNT_##opt
992 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
993 EXT4_MOUNT_##opt
994 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
995 EXT4_MOUNT_##opt)
996
997 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
998 ~EXT4_MOUNT2_##opt
999 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1000 EXT4_MOUNT2_##opt
1001 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1002 EXT4_MOUNT2_##opt)
1003
1004 #define ext4_test_and_set_bit __test_and_set_bit_le
1005 #define ext4_set_bit __set_bit_le
1006 #define ext4_set_bit_atomic ext2_set_bit_atomic
1007 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1008 #define ext4_clear_bit __clear_bit_le
1009 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1010 #define ext4_test_bit test_bit_le
1011 #define ext4_find_next_zero_bit find_next_zero_bit_le
1012 #define ext4_find_next_bit find_next_bit_le
1013
1014 extern void ext4_set_bits(void *bm, int cur, int len);
1015
1016 /*
1017 * Maximal mount counts between two filesystem checks
1018 */
1019 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1020 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1021
1022 /*
1023 * Behaviour when detecting errors
1024 */
1025 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1026 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1027 #define EXT4_ERRORS_PANIC 3 /* Panic */
1028 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1029
1030 /* Metadata checksum algorithm codes */
1031 #define EXT4_CRC32C_CHKSUM 1
1032
1033 /*
1034 * Structure of the super block
1035 */
1036 struct ext4_super_block {
1037 /*00*/ __le32 s_inodes_count; /* Inodes count */
1038 __le32 s_blocks_count_lo; /* Blocks count */
1039 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1040 __le32 s_free_blocks_count_lo; /* Free blocks count */
1041 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1042 __le32 s_first_data_block; /* First Data Block */
1043 __le32 s_log_block_size; /* Block size */
1044 __le32 s_log_cluster_size; /* Allocation cluster size */
1045 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1046 __le32 s_clusters_per_group; /* # Clusters per group */
1047 __le32 s_inodes_per_group; /* # Inodes per group */
1048 __le32 s_mtime; /* Mount time */
1049 /*30*/ __le32 s_wtime; /* Write time */
1050 __le16 s_mnt_count; /* Mount count */
1051 __le16 s_max_mnt_count; /* Maximal mount count */
1052 __le16 s_magic; /* Magic signature */
1053 __le16 s_state; /* File system state */
1054 __le16 s_errors; /* Behaviour when detecting errors */
1055 __le16 s_minor_rev_level; /* minor revision level */
1056 /*40*/ __le32 s_lastcheck; /* time of last check */
1057 __le32 s_checkinterval; /* max. time between checks */
1058 __le32 s_creator_os; /* OS */
1059 __le32 s_rev_level; /* Revision level */
1060 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1061 __le16 s_def_resgid; /* Default gid for reserved blocks */
1062 /*
1063 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1064 *
1065 * Note: the difference between the compatible feature set and
1066 * the incompatible feature set is that if there is a bit set
1067 * in the incompatible feature set that the kernel doesn't
1068 * know about, it should refuse to mount the filesystem.
1069 *
1070 * e2fsck's requirements are more strict; if it doesn't know
1071 * about a feature in either the compatible or incompatible
1072 * feature set, it must abort and not try to meddle with
1073 * things it doesn't understand...
1074 */
1075 __le32 s_first_ino; /* First non-reserved inode */
1076 __le16 s_inode_size; /* size of inode structure */
1077 __le16 s_block_group_nr; /* block group # of this superblock */
1078 __le32 s_feature_compat; /* compatible feature set */
1079 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1080 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1081 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1082 /*78*/ char s_volume_name[16]; /* volume name */
1083 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1084 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1085 /*
1086 * Performance hints. Directory preallocation should only
1087 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1088 */
1089 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1090 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1091 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1092 /*
1093 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1094 */
1095 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1096 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1097 __le32 s_journal_dev; /* device number of journal file */
1098 __le32 s_last_orphan; /* start of list of inodes to delete */
1099 __le32 s_hash_seed[4]; /* HTREE hash seed */
1100 __u8 s_def_hash_version; /* Default hash version to use */
1101 __u8 s_jnl_backup_type;
1102 __le16 s_desc_size; /* size of group descriptor */
1103 /*100*/ __le32 s_default_mount_opts;
1104 __le32 s_first_meta_bg; /* First metablock block group */
1105 __le32 s_mkfs_time; /* When the filesystem was created */
1106 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1107 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1108 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1109 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1110 __le32 s_free_blocks_count_hi; /* Free blocks count */
1111 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1112 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1113 __le32 s_flags; /* Miscellaneous flags */
1114 __le16 s_raid_stride; /* RAID stride */
1115 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1116 __le64 s_mmp_block; /* Block for multi-mount protection */
1117 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1118 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1119 __u8 s_checksum_type; /* metadata checksum algorithm used */
1120 __le16 s_reserved_pad;
1121 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1122 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1123 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1124 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1125 snapshot's future use */
1126 __le32 s_snapshot_list; /* inode number of the head of the
1127 on-disk snapshot list */
1128 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1129 __le32 s_error_count; /* number of fs errors */
1130 __le32 s_first_error_time; /* first time an error happened */
1131 __le32 s_first_error_ino; /* inode involved in first error */
1132 __le64 s_first_error_block; /* block involved of first error */
1133 __u8 s_first_error_func[32]; /* function where the error happened */
1134 __le32 s_first_error_line; /* line number where error happened */
1135 __le32 s_last_error_time; /* most recent time of an error */
1136 __le32 s_last_error_ino; /* inode involved in last error */
1137 __le32 s_last_error_line; /* line number where error happened */
1138 __le64 s_last_error_block; /* block involved of last error */
1139 __u8 s_last_error_func[32]; /* function where the error happened */
1140 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1141 __u8 s_mount_opts[64];
1142 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1143 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1144 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1145 __le32 s_reserved[108]; /* Padding to the end of the block */
1146 __le32 s_checksum; /* crc32c(superblock) */
1147 };
1148
1149 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1150
1151 #ifdef __KERNEL__
1152
1153 /*
1154 * run-time mount flags
1155 */
1156 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1157 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1158
1159 /*
1160 * fourth extended-fs super-block data in memory
1161 */
1162 struct ext4_sb_info {
1163 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1164 unsigned long s_inodes_per_block;/* Number of inodes per block */
1165 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1166 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1167 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1168 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1169 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1170 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1171 ext4_group_t s_groups_count; /* Number of groups in the fs */
1172 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1173 unsigned long s_overhead; /* # of fs overhead clusters */
1174 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1175 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1176 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1177 struct buffer_head * s_sbh; /* Buffer containing the super block */
1178 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1179 struct buffer_head **s_group_desc;
1180 unsigned int s_mount_opt;
1181 unsigned int s_mount_opt2;
1182 unsigned int s_mount_flags;
1183 unsigned int s_def_mount_opt;
1184 ext4_fsblk_t s_sb_block;
1185 atomic64_t s_resv_clusters;
1186 kuid_t s_resuid;
1187 kgid_t s_resgid;
1188 unsigned short s_mount_state;
1189 unsigned short s_pad;
1190 int s_addr_per_block_bits;
1191 int s_desc_per_block_bits;
1192 int s_inode_size;
1193 int s_first_ino;
1194 unsigned int s_inode_readahead_blks;
1195 unsigned int s_inode_goal;
1196 spinlock_t s_next_gen_lock;
1197 u32 s_next_generation;
1198 u32 s_hash_seed[4];
1199 int s_def_hash_version;
1200 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1201 struct percpu_counter s_freeclusters_counter;
1202 struct percpu_counter s_freeinodes_counter;
1203 struct percpu_counter s_dirs_counter;
1204 struct percpu_counter s_dirtyclusters_counter;
1205 struct blockgroup_lock *s_blockgroup_lock;
1206 struct proc_dir_entry *s_proc;
1207 struct kobject s_kobj;
1208 struct completion s_kobj_unregister;
1209 struct super_block *s_sb;
1210
1211 /* Journaling */
1212 struct journal_s *s_journal;
1213 struct list_head s_orphan;
1214 struct mutex s_orphan_lock;
1215 unsigned long s_resize_flags; /* Flags indicating if there
1216 is a resizer */
1217 unsigned long s_commit_interval;
1218 u32 s_max_batch_time;
1219 u32 s_min_batch_time;
1220 struct block_device *journal_bdev;
1221 #ifdef CONFIG_QUOTA
1222 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1223 int s_jquota_fmt; /* Format of quota to use */
1224 #endif
1225 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1226 struct rb_root system_blks;
1227
1228 #ifdef EXTENTS_STATS
1229 /* ext4 extents stats */
1230 unsigned long s_ext_min;
1231 unsigned long s_ext_max;
1232 unsigned long s_depth_max;
1233 spinlock_t s_ext_stats_lock;
1234 unsigned long s_ext_blocks;
1235 unsigned long s_ext_extents;
1236 #endif
1237
1238 /* for buddy allocator */
1239 struct ext4_group_info ***s_group_info;
1240 struct inode *s_buddy_cache;
1241 spinlock_t s_md_lock;
1242 unsigned short *s_mb_offsets;
1243 unsigned int *s_mb_maxs;
1244 unsigned int s_group_info_size;
1245
1246 /* tunables */
1247 unsigned long s_stripe;
1248 unsigned int s_mb_stream_request;
1249 unsigned int s_mb_max_to_scan;
1250 unsigned int s_mb_min_to_scan;
1251 unsigned int s_mb_stats;
1252 unsigned int s_mb_order2_reqs;
1253 unsigned int s_mb_group_prealloc;
1254 unsigned int s_max_dir_size_kb;
1255 /* where last allocation was done - for stream allocation */
1256 unsigned long s_mb_last_group;
1257 unsigned long s_mb_last_start;
1258
1259 /* stats for buddy allocator */
1260 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1261 atomic_t s_bal_success; /* we found long enough chunks */
1262 atomic_t s_bal_allocated; /* in blocks */
1263 atomic_t s_bal_ex_scanned; /* total extents scanned */
1264 atomic_t s_bal_goals; /* goal hits */
1265 atomic_t s_bal_breaks; /* too long searches */
1266 atomic_t s_bal_2orders; /* 2^order hits */
1267 spinlock_t s_bal_lock;
1268 unsigned long s_mb_buddies_generated;
1269 unsigned long long s_mb_generation_time;
1270 atomic_t s_mb_lost_chunks;
1271 atomic_t s_mb_preallocated;
1272 atomic_t s_mb_discarded;
1273 atomic_t s_lock_busy;
1274
1275 /* locality groups */
1276 struct ext4_locality_group __percpu *s_locality_groups;
1277
1278 /* for write statistics */
1279 unsigned long s_sectors_written_start;
1280 u64 s_kbytes_written;
1281
1282 /* the size of zero-out chunk */
1283 unsigned int s_extent_max_zeroout_kb;
1284
1285 unsigned int s_log_groups_per_flex;
1286 struct flex_groups *s_flex_groups;
1287 ext4_group_t s_flex_groups_allocated;
1288
1289 /* workqueue for reserved extent conversions (buffered io) */
1290 struct workqueue_struct *rsv_conversion_wq;
1291
1292 /* timer for periodic error stats printing */
1293 struct timer_list s_err_report;
1294
1295 /* Lazy inode table initialization info */
1296 struct ext4_li_request *s_li_request;
1297 /* Wait multiplier for lazy initialization thread */
1298 unsigned int s_li_wait_mult;
1299
1300 /* Kernel thread for multiple mount protection */
1301 struct task_struct *s_mmp_tsk;
1302
1303 /* record the last minlen when FITRIM is called. */
1304 atomic_t s_last_trim_minblks;
1305
1306 /* Reference to checksum algorithm driver via cryptoapi */
1307 struct crypto_shash *s_chksum_driver;
1308
1309 /* Precomputed FS UUID checksum for seeding other checksums */
1310 __u32 s_csum_seed;
1311
1312 /* Reclaim extents from extent status tree */
1313 struct shrinker s_es_shrinker;
1314 struct list_head s_es_lru;
1315 unsigned long s_es_last_sorted;
1316 struct percpu_counter s_extent_cache_cnt;
1317 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1318
1319 /* Ratelimit ext4 messages. */
1320 struct ratelimit_state s_err_ratelimit_state;
1321 struct ratelimit_state s_warning_ratelimit_state;
1322 struct ratelimit_state s_msg_ratelimit_state;
1323 };
1324
1325 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1326 {
1327 return sb->s_fs_info;
1328 }
1329 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1330 {
1331 return container_of(inode, struct ext4_inode_info, vfs_inode);
1332 }
1333
1334 static inline struct timespec ext4_current_time(struct inode *inode)
1335 {
1336 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1337 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1338 }
1339
1340 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1341 {
1342 return ino == EXT4_ROOT_INO ||
1343 ino == EXT4_USR_QUOTA_INO ||
1344 ino == EXT4_GRP_QUOTA_INO ||
1345 ino == EXT4_BOOT_LOADER_INO ||
1346 ino == EXT4_JOURNAL_INO ||
1347 ino == EXT4_RESIZE_INO ||
1348 (ino >= EXT4_FIRST_INO(sb) &&
1349 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1350 }
1351
1352 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1353 struct ext4_io_end *io_end)
1354 {
1355 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1356 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1357 atomic_inc(&EXT4_I(inode)->i_unwritten);
1358 }
1359 }
1360
1361 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1362 {
1363 return inode->i_private;
1364 }
1365
1366 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1367 {
1368 inode->i_private = io;
1369 }
1370
1371 /*
1372 * Inode dynamic state flags
1373 */
1374 enum {
1375 EXT4_STATE_JDATA, /* journaled data exists */
1376 EXT4_STATE_NEW, /* inode is newly created */
1377 EXT4_STATE_XATTR, /* has in-inode xattrs */
1378 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1379 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1380 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1381 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1382 EXT4_STATE_NEWENTRY, /* File just added to dir */
1383 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1384 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1385 nolocking */
1386 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1387 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1388 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1389 };
1390
1391 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1392 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1393 { \
1394 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1395 } \
1396 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1397 { \
1398 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1399 } \
1400 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1401 { \
1402 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1403 }
1404
1405 /* Add these declarations here only so that these functions can be
1406 * found by name. Otherwise, they are very hard to locate. */
1407 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1408 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1409 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1410 EXT4_INODE_BIT_FNS(flag, flags, 0)
1411
1412 /* Add these declarations here only so that these functions can be
1413 * found by name. Otherwise, they are very hard to locate. */
1414 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1415 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1416 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1417 #if (BITS_PER_LONG < 64)
1418 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1419
1420 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1421 {
1422 (ei)->i_state_flags = 0;
1423 }
1424 #else
1425 EXT4_INODE_BIT_FNS(state, flags, 32)
1426
1427 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1428 {
1429 /* We depend on the fact that callers will set i_flags */
1430 }
1431 #endif
1432 #else
1433 /* Assume that user mode programs are passing in an ext4fs superblock, not
1434 * a kernel struct super_block. This will allow us to call the feature-test
1435 * macros from user land. */
1436 #define EXT4_SB(sb) (sb)
1437 #endif
1438
1439 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1440
1441 /*
1442 * Codes for operating systems
1443 */
1444 #define EXT4_OS_LINUX 0
1445 #define EXT4_OS_HURD 1
1446 #define EXT4_OS_MASIX 2
1447 #define EXT4_OS_FREEBSD 3
1448 #define EXT4_OS_LITES 4
1449
1450 /*
1451 * Revision levels
1452 */
1453 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1454 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1455
1456 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1457 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1458
1459 #define EXT4_GOOD_OLD_INODE_SIZE 128
1460
1461 /*
1462 * Feature set definitions
1463 */
1464
1465 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1466 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1467 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1468 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1469 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1470 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1471 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1472 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1473 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1474 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1475 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1476 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1477 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1478 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1479 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1480 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1481 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1482 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1483
1484 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1485 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1486 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1487 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1488 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1489 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1490
1491 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1492 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1493 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1494 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1495 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1496 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1497 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1498 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1499 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1500 /*
1501 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1502 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1503 * all other data structures' checksums. However, the METADATA_CSUM and
1504 * GDT_CSUM bits are mutually exclusive.
1505 */
1506 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1507
1508 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1509 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1510 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1511 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1512 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1513 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1514 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1515 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1516 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1517 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1518 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1519 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1520 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1521 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1522
1523 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1524 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1525 EXT4_FEATURE_INCOMPAT_META_BG)
1526 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1527 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1528 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1529
1530 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1531 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1532 EXT4_FEATURE_INCOMPAT_RECOVER| \
1533 EXT4_FEATURE_INCOMPAT_META_BG)
1534 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1535 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1536 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1537
1538 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1539 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1540 EXT4_FEATURE_INCOMPAT_RECOVER| \
1541 EXT4_FEATURE_INCOMPAT_META_BG| \
1542 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1543 EXT4_FEATURE_INCOMPAT_64BIT| \
1544 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1545 EXT4_FEATURE_INCOMPAT_MMP | \
1546 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1547 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1548 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1549 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1550 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1551 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1552 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1553 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1554 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1555 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1556 EXT4_FEATURE_RO_COMPAT_QUOTA)
1557
1558 /*
1559 * Default values for user and/or group using reserved blocks
1560 */
1561 #define EXT4_DEF_RESUID 0
1562 #define EXT4_DEF_RESGID 0
1563
1564 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1565
1566 /*
1567 * Default mount options
1568 */
1569 #define EXT4_DEFM_DEBUG 0x0001
1570 #define EXT4_DEFM_BSDGROUPS 0x0002
1571 #define EXT4_DEFM_XATTR_USER 0x0004
1572 #define EXT4_DEFM_ACL 0x0008
1573 #define EXT4_DEFM_UID16 0x0010
1574 #define EXT4_DEFM_JMODE 0x0060
1575 #define EXT4_DEFM_JMODE_DATA 0x0020
1576 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1577 #define EXT4_DEFM_JMODE_WBACK 0x0060
1578 #define EXT4_DEFM_NOBARRIER 0x0100
1579 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1580 #define EXT4_DEFM_DISCARD 0x0400
1581 #define EXT4_DEFM_NODELALLOC 0x0800
1582
1583 /*
1584 * Default journal batch times
1585 */
1586 #define EXT4_DEF_MIN_BATCH_TIME 0
1587 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1588
1589 /*
1590 * Minimum number of groups in a flexgroup before we separate out
1591 * directories into the first block group of a flexgroup
1592 */
1593 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1594
1595 /*
1596 * Structure of a directory entry
1597 */
1598 #define EXT4_NAME_LEN 255
1599
1600 struct ext4_dir_entry {
1601 __le32 inode; /* Inode number */
1602 __le16 rec_len; /* Directory entry length */
1603 __le16 name_len; /* Name length */
1604 char name[EXT4_NAME_LEN]; /* File name */
1605 };
1606
1607 /*
1608 * The new version of the directory entry. Since EXT4 structures are
1609 * stored in intel byte order, and the name_len field could never be
1610 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1611 * file_type field.
1612 */
1613 struct ext4_dir_entry_2 {
1614 __le32 inode; /* Inode number */
1615 __le16 rec_len; /* Directory entry length */
1616 __u8 name_len; /* Name length */
1617 __u8 file_type;
1618 char name[EXT4_NAME_LEN]; /* File name */
1619 };
1620
1621 /*
1622 * This is a bogus directory entry at the end of each leaf block that
1623 * records checksums.
1624 */
1625 struct ext4_dir_entry_tail {
1626 __le32 det_reserved_zero1; /* Pretend to be unused */
1627 __le16 det_rec_len; /* 12 */
1628 __u8 det_reserved_zero2; /* Zero name length */
1629 __u8 det_reserved_ft; /* 0xDE, fake file type */
1630 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1631 };
1632
1633 #define EXT4_DIRENT_TAIL(block, blocksize) \
1634 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1635 ((blocksize) - \
1636 sizeof(struct ext4_dir_entry_tail))))
1637
1638 /*
1639 * Ext4 directory file types. Only the low 3 bits are used. The
1640 * other bits are reserved for now.
1641 */
1642 #define EXT4_FT_UNKNOWN 0
1643 #define EXT4_FT_REG_FILE 1
1644 #define EXT4_FT_DIR 2
1645 #define EXT4_FT_CHRDEV 3
1646 #define EXT4_FT_BLKDEV 4
1647 #define EXT4_FT_FIFO 5
1648 #define EXT4_FT_SOCK 6
1649 #define EXT4_FT_SYMLINK 7
1650
1651 #define EXT4_FT_MAX 8
1652
1653 #define EXT4_FT_DIR_CSUM 0xDE
1654
1655 /*
1656 * EXT4_DIR_PAD defines the directory entries boundaries
1657 *
1658 * NOTE: It must be a multiple of 4
1659 */
1660 #define EXT4_DIR_PAD 4
1661 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1662 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1663 ~EXT4_DIR_ROUND)
1664 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1665
1666 /*
1667 * If we ever get support for fs block sizes > page_size, we'll need
1668 * to remove the #if statements in the next two functions...
1669 */
1670 static inline unsigned int
1671 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1672 {
1673 unsigned len = le16_to_cpu(dlen);
1674
1675 #if (PAGE_CACHE_SIZE >= 65536)
1676 if (len == EXT4_MAX_REC_LEN || len == 0)
1677 return blocksize;
1678 return (len & 65532) | ((len & 3) << 16);
1679 #else
1680 return len;
1681 #endif
1682 }
1683
1684 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1685 {
1686 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1687 BUG();
1688 #if (PAGE_CACHE_SIZE >= 65536)
1689 if (len < 65536)
1690 return cpu_to_le16(len);
1691 if (len == blocksize) {
1692 if (blocksize == 65536)
1693 return cpu_to_le16(EXT4_MAX_REC_LEN);
1694 else
1695 return cpu_to_le16(0);
1696 }
1697 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1698 #else
1699 return cpu_to_le16(len);
1700 #endif
1701 }
1702
1703 /*
1704 * Hash Tree Directory indexing
1705 * (c) Daniel Phillips, 2001
1706 */
1707
1708 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1709 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1710 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1711 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1712 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1713
1714 /* Legal values for the dx_root hash_version field: */
1715
1716 #define DX_HASH_LEGACY 0
1717 #define DX_HASH_HALF_MD4 1
1718 #define DX_HASH_TEA 2
1719 #define DX_HASH_LEGACY_UNSIGNED 3
1720 #define DX_HASH_HALF_MD4_UNSIGNED 4
1721 #define DX_HASH_TEA_UNSIGNED 5
1722
1723 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1724 const void *address, unsigned int length)
1725 {
1726 struct {
1727 struct shash_desc shash;
1728 char ctx[4];
1729 } desc;
1730 int err;
1731
1732 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1733
1734 desc.shash.tfm = sbi->s_chksum_driver;
1735 desc.shash.flags = 0;
1736 *(u32 *)desc.ctx = crc;
1737
1738 err = crypto_shash_update(&desc.shash, address, length);
1739 BUG_ON(err);
1740
1741 return *(u32 *)desc.ctx;
1742 }
1743
1744 #ifdef __KERNEL__
1745
1746 /* hash info structure used by the directory hash */
1747 struct dx_hash_info
1748 {
1749 u32 hash;
1750 u32 minor_hash;
1751 int hash_version;
1752 u32 *seed;
1753 };
1754
1755
1756 /* 32 and 64 bit signed EOF for dx directories */
1757 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1758 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1759
1760
1761 /*
1762 * Control parameters used by ext4_htree_next_block
1763 */
1764 #define HASH_NB_ALWAYS 1
1765
1766
1767 /*
1768 * Describe an inode's exact location on disk and in memory
1769 */
1770 struct ext4_iloc
1771 {
1772 struct buffer_head *bh;
1773 unsigned long offset;
1774 ext4_group_t block_group;
1775 };
1776
1777 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1778 {
1779 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1780 }
1781
1782 /*
1783 * This structure is stuffed into the struct file's private_data field
1784 * for directories. It is where we put information so that we can do
1785 * readdir operations in hash tree order.
1786 */
1787 struct dir_private_info {
1788 struct rb_root root;
1789 struct rb_node *curr_node;
1790 struct fname *extra_fname;
1791 loff_t last_pos;
1792 __u32 curr_hash;
1793 __u32 curr_minor_hash;
1794 __u32 next_hash;
1795 };
1796
1797 /* calculate the first block number of the group */
1798 static inline ext4_fsblk_t
1799 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1800 {
1801 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1802 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1803 }
1804
1805 /*
1806 * Special error return code only used by dx_probe() and its callers.
1807 */
1808 #define ERR_BAD_DX_DIR -75000
1809
1810 /*
1811 * Timeout and state flag for lazy initialization inode thread.
1812 */
1813 #define EXT4_DEF_LI_WAIT_MULT 10
1814 #define EXT4_DEF_LI_MAX_START_DELAY 5
1815 #define EXT4_LAZYINIT_QUIT 0x0001
1816 #define EXT4_LAZYINIT_RUNNING 0x0002
1817
1818 /*
1819 * Lazy inode table initialization info
1820 */
1821 struct ext4_lazy_init {
1822 unsigned long li_state;
1823 struct list_head li_request_list;
1824 struct mutex li_list_mtx;
1825 };
1826
1827 struct ext4_li_request {
1828 struct super_block *lr_super;
1829 struct ext4_sb_info *lr_sbi;
1830 ext4_group_t lr_next_group;
1831 struct list_head lr_request;
1832 unsigned long lr_next_sched;
1833 unsigned long lr_timeout;
1834 };
1835
1836 struct ext4_features {
1837 struct kobject f_kobj;
1838 struct completion f_kobj_unregister;
1839 };
1840
1841 /*
1842 * This structure will be used for multiple mount protection. It will be
1843 * written into the block number saved in the s_mmp_block field in the
1844 * superblock. Programs that check MMP should assume that if
1845 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1846 * to use the filesystem, regardless of how old the timestamp is.
1847 */
1848 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1849 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1850 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1851 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1852
1853 struct mmp_struct {
1854 __le32 mmp_magic; /* Magic number for MMP */
1855 __le32 mmp_seq; /* Sequence no. updated periodically */
1856
1857 /*
1858 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1859 * purposes and do not affect the correctness of the algorithm
1860 */
1861 __le64 mmp_time; /* Time last updated */
1862 char mmp_nodename[64]; /* Node which last updated MMP block */
1863 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1864
1865 /*
1866 * mmp_check_interval is used to verify if the MMP block has been
1867 * updated on the block device. The value is updated based on the
1868 * maximum time to write the MMP block during an update cycle.
1869 */
1870 __le16 mmp_check_interval;
1871
1872 __le16 mmp_pad1;
1873 __le32 mmp_pad2[226];
1874 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1875 };
1876
1877 /* arguments passed to the mmp thread */
1878 struct mmpd_data {
1879 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1880 struct super_block *sb; /* super block of the fs */
1881 };
1882
1883 /*
1884 * Check interval multiplier
1885 * The MMP block is written every update interval and initially checked every
1886 * update interval x the multiplier (the value is then adapted based on the
1887 * write latency). The reason is that writes can be delayed under load and we
1888 * don't want readers to incorrectly assume that the filesystem is no longer
1889 * in use.
1890 */
1891 #define EXT4_MMP_CHECK_MULT 2UL
1892
1893 /*
1894 * Minimum interval for MMP checking in seconds.
1895 */
1896 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1897
1898 /*
1899 * Maximum interval for MMP checking in seconds.
1900 */
1901 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1902
1903 /*
1904 * Function prototypes
1905 */
1906
1907 /*
1908 * Ok, these declarations are also in <linux/kernel.h> but none of the
1909 * ext4 source programs needs to include it so they are duplicated here.
1910 */
1911 # define NORET_TYPE /**/
1912 # define ATTRIB_NORET __attribute__((noreturn))
1913 # define NORET_AND noreturn,
1914
1915 /* bitmap.c */
1916 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1917 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1918 struct ext4_group_desc *gdp,
1919 struct buffer_head *bh, int sz);
1920 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1921 struct ext4_group_desc *gdp,
1922 struct buffer_head *bh, int sz);
1923 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1924 struct ext4_group_desc *gdp,
1925 struct buffer_head *bh);
1926 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1927 struct ext4_group_desc *gdp,
1928 struct buffer_head *bh);
1929
1930 /* balloc.c */
1931 extern void ext4_get_group_no_and_offset(struct super_block *sb,
1932 ext4_fsblk_t blocknr,
1933 ext4_group_t *blockgrpp,
1934 ext4_grpblk_t *offsetp);
1935 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1936 ext4_fsblk_t block);
1937
1938 extern void ext4_validate_block_bitmap(struct super_block *sb,
1939 struct ext4_group_desc *desc,
1940 ext4_group_t block_group,
1941 struct buffer_head *bh);
1942 extern unsigned int ext4_block_group(struct super_block *sb,
1943 ext4_fsblk_t blocknr);
1944 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1945 ext4_fsblk_t blocknr);
1946 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1947 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1948 ext4_group_t group);
1949 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1950 ext4_fsblk_t goal,
1951 unsigned int flags,
1952 unsigned long *count,
1953 int *errp);
1954 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1955 s64 nclusters, unsigned int flags);
1956 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1957 extern void ext4_check_blocks_bitmap(struct super_block *);
1958 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1959 ext4_group_t block_group,
1960 struct buffer_head ** bh);
1961 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1962
1963 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1964 ext4_group_t block_group);
1965 extern int ext4_wait_block_bitmap(struct super_block *sb,
1966 ext4_group_t block_group,
1967 struct buffer_head *bh);
1968 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1969 ext4_group_t block_group);
1970 extern void ext4_init_block_bitmap(struct super_block *sb,
1971 struct buffer_head *bh,
1972 ext4_group_t group,
1973 struct ext4_group_desc *desc);
1974 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1975 ext4_group_t block_group,
1976 struct ext4_group_desc *gdp);
1977 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1978 ext4_group_t block_group,
1979 struct ext4_group_desc *gdp);
1980 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1981
1982 /* dir.c */
1983 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1984 struct file *,
1985 struct ext4_dir_entry_2 *,
1986 struct buffer_head *, char *, int,
1987 unsigned int);
1988 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1989 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1990 (de), (bh), (buf), (size), (offset)))
1991 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1992 __u32 minor_hash,
1993 struct ext4_dir_entry_2 *dirent);
1994 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1995 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1996 struct buffer_head *bh,
1997 void *buf, int buf_size,
1998 const char *name, int namelen,
1999 struct ext4_dir_entry_2 **dest_de);
2000 void ext4_insert_dentry(struct inode *inode,
2001 struct ext4_dir_entry_2 *de,
2002 int buf_size,
2003 const char *name, int namelen);
2004 static inline void ext4_update_dx_flag(struct inode *inode)
2005 {
2006 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2007 EXT4_FEATURE_COMPAT_DIR_INDEX))
2008 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2009 }
2010 static unsigned char ext4_filetype_table[] = {
2011 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2012 };
2013
2014 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2015 {
2016 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2017 (filetype >= EXT4_FT_MAX))
2018 return DT_UNKNOWN;
2019
2020 return ext4_filetype_table[filetype];
2021 }
2022
2023 /* fsync.c */
2024 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2025
2026 /* hash.c */
2027 extern int ext4fs_dirhash(const char *name, int len, struct
2028 dx_hash_info *hinfo);
2029
2030 /* ialloc.c */
2031 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2032 const struct qstr *qstr, __u32 goal,
2033 uid_t *owner, int handle_type,
2034 unsigned int line_no, int nblocks);
2035
2036 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2037 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2038 0, 0, 0)
2039 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2040 type, nblocks) \
2041 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2042 (type), __LINE__, (nblocks))
2043
2044
2045 extern void ext4_free_inode(handle_t *, struct inode *);
2046 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2047 extern unsigned long ext4_count_free_inodes(struct super_block *);
2048 extern unsigned long ext4_count_dirs(struct super_block *);
2049 extern void ext4_check_inodes_bitmap(struct super_block *);
2050 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2051 extern int ext4_init_inode_table(struct super_block *sb,
2052 ext4_group_t group, int barrier);
2053 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2054
2055 /* mballoc.c */
2056 extern long ext4_mb_stats;
2057 extern long ext4_mb_max_to_scan;
2058 extern int ext4_mb_init(struct super_block *);
2059 extern int ext4_mb_release(struct super_block *);
2060 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2061 struct ext4_allocation_request *, int *);
2062 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2063 extern void ext4_discard_preallocations(struct inode *);
2064 extern int __init ext4_init_mballoc(void);
2065 extern void ext4_exit_mballoc(void);
2066 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2067 struct buffer_head *bh, ext4_fsblk_t block,
2068 unsigned long count, int flags);
2069 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2070 ext4_group_t ngroups);
2071 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2072 ext4_group_t i, struct ext4_group_desc *desc);
2073 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2074 ext4_fsblk_t block, unsigned long count);
2075 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2076
2077 /* inode.c */
2078 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2079 ext4_lblk_t, int, int *);
2080 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2081 ext4_lblk_t, int, int *);
2082 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2083 struct buffer_head *bh_result, int create);
2084 int ext4_get_block(struct inode *inode, sector_t iblock,
2085 struct buffer_head *bh_result, int create);
2086 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2087 struct buffer_head *bh, int create);
2088 int ext4_walk_page_buffers(handle_t *handle,
2089 struct buffer_head *head,
2090 unsigned from,
2091 unsigned to,
2092 int *partial,
2093 int (*fn)(handle_t *handle,
2094 struct buffer_head *bh));
2095 int do_journal_get_write_access(handle_t *handle,
2096 struct buffer_head *bh);
2097 #define FALL_BACK_TO_NONDELALLOC 1
2098 #define CONVERT_INLINE_DATA 2
2099
2100 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2101 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2102 extern int ext4_setattr(struct dentry *, struct iattr *);
2103 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2104 struct kstat *stat);
2105 extern void ext4_evict_inode(struct inode *);
2106 extern void ext4_clear_inode(struct inode *);
2107 extern int ext4_sync_inode(handle_t *, struct inode *);
2108 extern void ext4_dirty_inode(struct inode *, int);
2109 extern int ext4_change_inode_journal_flag(struct inode *, int);
2110 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2111 extern int ext4_inode_attach_jinode(struct inode *inode);
2112 extern int ext4_can_truncate(struct inode *inode);
2113 extern void ext4_truncate(struct inode *);
2114 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2115 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2116 extern void ext4_set_inode_flags(struct inode *);
2117 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2118 extern int ext4_alloc_da_blocks(struct inode *inode);
2119 extern void ext4_set_aops(struct inode *inode);
2120 extern int ext4_writepage_trans_blocks(struct inode *);
2121 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2122 extern int ext4_block_truncate_page(handle_t *handle,
2123 struct address_space *mapping, loff_t from);
2124 extern int ext4_block_zero_page_range(handle_t *handle,
2125 struct address_space *mapping, loff_t from, loff_t length);
2126 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2127 loff_t lstart, loff_t lend);
2128 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2129 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2130 extern void ext4_da_update_reserve_space(struct inode *inode,
2131 int used, int quota_claim);
2132
2133 /* indirect.c */
2134 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2135 struct ext4_map_blocks *map, int flags);
2136 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2137 const struct iovec *iov, loff_t offset,
2138 unsigned long nr_segs);
2139 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2140 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2141 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2142 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
2143 ext4_lblk_t first, ext4_lblk_t stop);
2144
2145 /* ioctl.c */
2146 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2147 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2148
2149 /* migrate.c */
2150 extern int ext4_ext_migrate(struct inode *);
2151 extern int ext4_ind_migrate(struct inode *inode);
2152
2153 /* namei.c */
2154 extern int ext4_dirent_csum_verify(struct inode *inode,
2155 struct ext4_dir_entry *dirent);
2156 extern int ext4_orphan_add(handle_t *, struct inode *);
2157 extern int ext4_orphan_del(handle_t *, struct inode *);
2158 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2159 __u32 start_minor_hash, __u32 *next_hash);
2160 extern int search_dir(struct buffer_head *bh,
2161 char *search_buf,
2162 int buf_size,
2163 struct inode *dir,
2164 const struct qstr *d_name,
2165 unsigned int offset,
2166 struct ext4_dir_entry_2 **res_dir);
2167 extern int ext4_generic_delete_entry(handle_t *handle,
2168 struct inode *dir,
2169 struct ext4_dir_entry_2 *de_del,
2170 struct buffer_head *bh,
2171 void *entry_buf,
2172 int buf_size,
2173 int csum_size);
2174
2175 /* resize.c */
2176 extern int ext4_group_add(struct super_block *sb,
2177 struct ext4_new_group_data *input);
2178 extern int ext4_group_extend(struct super_block *sb,
2179 struct ext4_super_block *es,
2180 ext4_fsblk_t n_blocks_count);
2181 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2182
2183 /* super.c */
2184 extern int ext4_calculate_overhead(struct super_block *sb);
2185 extern int ext4_superblock_csum_verify(struct super_block *sb,
2186 struct ext4_super_block *es);
2187 extern void ext4_superblock_csum_set(struct super_block *sb);
2188 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2189 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2190 extern void ext4_kvfree(void *ptr);
2191 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2192 ext4_group_t ngroup);
2193 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2194 char nbuf[16]);
2195
2196 extern __printf(4, 5)
2197 void __ext4_error(struct super_block *, const char *, unsigned int,
2198 const char *, ...);
2199 extern __printf(5, 6)
2200 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2201 const char *, ...);
2202 extern __printf(5, 6)
2203 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2204 const char *, ...);
2205 extern void __ext4_std_error(struct super_block *, const char *,
2206 unsigned int, int);
2207 extern __printf(4, 5)
2208 void __ext4_abort(struct super_block *, const char *, unsigned int,
2209 const char *, ...);
2210 extern __printf(4, 5)
2211 void __ext4_warning(struct super_block *, const char *, unsigned int,
2212 const char *, ...);
2213 extern __printf(3, 4)
2214 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2215 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2216 const char *, unsigned int, const char *);
2217 extern __printf(7, 8)
2218 void __ext4_grp_locked_error(const char *, unsigned int,
2219 struct super_block *, ext4_group_t,
2220 unsigned long, ext4_fsblk_t,
2221 const char *, ...);
2222
2223 #ifdef CONFIG_PRINTK
2224
2225 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2226 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2227 #define ext4_error_file(file, func, line, block, fmt, ...) \
2228 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2229 #define ext4_error(sb, fmt, ...) \
2230 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2231 #define ext4_abort(sb, fmt, ...) \
2232 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2233 #define ext4_warning(sb, fmt, ...) \
2234 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2235 #define ext4_msg(sb, level, fmt, ...) \
2236 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2237 #define dump_mmp_msg(sb, mmp, msg) \
2238 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2239 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2240 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2241 fmt, ##__VA_ARGS__)
2242
2243 #else
2244
2245 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2246 do { \
2247 no_printk(fmt, ##__VA_ARGS__); \
2248 __ext4_error_inode(inode, "", 0, block, " "); \
2249 } while (0)
2250 #define ext4_error_file(file, func, line, block, fmt, ...) \
2251 do { \
2252 no_printk(fmt, ##__VA_ARGS__); \
2253 __ext4_error_file(file, "", 0, block, " "); \
2254 } while (0)
2255 #define ext4_error(sb, fmt, ...) \
2256 do { \
2257 no_printk(fmt, ##__VA_ARGS__); \
2258 __ext4_error(sb, "", 0, " "); \
2259 } while (0)
2260 #define ext4_abort(sb, fmt, ...) \
2261 do { \
2262 no_printk(fmt, ##__VA_ARGS__); \
2263 __ext4_abort(sb, "", 0, " "); \
2264 } while (0)
2265 #define ext4_warning(sb, fmt, ...) \
2266 do { \
2267 no_printk(fmt, ##__VA_ARGS__); \
2268 __ext4_warning(sb, "", 0, " "); \
2269 } while (0)
2270 #define ext4_msg(sb, level, fmt, ...) \
2271 do { \
2272 no_printk(fmt, ##__VA_ARGS__); \
2273 __ext4_msg(sb, "", " "); \
2274 } while (0)
2275 #define dump_mmp_msg(sb, mmp, msg) \
2276 __dump_mmp_msg(sb, mmp, "", 0, "")
2277 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2278 do { \
2279 no_printk(fmt, ##__VA_ARGS__); \
2280 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2281 } while (0)
2282
2283 #endif
2284
2285 extern void ext4_update_dynamic_rev(struct super_block *sb);
2286 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2287 __u32 compat);
2288 extern int ext4_update_rocompat_feature(handle_t *handle,
2289 struct super_block *sb, __u32 rocompat);
2290 extern int ext4_update_incompat_feature(handle_t *handle,
2291 struct super_block *sb, __u32 incompat);
2292 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2293 struct ext4_group_desc *bg);
2294 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2295 struct ext4_group_desc *bg);
2296 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2297 struct ext4_group_desc *bg);
2298 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2299 struct ext4_group_desc *bg);
2300 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2301 struct ext4_group_desc *bg);
2302 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2303 struct ext4_group_desc *bg);
2304 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2305 struct ext4_group_desc *bg);
2306 extern void ext4_block_bitmap_set(struct super_block *sb,
2307 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2308 extern void ext4_inode_bitmap_set(struct super_block *sb,
2309 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2310 extern void ext4_inode_table_set(struct super_block *sb,
2311 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2312 extern void ext4_free_group_clusters_set(struct super_block *sb,
2313 struct ext4_group_desc *bg,
2314 __u32 count);
2315 extern void ext4_free_inodes_set(struct super_block *sb,
2316 struct ext4_group_desc *bg, __u32 count);
2317 extern void ext4_used_dirs_set(struct super_block *sb,
2318 struct ext4_group_desc *bg, __u32 count);
2319 extern void ext4_itable_unused_set(struct super_block *sb,
2320 struct ext4_group_desc *bg, __u32 count);
2321 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2322 struct ext4_group_desc *gdp);
2323 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2324 struct ext4_group_desc *gdp);
2325 extern int ext4_register_li_request(struct super_block *sb,
2326 ext4_group_t first_not_zeroed);
2327
2328 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2329 {
2330 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2331 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2332 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2333 }
2334
2335 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2336 {
2337 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2338 le32_to_cpu(es->s_blocks_count_lo);
2339 }
2340
2341 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2342 {
2343 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2344 le32_to_cpu(es->s_r_blocks_count_lo);
2345 }
2346
2347 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2348 {
2349 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2350 le32_to_cpu(es->s_free_blocks_count_lo);
2351 }
2352
2353 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2354 ext4_fsblk_t blk)
2355 {
2356 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2357 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2358 }
2359
2360 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2361 ext4_fsblk_t blk)
2362 {
2363 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2364 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2365 }
2366
2367 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2368 ext4_fsblk_t blk)
2369 {
2370 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2371 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2372 }
2373
2374 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2375 {
2376 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2377 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2378 le32_to_cpu(raw_inode->i_size_lo);
2379 else
2380 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2381 }
2382
2383 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2384 {
2385 raw_inode->i_size_lo = cpu_to_le32(i_size);
2386 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2387 }
2388
2389 static inline
2390 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2391 ext4_group_t group)
2392 {
2393 struct ext4_group_info ***grp_info;
2394 long indexv, indexh;
2395 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2396 grp_info = EXT4_SB(sb)->s_group_info;
2397 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2398 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2399 return grp_info[indexv][indexh];
2400 }
2401
2402 /*
2403 * Reading s_groups_count requires using smp_rmb() afterwards. See
2404 * the locking protocol documented in the comments of ext4_group_add()
2405 * in resize.c
2406 */
2407 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2408 {
2409 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2410
2411 smp_rmb();
2412 return ngroups;
2413 }
2414
2415 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2416 ext4_group_t block_group)
2417 {
2418 return block_group >> sbi->s_log_groups_per_flex;
2419 }
2420
2421 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2422 {
2423 return 1 << sbi->s_log_groups_per_flex;
2424 }
2425
2426 #define ext4_std_error(sb, errno) \
2427 do { \
2428 if ((errno)) \
2429 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2430 } while (0)
2431
2432 #ifdef CONFIG_SMP
2433 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2434 * counters. So we need to make sure we have free clusters more
2435 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2436 */
2437 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2438 #else
2439 #define EXT4_FREECLUSTERS_WATERMARK 0
2440 #endif
2441
2442 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2443 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2444 {
2445 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2446 !mutex_is_locked(&inode->i_mutex));
2447 down_write(&EXT4_I(inode)->i_data_sem);
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 /*
2454 * Update i_disksize after writeback has been started. Races with truncate
2455 * are avoided by checking i_size under i_data_sem.
2456 */
2457 static inline void ext4_wb_update_i_disksize(struct inode *inode, loff_t newsize)
2458 {
2459 loff_t i_size;
2460
2461 down_write(&EXT4_I(inode)->i_data_sem);
2462 i_size = i_size_read(inode);
2463 if (newsize > i_size)
2464 newsize = i_size;
2465 if (newsize > EXT4_I(inode)->i_disksize)
2466 EXT4_I(inode)->i_disksize = newsize;
2467 up_write(&EXT4_I(inode)->i_data_sem);
2468 }
2469
2470 struct ext4_group_info {
2471 unsigned long bb_state;
2472 struct rb_root bb_free_root;
2473 ext4_grpblk_t bb_first_free; /* first free block */
2474 ext4_grpblk_t bb_free; /* total free blocks */
2475 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2476 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2477 struct list_head bb_prealloc_list;
2478 #ifdef DOUBLE_CHECK
2479 void *bb_bitmap;
2480 #endif
2481 struct rw_semaphore alloc_sem;
2482 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2483 * regions, index is order.
2484 * bb_counters[3] = 5 means
2485 * 5 free 8-block regions. */
2486 };
2487
2488 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2489 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2490 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2491 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2492
2493 #define EXT4_MB_GRP_NEED_INIT(grp) \
2494 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2495 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2496 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2497 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2498 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2499
2500 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2501 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2502 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2503 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2504 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2505 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2506
2507 #define EXT4_MAX_CONTENTION 8
2508 #define EXT4_CONTENTION_THRESHOLD 2
2509
2510 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2511 ext4_group_t group)
2512 {
2513 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2514 }
2515
2516 /*
2517 * Returns true if the filesystem is busy enough that attempts to
2518 * access the block group locks has run into contention.
2519 */
2520 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2521 {
2522 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2523 }
2524
2525 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2526 {
2527 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2528 if (spin_trylock(lock))
2529 /*
2530 * We're able to grab the lock right away, so drop the
2531 * lock contention counter.
2532 */
2533 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2534 else {
2535 /*
2536 * The lock is busy, so bump the contention counter,
2537 * and then wait on the spin lock.
2538 */
2539 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2540 EXT4_MAX_CONTENTION);
2541 spin_lock(lock);
2542 }
2543 }
2544
2545 static inline void ext4_unlock_group(struct super_block *sb,
2546 ext4_group_t group)
2547 {
2548 spin_unlock(ext4_group_lock_ptr(sb, group));
2549 }
2550
2551 /*
2552 * Block validity checking
2553 */
2554 #define ext4_check_indirect_blockref(inode, bh) \
2555 ext4_check_blockref(__func__, __LINE__, inode, \
2556 (__le32 *)(bh)->b_data, \
2557 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2558
2559 #define ext4_ind_check_inode(inode) \
2560 ext4_check_blockref(__func__, __LINE__, inode, \
2561 EXT4_I(inode)->i_data, \
2562 EXT4_NDIR_BLOCKS)
2563
2564 /*
2565 * Inodes and files operations
2566 */
2567
2568 /* dir.c */
2569 extern const struct file_operations ext4_dir_operations;
2570
2571 /* file.c */
2572 extern const struct inode_operations ext4_file_inode_operations;
2573 extern const struct file_operations ext4_file_operations;
2574 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2575 extern void ext4_unwritten_wait(struct inode *inode);
2576
2577 /* inline.c */
2578 extern int ext4_has_inline_data(struct inode *inode);
2579 extern int ext4_get_inline_size(struct inode *inode);
2580 extern int ext4_get_max_inline_size(struct inode *inode);
2581 extern int ext4_find_inline_data_nolock(struct inode *inode);
2582 extern void ext4_write_inline_data(struct inode *inode,
2583 struct ext4_iloc *iloc,
2584 void *buffer, loff_t pos,
2585 unsigned int len);
2586 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2587 unsigned int len);
2588 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2589 unsigned int len);
2590 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2591
2592 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2593 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2594 struct inode *inode,
2595 loff_t pos, unsigned len,
2596 unsigned flags,
2597 struct page **pagep);
2598 extern int ext4_write_inline_data_end(struct inode *inode,
2599 loff_t pos, unsigned len,
2600 unsigned copied,
2601 struct page *page);
2602 extern struct buffer_head *
2603 ext4_journalled_write_inline_data(struct inode *inode,
2604 unsigned len,
2605 struct page *page);
2606 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2607 struct inode *inode,
2608 loff_t pos, unsigned len,
2609 unsigned flags,
2610 struct page **pagep,
2611 void **fsdata);
2612 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2613 unsigned len, unsigned copied,
2614 struct page *page);
2615 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2616 struct inode *inode);
2617 extern int ext4_try_create_inline_dir(handle_t *handle,
2618 struct inode *parent,
2619 struct inode *inode);
2620 extern int ext4_read_inline_dir(struct file *filp,
2621 struct dir_context *ctx,
2622 int *has_inline_data);
2623 extern int htree_inlinedir_to_tree(struct file *dir_file,
2624 struct inode *dir, ext4_lblk_t block,
2625 struct dx_hash_info *hinfo,
2626 __u32 start_hash, __u32 start_minor_hash,
2627 int *has_inline_data);
2628 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2629 const struct qstr *d_name,
2630 struct ext4_dir_entry_2 **res_dir,
2631 int *has_inline_data);
2632 extern int ext4_delete_inline_entry(handle_t *handle,
2633 struct inode *dir,
2634 struct ext4_dir_entry_2 *de_del,
2635 struct buffer_head *bh,
2636 int *has_inline_data);
2637 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2638 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2639 struct ext4_dir_entry_2 **parent_de,
2640 int *retval);
2641 extern int ext4_inline_data_fiemap(struct inode *inode,
2642 struct fiemap_extent_info *fieinfo,
2643 int *has_inline);
2644 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2645 struct inode *inode,
2646 int needed);
2647 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2648
2649 extern int ext4_convert_inline_data(struct inode *inode);
2650
2651 /* namei.c */
2652 extern const struct inode_operations ext4_dir_inode_operations;
2653 extern const struct inode_operations ext4_special_inode_operations;
2654 extern struct dentry *ext4_get_parent(struct dentry *child);
2655 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2656 struct ext4_dir_entry_2 *de,
2657 int blocksize, int csum_size,
2658 unsigned int parent_ino, int dotdot_real_len);
2659 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2660 unsigned int blocksize);
2661 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2662 struct inode *inode,
2663 struct buffer_head *bh);
2664 #define S_SHIFT 12
2665 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2666 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2667 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2668 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2669 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2670 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2671 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2672 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2673 };
2674
2675 static inline void ext4_set_de_type(struct super_block *sb,
2676 struct ext4_dir_entry_2 *de,
2677 umode_t mode) {
2678 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2679 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2680 }
2681
2682
2683 /* symlink.c */
2684 extern const struct inode_operations ext4_symlink_inode_operations;
2685 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2686
2687 /* block_validity */
2688 extern void ext4_release_system_zone(struct super_block *sb);
2689 extern int ext4_setup_system_zone(struct super_block *sb);
2690 extern int __init ext4_init_system_zone(void);
2691 extern void ext4_exit_system_zone(void);
2692 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2693 ext4_fsblk_t start_blk,
2694 unsigned int count);
2695 extern int ext4_check_blockref(const char *, unsigned int,
2696 struct inode *, __le32 *, unsigned int);
2697
2698 /* extents.c */
2699 struct ext4_ext_path;
2700 struct ext4_extent;
2701
2702 /*
2703 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2704 * __le32.
2705 */
2706 #define EXT_MAX_BLOCKS 0xffffffff
2707
2708 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2709 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2710 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2711 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2712 struct ext4_map_blocks *map, int flags);
2713 extern void ext4_ext_truncate(handle_t *, struct inode *);
2714 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2715 ext4_lblk_t end);
2716 extern void ext4_ext_init(struct super_block *);
2717 extern void ext4_ext_release(struct super_block *);
2718 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2719 loff_t len);
2720 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2721 loff_t offset, ssize_t len);
2722 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2723 struct ext4_map_blocks *map, int flags);
2724 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2725 ext4_lblk_t lblocks);
2726 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2727 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2728 int num,
2729 struct ext4_ext_path *path);
2730 extern int ext4_can_extents_be_merged(struct inode *inode,
2731 struct ext4_extent *ex1,
2732 struct ext4_extent *ex2);
2733 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2734 struct ext4_ext_path *,
2735 struct ext4_extent *, int);
2736 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2737 struct ext4_ext_path *,
2738 int flags);
2739 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2740 extern int ext4_ext_check_inode(struct inode *inode);
2741 extern int ext4_find_delalloc_range(struct inode *inode,
2742 ext4_lblk_t lblk_start,
2743 ext4_lblk_t lblk_end);
2744 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2745 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2746 __u64 start, __u64 len);
2747 extern int ext4_ext_precache(struct inode *inode);
2748
2749 /* move_extent.c */
2750 extern void ext4_double_down_write_data_sem(struct inode *first,
2751 struct inode *second);
2752 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2753 struct inode *donor_inode);
2754 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2755 __u64 start_orig, __u64 start_donor,
2756 __u64 len, __u64 *moved_len);
2757
2758 /* page-io.c */
2759 extern int __init ext4_init_pageio(void);
2760 extern void ext4_exit_pageio(void);
2761 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2762 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2763 extern int ext4_put_io_end(ext4_io_end_t *io_end);
2764 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2765 extern void ext4_io_submit_init(struct ext4_io_submit *io,
2766 struct writeback_control *wbc);
2767 extern void ext4_end_io_rsv_work(struct work_struct *work);
2768 extern void ext4_io_submit(struct ext4_io_submit *io);
2769 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2770 struct page *page,
2771 int len,
2772 struct writeback_control *wbc);
2773
2774 /* mmp.c */
2775 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2776 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2777 extern int ext4_mmp_csum_verify(struct super_block *sb,
2778 struct mmp_struct *mmp);
2779
2780 /*
2781 * Note that these flags will never ever appear in a buffer_head's state flag.
2782 * See EXT4_MAP_... to see where this is used.
2783 */
2784 enum ext4_state_bits {
2785 BH_Uninit /* blocks are allocated but uninitialized on disk */
2786 = BH_JBDPrivateStart,
2787 BH_AllocFromCluster, /* allocated blocks were part of already
2788 * allocated cluster. */
2789 };
2790
2791 /*
2792 * Add new method to test whether block and inode bitmaps are properly
2793 * initialized. With uninit_bg reading the block from disk is not enough
2794 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2795 */
2796 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2797
2798 static inline int bitmap_uptodate(struct buffer_head *bh)
2799 {
2800 return (buffer_uptodate(bh) &&
2801 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2802 }
2803 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2804 {
2805 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2806 }
2807
2808 /*
2809 * Disable DIO read nolock optimization, so new dioreaders will be forced
2810 * to grab i_mutex
2811 */
2812 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2813 {
2814 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2815 smp_mb();
2816 }
2817 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2818 {
2819 smp_mb();
2820 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2821 }
2822
2823 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2824
2825 /* For ioend & aio unwritten conversion wait queues */
2826 #define EXT4_WQ_HASH_SZ 37
2827 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2828 EXT4_WQ_HASH_SZ])
2829 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2830 EXT4_WQ_HASH_SZ])
2831 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2832 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2833
2834 #define EXT4_RESIZING 0
2835 extern int ext4_resize_begin(struct super_block *sb);
2836 extern void ext4_resize_end(struct super_block *sb);
2837
2838 #endif /* __KERNEL__ */
2839
2840 #endif /* _EXT4_H */
Linus' kernel tree