?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
sctp: translate host order to network order when setting a hmacid
[linux/fpc-iii.git] / fs / ext4 / ext4.h
blob2a71466b0115005dc940ca85f194ba1eb83ea046
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #include <crypto/hash.h>
33 #ifdef __KERNEL__
34 #include <linux/compat.h>
35 #endif
36
37 /*
38 * The fourth extended filesystem constants/structures
39 */
40
41 /*
42 * Define EXT4FS_DEBUG to produce debug messages
43 */
44 #undef EXT4FS_DEBUG
45
46 /*
47 * Debug code
48 */
49 #ifdef EXT4FS_DEBUG
50 #define ext4_debug(f, a...) \
51 do { \
52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
53 __FILE__, __LINE__, __func__); \
54 printk(KERN_DEBUG f, ## a); \
55 } while (0)
56 #else
57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
58 #endif
59
60 /*
61 * Turn on EXT_DEBUG to get lots of info about extents operations.
62 */
63 #define EXT_DEBUG__
64 #ifdef EXT_DEBUG
65 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
66 #else
67 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
68 #endif
69
70 #define EXT4_ERROR_INODE(inode, fmt, a...) \
71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
72
73 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
75
76 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
78
79 /* data type for block offset of block group */
80 typedef int ext4_grpblk_t;
81
82 /* data type for filesystem-wide blocks number */
83 typedef unsigned long long ext4_fsblk_t;
84
85 /* data type for file logical block number */
86 typedef __u32 ext4_lblk_t;
87
88 /* data type for block group number */
89 typedef unsigned int ext4_group_t;
90
91 /*
92 * Flags used in mballoc's allocation_context flags field.
93 *
94 * Also used to show what's going on for debugging purposes when the
95 * flag field is exported via the traceport interface
96 */
97
98 /* prefer goal again. length */
99 #define EXT4_MB_HINT_MERGE 0x0001
100 /* blocks already reserved */
101 #define EXT4_MB_HINT_RESERVED 0x0002
102 /* metadata is being allocated */
103 #define EXT4_MB_HINT_METADATA 0x0004
104 /* first blocks in the file */
105 #define EXT4_MB_HINT_FIRST 0x0008
106 /* search for the best chunk */
107 #define EXT4_MB_HINT_BEST 0x0010
108 /* data is being allocated */
109 #define EXT4_MB_HINT_DATA 0x0020
110 /* don't preallocate (for tails) */
111 #define EXT4_MB_HINT_NOPREALLOC 0x0040
112 /* allocate for locality group */
113 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
114 /* allocate goal blocks or none */
115 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
116 /* goal is meaningful */
117 #define EXT4_MB_HINT_TRY_GOAL 0x0200
118 /* blocks already pre-reserved by delayed allocation */
119 #define EXT4_MB_DELALLOC_RESERVED 0x0400
120 /* We are doing stream allocation */
121 #define EXT4_MB_STREAM_ALLOC 0x0800
122 /* Use reserved root blocks if needed */
123 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
124 /* Use blocks from reserved pool */
125 #define EXT4_MB_USE_RESERVED 0x2000
126
127 struct ext4_allocation_request {
128 /* target inode for block we're allocating */
129 struct inode *inode;
130 /* how many blocks we want to allocate */
131 unsigned int len;
132 /* logical block in target inode */
133 ext4_lblk_t logical;
134 /* the closest logical allocated block to the left */
135 ext4_lblk_t lleft;
136 /* the closest logical allocated block to the right */
137 ext4_lblk_t lright;
138 /* phys. target (a hint) */
139 ext4_fsblk_t goal;
140 /* phys. block for the closest logical allocated block to the left */
141 ext4_fsblk_t pleft;
142 /* phys. block for the closest logical allocated block to the right */
143 ext4_fsblk_t pright;
144 /* flags. see above EXT4_MB_HINT_* */
145 unsigned int flags;
146 };
147
148 /*
149 * Logical to physical block mapping, used by ext4_map_blocks()
150 *
151 * This structure is used to pass requests into ext4_map_blocks() as
152 * well as to store the information returned by ext4_map_blocks(). It
153 * takes less room on the stack than a struct buffer_head.
154 */
155 #define EXT4_MAP_NEW (1 << BH_New)
156 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
157 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
158 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
159 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
160 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
161 * ext4_map_blocks wants to know whether or not the underlying cluster has
162 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
163 * the requested mapping was from previously mapped (or delayed allocated)
164 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
165 * should never appear on buffer_head's state flags.
166 */
167 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
168 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
169 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
170 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
171
172 struct ext4_map_blocks {
173 ext4_fsblk_t m_pblk;
174 ext4_lblk_t m_lblk;
175 unsigned int m_len;
176 unsigned int m_flags;
177 };
178
179 /*
180 * For delayed allocation tracking
181 */
182 struct mpage_da_data {
183 struct inode *inode;
184 sector_t b_blocknr; /* start block number of extent */
185 size_t b_size; /* size of extent */
186 unsigned long b_state; /* state of the extent */
187 unsigned long first_page, next_page; /* extent of pages */
188 struct writeback_control *wbc;
189 int io_done;
190 int pages_written;
191 int retval;
192 };
193
194 /*
195 * Flags for ext4_io_end->flags
196 */
197 #define EXT4_IO_END_UNWRITTEN 0x0001
198 #define EXT4_IO_END_ERROR 0x0002
199 #define EXT4_IO_END_DIRECT 0x0004
200
201 /*
202 * For converting uninitialized extents on a work queue.
203 */
204 typedef struct ext4_io_end {
205 struct list_head list; /* per-file finished IO list */
206 struct inode *inode; /* file being written to */
207 unsigned int flag; /* unwritten or not */
208 loff_t offset; /* offset in the file */
209 ssize_t size; /* size of the extent */
210 struct kiocb *iocb; /* iocb struct for AIO */
211 int result; /* error value for AIO */
212 } ext4_io_end_t;
213
214 struct ext4_io_submit {
215 int io_op;
216 struct bio *io_bio;
217 ext4_io_end_t *io_end;
218 sector_t io_next_block;
219 };
220
221 /*
222 * Special inodes numbers
223 */
224 #define EXT4_BAD_INO 1 /* Bad blocks inode */
225 #define EXT4_ROOT_INO 2 /* Root inode */
226 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
227 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
228 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
229 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
230 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
231 #define EXT4_JOURNAL_INO 8 /* Journal inode */
232
233 /* First non-reserved inode for old ext4 filesystems */
234 #define EXT4_GOOD_OLD_FIRST_INO 11
235
236 /*
237 * Maximal count of links to a file
238 */
239 #define EXT4_LINK_MAX 65000
240
241 /*
242 * Macro-instructions used to manage several block sizes
243 */
244 #define EXT4_MIN_BLOCK_SIZE 1024
245 #define EXT4_MAX_BLOCK_SIZE 65536
246 #define EXT4_MIN_BLOCK_LOG_SIZE 10
247 #define EXT4_MAX_BLOCK_LOG_SIZE 16
248 #ifdef __KERNEL__
249 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
250 #else
251 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
252 #endif
253 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
254 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
255 EXT4_SB(s)->s_cluster_bits)
256 #ifdef __KERNEL__
257 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
258 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
259 #else
260 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
261 #endif
262 #ifdef __KERNEL__
263 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
264 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
265 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
266 #else
267 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
268 EXT4_GOOD_OLD_INODE_SIZE : \
269 (s)->s_inode_size)
270 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
271 EXT4_GOOD_OLD_FIRST_INO : \
272 (s)->s_first_ino)
273 #endif
274 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
275
276 /* Translate a block number to a cluster number */
277 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
278 /* Translate a cluster number to a block number */
279 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
280 /* Translate # of blks to # of clusters */
281 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
282 (sbi)->s_cluster_bits)
283 /* Mask out the low bits to get the starting block of the cluster */
284 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
285 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
286 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
287 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
288 /* Get the cluster offset */
289 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
290 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
291 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
292 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
293
294 /*
295 * Structure of a blocks group descriptor
296 */
297 struct ext4_group_desc
298 {
299 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
300 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
301 __le32 bg_inode_table_lo; /* Inodes table block */
302 __le16 bg_free_blocks_count_lo;/* Free blocks count */
303 __le16 bg_free_inodes_count_lo;/* Free inodes count */
304 __le16 bg_used_dirs_count_lo; /* Directories count */
305 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
306 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
307 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
308 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
309 __le16 bg_itable_unused_lo; /* Unused inodes count */
310 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
311 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
312 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
313 __le32 bg_inode_table_hi; /* Inodes table block MSB */
314 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
315 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
316 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
317 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
318 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
319 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
320 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
321 __u32 bg_reserved;
322 };
323
324 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
325 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
326 sizeof(__le16))
327 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
328 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
329 sizeof(__le16))
330
331 /*
332 * Structure of a flex block group info
333 */
334
335 struct flex_groups {
336 atomic64_t free_clusters;
337 atomic_t free_inodes;
338 atomic_t used_dirs;
339 };
340
341 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
342 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
343 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
344
345 /*
346 * Macro-instructions used to manage group descriptors
347 */
348 #define EXT4_MIN_DESC_SIZE 32
349 #define EXT4_MIN_DESC_SIZE_64BIT 64
350 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
351 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
352 #ifdef __KERNEL__
353 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
354 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
356 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
357 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
358 #else
359 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
361 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
362 #endif
363
364 /*
365 * Constants relative to the data blocks
366 */
367 #define EXT4_NDIR_BLOCKS 12
368 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
369 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
370 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
371 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
372
373 /*
374 * Inode flags
375 */
376 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
377 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
378 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
379 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
380 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
381 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
382 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
383 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
384 /* Reserved for compression usage... */
385 #define EXT4_DIRTY_FL 0x00000100
386 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
387 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
388 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
389 /* End compression flags --- maybe not all used */
390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
401 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
402
403 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
404 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
405
406 /* Flags that should be inherited by new inodes from their parent. */
407 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
408 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
409 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
410 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
411
412 /* Flags that are appropriate for regular files (all but dir-specific ones). */
413 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
414
415 /* Flags that are appropriate for non-directories/regular files. */
416 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
417
418 /* Mask out flags that are inappropriate for the given type of inode. */
419 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
420 {
421 if (S_ISDIR(mode))
422 return flags;
423 else if (S_ISREG(mode))
424 return flags & EXT4_REG_FLMASK;
425 else
426 return flags & EXT4_OTHER_FLMASK;
427 }
428
429 /*
430 * Inode flags used for atomic set/get
431 */
432 enum {
433 EXT4_INODE_SECRM = 0, /* Secure deletion */
434 EXT4_INODE_UNRM = 1, /* Undelete */
435 EXT4_INODE_COMPR = 2, /* Compress file */
436 EXT4_INODE_SYNC = 3, /* Synchronous updates */
437 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
438 EXT4_INODE_APPEND = 5, /* writes to file may only append */
439 EXT4_INODE_NODUMP = 6, /* do not dump file */
440 EXT4_INODE_NOATIME = 7, /* do not update atime */
441 /* Reserved for compression usage... */
442 EXT4_INODE_DIRTY = 8,
443 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
444 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
445 EXT4_INODE_ECOMPR = 11, /* Compression error */
446 /* End compression flags --- maybe not all used */
447 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
448 EXT4_INODE_IMAGIC = 13, /* AFS directory */
449 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
450 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
451 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
452 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
453 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
454 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
455 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
456 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
457 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
458 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
459 };
460
461 /*
462 * Since it's pretty easy to mix up bit numbers and hex values, we use a
463 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
464 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
465 * any extra space in the compiled kernel image, otherwise, the build will fail.
466 * It's important that these values are the same, since we are using
467 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
468 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
469 * values found in ext2, ext3 and ext4 filesystems, and of course the values
470 * defined in e2fsprogs.
471 *
472 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
473 */
474 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
475 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
476
477 static inline void ext4_check_flag_values(void)
478 {
479 CHECK_FLAG_VALUE(SECRM);
480 CHECK_FLAG_VALUE(UNRM);
481 CHECK_FLAG_VALUE(COMPR);
482 CHECK_FLAG_VALUE(SYNC);
483 CHECK_FLAG_VALUE(IMMUTABLE);
484 CHECK_FLAG_VALUE(APPEND);
485 CHECK_FLAG_VALUE(NODUMP);
486 CHECK_FLAG_VALUE(NOATIME);
487 CHECK_FLAG_VALUE(DIRTY);
488 CHECK_FLAG_VALUE(COMPRBLK);
489 CHECK_FLAG_VALUE(NOCOMPR);
490 CHECK_FLAG_VALUE(ECOMPR);
491 CHECK_FLAG_VALUE(INDEX);
492 CHECK_FLAG_VALUE(IMAGIC);
493 CHECK_FLAG_VALUE(JOURNAL_DATA);
494 CHECK_FLAG_VALUE(NOTAIL);
495 CHECK_FLAG_VALUE(DIRSYNC);
496 CHECK_FLAG_VALUE(TOPDIR);
497 CHECK_FLAG_VALUE(HUGE_FILE);
498 CHECK_FLAG_VALUE(EXTENTS);
499 CHECK_FLAG_VALUE(EA_INODE);
500 CHECK_FLAG_VALUE(EOFBLOCKS);
501 CHECK_FLAG_VALUE(INLINE_DATA);
502 CHECK_FLAG_VALUE(RESERVED);
503 }
504
505 /* Used to pass group descriptor data when online resize is done */
506 struct ext4_new_group_input {
507 __u32 group; /* Group number for this data */
508 __u64 block_bitmap; /* Absolute block number of block bitmap */
509 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
510 __u64 inode_table; /* Absolute block number of inode table start */
511 __u32 blocks_count; /* Total number of blocks in this group */
512 __u16 reserved_blocks; /* Number of reserved blocks in this group */
513 __u16 unused;
514 };
515
516 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
517 struct compat_ext4_new_group_input {
518 u32 group;
519 compat_u64 block_bitmap;
520 compat_u64 inode_bitmap;
521 compat_u64 inode_table;
522 u32 blocks_count;
523 u16 reserved_blocks;
524 u16 unused;
525 };
526 #endif
527
528 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
529 struct ext4_new_group_data {
530 __u32 group;
531 __u64 block_bitmap;
532 __u64 inode_bitmap;
533 __u64 inode_table;
534 __u32 blocks_count;
535 __u16 reserved_blocks;
536 __u16 unused;
537 __u32 free_blocks_count;
538 };
539
540 /* Indexes used to index group tables in ext4_new_group_data */
541 enum {
542 BLOCK_BITMAP = 0, /* block bitmap */
543 INODE_BITMAP, /* inode bitmap */
544 INODE_TABLE, /* inode tables */
545 GROUP_TABLE_COUNT,
546 };
547
548 /*
549 * Flags used by ext4_map_blocks()
550 */
551 /* Allocate any needed blocks and/or convert an unitialized
552 extent to be an initialized ext4 */
553 #define EXT4_GET_BLOCKS_CREATE 0x0001
554 /* Request the creation of an unitialized extent */
555 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
556 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
557 EXT4_GET_BLOCKS_CREATE)
558 /* Caller is from the delayed allocation writeout path
559 * finally doing the actual allocation of delayed blocks */
560 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
561 /* caller is from the direct IO path, request to creation of an
562 unitialized extents if not allocated, split the uninitialized
563 extent if blocks has been preallocated already*/
564 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
565 #define EXT4_GET_BLOCKS_CONVERT 0x0010
566 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
567 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
568 /* Convert extent to initialized after IO complete */
569 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
570 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
571 /* Eventual metadata allocation (due to growing extent tree)
572 * should not fail, so try to use reserved blocks for that.*/
573 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
574 /* Don't normalize allocation size (used for fallocate) */
575 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
576 /* Request will not result in inode size update (user for fallocate) */
577 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
578 /* Do not take i_data_sem locking in ext4_map_blocks */
579 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
580 /* Do not put hole in extent cache */
581 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
582
583 /*
584 * Flags used by ext4_free_blocks
585 */
586 #define EXT4_FREE_BLOCKS_METADATA 0x0001
587 #define EXT4_FREE_BLOCKS_FORGET 0x0002
588 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
589 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
590 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
591 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
592 #define EXT4_FREE_BLOCKS_RESERVE 0x0040
593
594 /*
595 * Flags used by ext4_discard_partial_page_buffers
596 */
597 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
598
599 /*
600 * ioctl commands
601 */
602 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
603 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
604 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
605 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
606 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
607 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
608 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
609 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
610 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
611 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
612 #define EXT4_IOC_MIGRATE _IO('f', 9)
613 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
614 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
615 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
616 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
617 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
618 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
619
620 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
621 /*
622 * ioctl commands in 32 bit emulation
623 */
624 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
625 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
626 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
627 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
628 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
629 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
630 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
631 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
632 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
633 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
634 #endif
635
636 /* Max physical block we can address w/o extents */
637 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
638
639 /*
640 * Structure of an inode on the disk
641 */
642 struct ext4_inode {
643 __le16 i_mode; /* File mode */
644 __le16 i_uid; /* Low 16 bits of Owner Uid */
645 __le32 i_size_lo; /* Size in bytes */
646 __le32 i_atime; /* Access time */
647 __le32 i_ctime; /* Inode Change time */
648 __le32 i_mtime; /* Modification time */
649 __le32 i_dtime; /* Deletion Time */
650 __le16 i_gid; /* Low 16 bits of Group Id */
651 __le16 i_links_count; /* Links count */
652 __le32 i_blocks_lo; /* Blocks count */
653 __le32 i_flags; /* File flags */
654 union {
655 struct {
656 __le32 l_i_version;
657 } linux1;
658 struct {
659 __u32 h_i_translator;
660 } hurd1;
661 struct {
662 __u32 m_i_reserved1;
663 } masix1;
664 } osd1; /* OS dependent 1 */
665 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
666 __le32 i_generation; /* File version (for NFS) */
667 __le32 i_file_acl_lo; /* File ACL */
668 __le32 i_size_high;
669 __le32 i_obso_faddr; /* Obsoleted fragment address */
670 union {
671 struct {
672 __le16 l_i_blocks_high; /* were l_i_reserved1 */
673 __le16 l_i_file_acl_high;
674 __le16 l_i_uid_high; /* these 2 fields */
675 __le16 l_i_gid_high; /* were reserved2[0] */
676 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
677 __le16 l_i_reserved;
678 } linux2;
679 struct {
680 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
681 __u16 h_i_mode_high;
682 __u16 h_i_uid_high;
683 __u16 h_i_gid_high;
684 __u32 h_i_author;
685 } hurd2;
686 struct {
687 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
688 __le16 m_i_file_acl_high;
689 __u32 m_i_reserved2[2];
690 } masix2;
691 } osd2; /* OS dependent 2 */
692 __le16 i_extra_isize;
693 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
694 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
695 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
696 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
697 __le32 i_crtime; /* File Creation time */
698 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
699 __le32 i_version_hi; /* high 32 bits for 64-bit version */
700 };
701
702 struct move_extent {
703 __u32 reserved; /* should be zero */
704 __u32 donor_fd; /* donor file descriptor */
705 __u64 orig_start; /* logical start offset in block for orig */
706 __u64 donor_start; /* logical start offset in block for donor */
707 __u64 len; /* block length to be moved */
708 __u64 moved_len; /* moved block length */
709 };
710
711 #define EXT4_EPOCH_BITS 2
712 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
713 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
714
715 /*
716 * Extended fields will fit into an inode if the filesystem was formatted
717 * with large inodes (-I 256 or larger) and there are not currently any EAs
718 * consuming all of the available space. For new inodes we always reserve
719 * enough space for the kernel's known extended fields, but for inodes
720 * created with an old kernel this might not have been the case. None of
721 * the extended inode fields is critical for correct filesystem operation.
722 * This macro checks if a certain field fits in the inode. Note that
723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
724 */
725 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
726 ((offsetof(typeof(*ext4_inode), field) + \
727 sizeof((ext4_inode)->field)) \
728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
729 (einode)->i_extra_isize)) \
730
731 static inline __le32 ext4_encode_extra_time(struct timespec *time)
732 {
733 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
734 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
735 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
736 }
737
738 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
739 {
740 if (sizeof(time->tv_sec) > 4)
741 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
742 << 32;
743 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
744 }
745
746 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
747 do { \
748 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
749 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
750 (raw_inode)->xtime ## _extra = \
751 ext4_encode_extra_time(&(inode)->xtime); \
752 } while (0)
753
754 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
755 do { \
756 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
757 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
758 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
759 (raw_inode)->xtime ## _extra = \
760 ext4_encode_extra_time(&(einode)->xtime); \
761 } while (0)
762
763 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
764 do { \
765 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
766 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
767 ext4_decode_extra_time(&(inode)->xtime, \
768 raw_inode->xtime ## _extra); \
769 else \
770 (inode)->xtime.tv_nsec = 0; \
771 } while (0)
772
773 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
774 do { \
775 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
776 (einode)->xtime.tv_sec = \
777 (signed)le32_to_cpu((raw_inode)->xtime); \
778 else \
779 (einode)->xtime.tv_sec = 0; \
780 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
781 ext4_decode_extra_time(&(einode)->xtime, \
782 raw_inode->xtime ## _extra); \
783 else \
784 (einode)->xtime.tv_nsec = 0; \
785 } while (0)
786
787 #define i_disk_version osd1.linux1.l_i_version
788
789 #if defined(__KERNEL__) || defined(__linux__)
790 #define i_reserved1 osd1.linux1.l_i_reserved1
791 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
792 #define i_blocks_high osd2.linux2.l_i_blocks_high
793 #define i_uid_low i_uid
794 #define i_gid_low i_gid
795 #define i_uid_high osd2.linux2.l_i_uid_high
796 #define i_gid_high osd2.linux2.l_i_gid_high
797 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
798
799 #elif defined(__GNU__)
800
801 #define i_translator osd1.hurd1.h_i_translator
802 #define i_uid_high osd2.hurd2.h_i_uid_high
803 #define i_gid_high osd2.hurd2.h_i_gid_high
804 #define i_author osd2.hurd2.h_i_author
805
806 #elif defined(__masix__)
807
808 #define i_reserved1 osd1.masix1.m_i_reserved1
809 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
810 #define i_reserved2 osd2.masix2.m_i_reserved2
811
812 #endif /* defined(__KERNEL__) || defined(__linux__) */
813
814 #include "extents_status.h"
815
816 /*
817 * fourth extended file system inode data in memory
818 */
819 struct ext4_inode_info {
820 __le32 i_data[15]; /* unconverted */
821 __u32 i_dtime;
822 ext4_fsblk_t i_file_acl;
823
824 /*
825 * i_block_group is the number of the block group which contains
826 * this file's inode. Constant across the lifetime of the inode,
827 * it is ued for making block allocation decisions - we try to
828 * place a file's data blocks near its inode block, and new inodes
829 * near to their parent directory's inode.
830 */
831 ext4_group_t i_block_group;
832 ext4_lblk_t i_dir_start_lookup;
833 #if (BITS_PER_LONG < 64)
834 unsigned long i_state_flags; /* Dynamic state flags */
835 #endif
836 unsigned long i_flags;
837
838 /*
839 * Extended attributes can be read independently of the main file
840 * data. Taking i_mutex even when reading would cause contention
841 * between readers of EAs and writers of regular file data, so
842 * instead we synchronize on xattr_sem when reading or changing
843 * EAs.
844 */
845 struct rw_semaphore xattr_sem;
846
847 struct list_head i_orphan; /* unlinked but open inodes */
848
849 /*
850 * i_disksize keeps track of what the inode size is ON DISK, not
851 * in memory. During truncate, i_size is set to the new size by
852 * the VFS prior to calling ext4_truncate(), but the filesystem won't
853 * set i_disksize to 0 until the truncate is actually under way.
854 *
855 * The intent is that i_disksize always represents the blocks which
856 * are used by this file. This allows recovery to restart truncate
857 * on orphans if we crash during truncate. We actually write i_disksize
858 * into the on-disk inode when writing inodes out, instead of i_size.
859 *
860 * The only time when i_disksize and i_size may be different is when
861 * a truncate is in progress. The only things which change i_disksize
862 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
863 */
864 loff_t i_disksize;
865
866 /*
867 * i_data_sem is for serialising ext4_truncate() against
868 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
869 * data tree are chopped off during truncate. We can't do that in
870 * ext4 because whenever we perform intermediate commits during
871 * truncate, the inode and all the metadata blocks *must* be in a
872 * consistent state which allows truncation of the orphans to restart
873 * during recovery. Hence we must fix the get_block-vs-truncate race
874 * by other means, so we have i_data_sem.
875 */
876 struct rw_semaphore i_data_sem;
877 struct inode vfs_inode;
878 struct jbd2_inode *jinode;
879
880 /*
881 * File creation time. Its function is same as that of
882 * struct timespec i_{a,c,m}time in the generic inode.
883 */
884 struct timespec i_crtime;
885
886 /* mballoc */
887 struct list_head i_prealloc_list;
888 spinlock_t i_prealloc_lock;
889
890 /* extents status tree */
891 struct ext4_es_tree i_es_tree;
892 rwlock_t i_es_lock;
893 struct list_head i_es_lru;
894 unsigned int i_es_lru_nr; /* protected by i_es_lock */
895
896 /* ialloc */
897 ext4_group_t i_last_alloc_group;
898
899 /* allocation reservation info for delalloc */
900 /* In case of bigalloc, these refer to clusters rather than blocks */
901 unsigned int i_reserved_data_blocks;
902 unsigned int i_reserved_meta_blocks;
903 unsigned int i_allocated_meta_blocks;
904 ext4_lblk_t i_da_metadata_calc_last_lblock;
905 int i_da_metadata_calc_len;
906
907 /* on-disk additional length */
908 __u16 i_extra_isize;
909
910 /* Indicate the inline data space. */
911 u16 i_inline_off;
912 u16 i_inline_size;
913
914 #ifdef CONFIG_QUOTA
915 /* quota space reservation, managed internally by quota code */
916 qsize_t i_reserved_quota;
917 #endif
918
919 /* completed IOs that might need unwritten extents handling */
920 struct list_head i_completed_io_list;
921 spinlock_t i_completed_io_lock;
922 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
923 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
924 struct work_struct i_unwritten_work; /* deferred extent conversion */
925
926 spinlock_t i_block_reservation_lock;
927
928 /*
929 * Transactions that contain inode's metadata needed to complete
930 * fsync and fdatasync, respectively.
931 */
932 tid_t i_sync_tid;
933 tid_t i_datasync_tid;
934
935 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
936 __u32 i_csum_seed;
937 };
938
939 /*
940 * File system states
941 */
942 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
943 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
944 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
945
946 /*
947 * Misc. filesystem flags
948 */
949 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
950 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
951 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
952
953 /*
954 * Mount flags set via mount options or defaults
955 */
956 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
957 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
958 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
959 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
960 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
961 #define EXT4_MOUNT_ERRORS_MASK 0x00070
962 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
963 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
964 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
965 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
966 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
967 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
968 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
969 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
970 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
971 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
972 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
973 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
974 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
975 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
976 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
977 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
978 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
979 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
980 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
981 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
982 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
983 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
984 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
985
986 /*
987 * Mount flags set either automatically (could not be set by mount option)
988 * based on per file system feature or property or in special cases such as
989 * distinguishing between explicit mount option definition and default.
990 */
991 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
992 specified delalloc */
993 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
994 size of blocksize * 8
995 blocks */
996
997 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
998 ~EXT4_MOUNT_##opt
999 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1000 EXT4_MOUNT_##opt
1001 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1002 EXT4_MOUNT_##opt)
1003
1004 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1005 ~EXT4_MOUNT2_##opt
1006 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1007 EXT4_MOUNT2_##opt
1008 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1009 EXT4_MOUNT2_##opt)
1010
1011 #define ext4_test_and_set_bit __test_and_set_bit_le
1012 #define ext4_set_bit __set_bit_le
1013 #define ext4_set_bit_atomic ext2_set_bit_atomic
1014 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1015 #define ext4_clear_bit __clear_bit_le
1016 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1017 #define ext4_test_bit test_bit_le
1018 #define ext4_find_next_zero_bit find_next_zero_bit_le
1019 #define ext4_find_next_bit find_next_bit_le
1020
1021 extern void ext4_set_bits(void *bm, int cur, int len);
1022
1023 /*
1024 * Maximal mount counts between two filesystem checks
1025 */
1026 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1027 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1028
1029 /*
1030 * Behaviour when detecting errors
1031 */
1032 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1033 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1034 #define EXT4_ERRORS_PANIC 3 /* Panic */
1035 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1036
1037 /* Metadata checksum algorithm codes */
1038 #define EXT4_CRC32C_CHKSUM 1
1039
1040 /*
1041 * Structure of the super block
1042 */
1043 struct ext4_super_block {
1044 /*00*/ __le32 s_inodes_count; /* Inodes count */
1045 __le32 s_blocks_count_lo; /* Blocks count */
1046 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1047 __le32 s_free_blocks_count_lo; /* Free blocks count */
1048 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1049 __le32 s_first_data_block; /* First Data Block */
1050 __le32 s_log_block_size; /* Block size */
1051 __le32 s_log_cluster_size; /* Allocation cluster size */
1052 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1053 __le32 s_clusters_per_group; /* # Clusters per group */
1054 __le32 s_inodes_per_group; /* # Inodes per group */
1055 __le32 s_mtime; /* Mount time */
1056 /*30*/ __le32 s_wtime; /* Write time */
1057 __le16 s_mnt_count; /* Mount count */
1058 __le16 s_max_mnt_count; /* Maximal mount count */
1059 __le16 s_magic; /* Magic signature */
1060 __le16 s_state; /* File system state */
1061 __le16 s_errors; /* Behaviour when detecting errors */
1062 __le16 s_minor_rev_level; /* minor revision level */
1063 /*40*/ __le32 s_lastcheck; /* time of last check */
1064 __le32 s_checkinterval; /* max. time between checks */
1065 __le32 s_creator_os; /* OS */
1066 __le32 s_rev_level; /* Revision level */
1067 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1068 __le16 s_def_resgid; /* Default gid for reserved blocks */
1069 /*
1070 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1071 *
1072 * Note: the difference between the compatible feature set and
1073 * the incompatible feature set is that if there is a bit set
1074 * in the incompatible feature set that the kernel doesn't
1075 * know about, it should refuse to mount the filesystem.
1076 *
1077 * e2fsck's requirements are more strict; if it doesn't know
1078 * about a feature in either the compatible or incompatible
1079 * feature set, it must abort and not try to meddle with
1080 * things it doesn't understand...
1081 */
1082 __le32 s_first_ino; /* First non-reserved inode */
1083 __le16 s_inode_size; /* size of inode structure */
1084 __le16 s_block_group_nr; /* block group # of this superblock */
1085 __le32 s_feature_compat; /* compatible feature set */
1086 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1087 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1088 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1089 /*78*/ char s_volume_name[16]; /* volume name */
1090 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1091 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1092 /*
1093 * Performance hints. Directory preallocation should only
1094 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1095 */
1096 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1097 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1098 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1099 /*
1100 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1101 */
1102 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1103 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1104 __le32 s_journal_dev; /* device number of journal file */
1105 __le32 s_last_orphan; /* start of list of inodes to delete */
1106 __le32 s_hash_seed[4]; /* HTREE hash seed */
1107 __u8 s_def_hash_version; /* Default hash version to use */
1108 __u8 s_jnl_backup_type;
1109 __le16 s_desc_size; /* size of group descriptor */
1110 /*100*/ __le32 s_default_mount_opts;
1111 __le32 s_first_meta_bg; /* First metablock block group */
1112 __le32 s_mkfs_time; /* When the filesystem was created */
1113 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1114 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1115 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1116 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1117 __le32 s_free_blocks_count_hi; /* Free blocks count */
1118 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1119 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1120 __le32 s_flags; /* Miscellaneous flags */
1121 __le16 s_raid_stride; /* RAID stride */
1122 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1123 __le64 s_mmp_block; /* Block for multi-mount protection */
1124 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1125 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1126 __u8 s_checksum_type; /* metadata checksum algorithm used */
1127 __le16 s_reserved_pad;
1128 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1129 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1130 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1131 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1132 snapshot's future use */
1133 __le32 s_snapshot_list; /* inode number of the head of the
1134 on-disk snapshot list */
1135 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1136 __le32 s_error_count; /* number of fs errors */
1137 __le32 s_first_error_time; /* first time an error happened */
1138 __le32 s_first_error_ino; /* inode involved in first error */
1139 __le64 s_first_error_block; /* block involved of first error */
1140 __u8 s_first_error_func[32]; /* function where the error happened */
1141 __le32 s_first_error_line; /* line number where error happened */
1142 __le32 s_last_error_time; /* most recent time of an error */
1143 __le32 s_last_error_ino; /* inode involved in last error */
1144 __le32 s_last_error_line; /* line number where error happened */
1145 __le64 s_last_error_block; /* block involved of last error */
1146 __u8 s_last_error_func[32]; /* function where the error happened */
1147 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1148 __u8 s_mount_opts[64];
1149 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1150 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1151 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1152 __le32 s_reserved[108]; /* Padding to the end of the block */
1153 __le32 s_checksum; /* crc32c(superblock) */
1154 };
1155
1156 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1157
1158 #ifdef __KERNEL__
1159
1160 /*
1161 * run-time mount flags
1162 */
1163 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1164 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1165
1166 /*
1167 * fourth extended-fs super-block data in memory
1168 */
1169 struct ext4_sb_info {
1170 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1171 unsigned long s_inodes_per_block;/* Number of inodes per block */
1172 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1173 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1174 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1175 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1176 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1177 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1178 ext4_group_t s_groups_count; /* Number of groups in the fs */
1179 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1180 unsigned long s_overhead; /* # of fs overhead clusters */
1181 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1182 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1183 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1184 struct buffer_head * s_sbh; /* Buffer containing the super block */
1185 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1186 struct buffer_head **s_group_desc;
1187 unsigned int s_mount_opt;
1188 unsigned int s_mount_opt2;
1189 unsigned int s_mount_flags;
1190 unsigned int s_def_mount_opt;
1191 ext4_fsblk_t s_sb_block;
1192 atomic64_t s_resv_clusters;
1193 kuid_t s_resuid;
1194 kgid_t s_resgid;
1195 unsigned short s_mount_state;
1196 unsigned short s_pad;
1197 int s_addr_per_block_bits;
1198 int s_desc_per_block_bits;
1199 int s_inode_size;
1200 int s_first_ino;
1201 unsigned int s_inode_readahead_blks;
1202 unsigned int s_inode_goal;
1203 spinlock_t s_next_gen_lock;
1204 u32 s_next_generation;
1205 u32 s_hash_seed[4];
1206 int s_def_hash_version;
1207 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1208 struct percpu_counter s_freeclusters_counter;
1209 struct percpu_counter s_freeinodes_counter;
1210 struct percpu_counter s_dirs_counter;
1211 struct percpu_counter s_dirtyclusters_counter;
1212 struct blockgroup_lock *s_blockgroup_lock;
1213 struct proc_dir_entry *s_proc;
1214 struct kobject s_kobj;
1215 struct completion s_kobj_unregister;
1216 struct super_block *s_sb;
1217
1218 /* Journaling */
1219 struct journal_s *s_journal;
1220 struct list_head s_orphan;
1221 struct mutex s_orphan_lock;
1222 unsigned long s_resize_flags; /* Flags indicating if there
1223 is a resizer */
1224 unsigned long s_commit_interval;
1225 u32 s_max_batch_time;
1226 u32 s_min_batch_time;
1227 struct block_device *journal_bdev;
1228 #ifdef CONFIG_QUOTA
1229 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1230 int s_jquota_fmt; /* Format of quota to use */
1231 #endif
1232 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1233 struct rb_root system_blks;
1234
1235 #ifdef EXTENTS_STATS
1236 /* ext4 extents stats */
1237 unsigned long s_ext_min;
1238 unsigned long s_ext_max;
1239 unsigned long s_depth_max;
1240 spinlock_t s_ext_stats_lock;
1241 unsigned long s_ext_blocks;
1242 unsigned long s_ext_extents;
1243 #endif
1244
1245 /* for buddy allocator */
1246 struct ext4_group_info ***s_group_info;
1247 struct inode *s_buddy_cache;
1248 spinlock_t s_md_lock;
1249 unsigned short *s_mb_offsets;
1250 unsigned int *s_mb_maxs;
1251 unsigned int s_group_info_size;
1252
1253 /* tunables */
1254 unsigned long s_stripe;
1255 unsigned int s_mb_stream_request;
1256 unsigned int s_mb_max_to_scan;
1257 unsigned int s_mb_min_to_scan;
1258 unsigned int s_mb_stats;
1259 unsigned int s_mb_order2_reqs;
1260 unsigned int s_mb_group_prealloc;
1261 unsigned int s_max_writeback_mb_bump;
1262 unsigned int s_max_dir_size_kb;
1263 /* where last allocation was done - for stream allocation */
1264 unsigned long s_mb_last_group;
1265 unsigned long s_mb_last_start;
1266
1267 /* stats for buddy allocator */
1268 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1269 atomic_t s_bal_success; /* we found long enough chunks */
1270 atomic_t s_bal_allocated; /* in blocks */
1271 atomic_t s_bal_ex_scanned; /* total extents scanned */
1272 atomic_t s_bal_goals; /* goal hits */
1273 atomic_t s_bal_breaks; /* too long searches */
1274 atomic_t s_bal_2orders; /* 2^order hits */
1275 spinlock_t s_bal_lock;
1276 unsigned long s_mb_buddies_generated;
1277 unsigned long long s_mb_generation_time;
1278 atomic_t s_mb_lost_chunks;
1279 atomic_t s_mb_preallocated;
1280 atomic_t s_mb_discarded;
1281 atomic_t s_lock_busy;
1282
1283 /* locality groups */
1284 struct ext4_locality_group __percpu *s_locality_groups;
1285
1286 /* for write statistics */
1287 unsigned long s_sectors_written_start;
1288 u64 s_kbytes_written;
1289
1290 /* the size of zero-out chunk */
1291 unsigned int s_extent_max_zeroout_kb;
1292
1293 unsigned int s_log_groups_per_flex;
1294 struct flex_groups *s_flex_groups;
1295 ext4_group_t s_flex_groups_allocated;
1296
1297 /* workqueue for dio unwritten */
1298 struct workqueue_struct *dio_unwritten_wq;
1299
1300 /* timer for periodic error stats printing */
1301 struct timer_list s_err_report;
1302
1303 /* Lazy inode table initialization info */
1304 struct ext4_li_request *s_li_request;
1305 /* Wait multiplier for lazy initialization thread */
1306 unsigned int s_li_wait_mult;
1307
1308 /* Kernel thread for multiple mount protection */
1309 struct task_struct *s_mmp_tsk;
1310
1311 /* record the last minlen when FITRIM is called. */
1312 atomic_t s_last_trim_minblks;
1313
1314 /* Reference to checksum algorithm driver via cryptoapi */
1315 struct crypto_shash *s_chksum_driver;
1316
1317 /* Precomputed FS UUID checksum for seeding other checksums */
1318 __u32 s_csum_seed;
1319
1320 /* Reclaim extents from extent status tree */
1321 struct shrinker s_es_shrinker;
1322 struct list_head s_es_lru;
1323 struct percpu_counter s_extent_cache_cnt;
1324 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1325 };
1326
1327 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1328 {
1329 return sb->s_fs_info;
1330 }
1331 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1332 {
1333 return container_of(inode, struct ext4_inode_info, vfs_inode);
1334 }
1335
1336 static inline struct timespec ext4_current_time(struct inode *inode)
1337 {
1338 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1339 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1340 }
1341
1342 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1343 {
1344 return ino == EXT4_ROOT_INO ||
1345 ino == EXT4_USR_QUOTA_INO ||
1346 ino == EXT4_GRP_QUOTA_INO ||
1347 ino == EXT4_BOOT_LOADER_INO ||
1348 ino == EXT4_JOURNAL_INO ||
1349 ino == EXT4_RESIZE_INO ||
1350 (ino >= EXT4_FIRST_INO(sb) &&
1351 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1352 }
1353
1354 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1355 struct ext4_io_end *io_end)
1356 {
1357 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1358 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1359 atomic_inc(&EXT4_I(inode)->i_unwritten);
1360 }
1361 }
1362
1363 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1364 {
1365 return inode->i_private;
1366 }
1367
1368 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1369 {
1370 inode->i_private = io;
1371 }
1372
1373 /*
1374 * Inode dynamic state flags
1375 */
1376 enum {
1377 EXT4_STATE_JDATA, /* journaled data exists */
1378 EXT4_STATE_NEW, /* inode is newly created */
1379 EXT4_STATE_XATTR, /* has in-inode xattrs */
1380 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1381 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1382 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1383 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1384 EXT4_STATE_NEWENTRY, /* File just added to dir */
1385 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1386 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1387 nolocking */
1388 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1389 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1390 };
1391
1392 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1393 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1394 { \
1395 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1396 } \
1397 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1398 { \
1399 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1400 } \
1401 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1402 { \
1403 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1404 }
1405
1406 EXT4_INODE_BIT_FNS(flag, flags, 0)
1407 #if (BITS_PER_LONG < 64)
1408 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1409
1410 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1411 {
1412 (ei)->i_state_flags = 0;
1413 }
1414 #else
1415 EXT4_INODE_BIT_FNS(state, flags, 32)
1416
1417 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1418 {
1419 /* We depend on the fact that callers will set i_flags */
1420 }
1421 #endif
1422 #else
1423 /* Assume that user mode programs are passing in an ext4fs superblock, not
1424 * a kernel struct super_block. This will allow us to call the feature-test
1425 * macros from user land. */
1426 #define EXT4_SB(sb) (sb)
1427 #endif
1428
1429 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1430
1431 /*
1432 * Codes for operating systems
1433 */
1434 #define EXT4_OS_LINUX 0
1435 #define EXT4_OS_HURD 1
1436 #define EXT4_OS_MASIX 2
1437 #define EXT4_OS_FREEBSD 3
1438 #define EXT4_OS_LITES 4
1439
1440 /*
1441 * Revision levels
1442 */
1443 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1444 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1445
1446 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1447 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1448
1449 #define EXT4_GOOD_OLD_INODE_SIZE 128
1450
1451 /*
1452 * Feature set definitions
1453 */
1454
1455 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1456 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1457 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1458 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1459 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1460 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1461 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1462 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1463 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1464 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1465 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1466 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1467 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1468 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1469 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1470 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1471 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1472 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1473
1474 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1475 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1476 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1477 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1478 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1479 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1480
1481 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1482 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1483 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1484 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1485 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1486 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1487 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1488 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1489 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1490 /*
1491 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1492 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1493 * all other data structures' checksums. However, the METADATA_CSUM and
1494 * GDT_CSUM bits are mutually exclusive.
1495 */
1496 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1497
1498 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1499 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1500 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1501 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1502 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1503 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1504 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1505 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1506 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1507 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1508 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1509 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1510 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1511 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1512
1513 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1514 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1515 EXT4_FEATURE_INCOMPAT_META_BG)
1516 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1517 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1518 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1519
1520 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1521 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1522 EXT4_FEATURE_INCOMPAT_RECOVER| \
1523 EXT4_FEATURE_INCOMPAT_META_BG)
1524 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1525 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1526 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1527
1528 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1529 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1530 EXT4_FEATURE_INCOMPAT_RECOVER| \
1531 EXT4_FEATURE_INCOMPAT_META_BG| \
1532 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1533 EXT4_FEATURE_INCOMPAT_64BIT| \
1534 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1535 EXT4_FEATURE_INCOMPAT_MMP | \
1536 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1537 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1538 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1539 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1540 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1541 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1542 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1543 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1544 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1545 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1546 EXT4_FEATURE_RO_COMPAT_QUOTA)
1547
1548 /*
1549 * Default values for user and/or group using reserved blocks
1550 */
1551 #define EXT4_DEF_RESUID 0
1552 #define EXT4_DEF_RESGID 0
1553
1554 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1555
1556 /*
1557 * Default mount options
1558 */
1559 #define EXT4_DEFM_DEBUG 0x0001
1560 #define EXT4_DEFM_BSDGROUPS 0x0002
1561 #define EXT4_DEFM_XATTR_USER 0x0004
1562 #define EXT4_DEFM_ACL 0x0008
1563 #define EXT4_DEFM_UID16 0x0010
1564 #define EXT4_DEFM_JMODE 0x0060
1565 #define EXT4_DEFM_JMODE_DATA 0x0020
1566 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1567 #define EXT4_DEFM_JMODE_WBACK 0x0060
1568 #define EXT4_DEFM_NOBARRIER 0x0100
1569 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1570 #define EXT4_DEFM_DISCARD 0x0400
1571 #define EXT4_DEFM_NODELALLOC 0x0800
1572
1573 /*
1574 * Default journal batch times
1575 */
1576 #define EXT4_DEF_MIN_BATCH_TIME 0
1577 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1578
1579 /*
1580 * Minimum number of groups in a flexgroup before we separate out
1581 * directories into the first block group of a flexgroup
1582 */
1583 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1584
1585 /*
1586 * Structure of a directory entry
1587 */
1588 #define EXT4_NAME_LEN 255
1589
1590 struct ext4_dir_entry {
1591 __le32 inode; /* Inode number */
1592 __le16 rec_len; /* Directory entry length */
1593 __le16 name_len; /* Name length */
1594 char name[EXT4_NAME_LEN]; /* File name */
1595 };
1596
1597 /*
1598 * The new version of the directory entry. Since EXT4 structures are
1599 * stored in intel byte order, and the name_len field could never be
1600 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1601 * file_type field.
1602 */
1603 struct ext4_dir_entry_2 {
1604 __le32 inode; /* Inode number */
1605 __le16 rec_len; /* Directory entry length */
1606 __u8 name_len; /* Name length */
1607 __u8 file_type;
1608 char name[EXT4_NAME_LEN]; /* File name */
1609 };
1610
1611 /*
1612 * This is a bogus directory entry at the end of each leaf block that
1613 * records checksums.
1614 */
1615 struct ext4_dir_entry_tail {
1616 __le32 det_reserved_zero1; /* Pretend to be unused */
1617 __le16 det_rec_len; /* 12 */
1618 __u8 det_reserved_zero2; /* Zero name length */
1619 __u8 det_reserved_ft; /* 0xDE, fake file type */
1620 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1621 };
1622
1623 #define EXT4_DIRENT_TAIL(block, blocksize) \
1624 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1625 ((blocksize) - \
1626 sizeof(struct ext4_dir_entry_tail))))
1627
1628 /*
1629 * Ext4 directory file types. Only the low 3 bits are used. The
1630 * other bits are reserved for now.
1631 */
1632 #define EXT4_FT_UNKNOWN 0
1633 #define EXT4_FT_REG_FILE 1
1634 #define EXT4_FT_DIR 2
1635 #define EXT4_FT_CHRDEV 3
1636 #define EXT4_FT_BLKDEV 4
1637 #define EXT4_FT_FIFO 5
1638 #define EXT4_FT_SOCK 6
1639 #define EXT4_FT_SYMLINK 7
1640
1641 #define EXT4_FT_MAX 8
1642
1643 #define EXT4_FT_DIR_CSUM 0xDE
1644
1645 /*
1646 * EXT4_DIR_PAD defines the directory entries boundaries
1647 *
1648 * NOTE: It must be a multiple of 4
1649 */
1650 #define EXT4_DIR_PAD 4
1651 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1652 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1653 ~EXT4_DIR_ROUND)
1654 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1655
1656 /*
1657 * If we ever get support for fs block sizes > page_size, we'll need
1658 * to remove the #if statements in the next two functions...
1659 */
1660 static inline unsigned int
1661 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1662 {
1663 unsigned len = le16_to_cpu(dlen);
1664
1665 #if (PAGE_CACHE_SIZE >= 65536)
1666 if (len == EXT4_MAX_REC_LEN || len == 0)
1667 return blocksize;
1668 return (len & 65532) | ((len & 3) << 16);
1669 #else
1670 return len;
1671 #endif
1672 }
1673
1674 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1675 {
1676 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1677 BUG();
1678 #if (PAGE_CACHE_SIZE >= 65536)
1679 if (len < 65536)
1680 return cpu_to_le16(len);
1681 if (len == blocksize) {
1682 if (blocksize == 65536)
1683 return cpu_to_le16(EXT4_MAX_REC_LEN);
1684 else
1685 return cpu_to_le16(0);
1686 }
1687 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1688 #else
1689 return cpu_to_le16(len);
1690 #endif
1691 }
1692
1693 /*
1694 * Hash Tree Directory indexing
1695 * (c) Daniel Phillips, 2001
1696 */
1697
1698 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1699 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1700 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1701 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1702 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1703
1704 /* Legal values for the dx_root hash_version field: */
1705
1706 #define DX_HASH_LEGACY 0
1707 #define DX_HASH_HALF_MD4 1
1708 #define DX_HASH_TEA 2
1709 #define DX_HASH_LEGACY_UNSIGNED 3
1710 #define DX_HASH_HALF_MD4_UNSIGNED 4
1711 #define DX_HASH_TEA_UNSIGNED 5
1712
1713 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1714 const void *address, unsigned int length)
1715 {
1716 struct {
1717 struct shash_desc shash;
1718 char ctx[4];
1719 } desc;
1720 int err;
1721
1722 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1723
1724 desc.shash.tfm = sbi->s_chksum_driver;
1725 desc.shash.flags = 0;
1726 *(u32 *)desc.ctx = crc;
1727
1728 err = crypto_shash_update(&desc.shash, address, length);
1729 BUG_ON(err);
1730
1731 return *(u32 *)desc.ctx;
1732 }
1733
1734 #ifdef __KERNEL__
1735
1736 /* hash info structure used by the directory hash */
1737 struct dx_hash_info
1738 {
1739 u32 hash;
1740 u32 minor_hash;
1741 int hash_version;
1742 u32 *seed;
1743 };
1744
1745
1746 /* 32 and 64 bit signed EOF for dx directories */
1747 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1748 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1749
1750
1751 /*
1752 * Control parameters used by ext4_htree_next_block
1753 */
1754 #define HASH_NB_ALWAYS 1
1755
1756
1757 /*
1758 * Describe an inode's exact location on disk and in memory
1759 */
1760 struct ext4_iloc
1761 {
1762 struct buffer_head *bh;
1763 unsigned long offset;
1764 ext4_group_t block_group;
1765 };
1766
1767 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1768 {
1769 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1770 }
1771
1772 /*
1773 * This structure is stuffed into the struct file's private_data field
1774 * for directories. It is where we put information so that we can do
1775 * readdir operations in hash tree order.
1776 */
1777 struct dir_private_info {
1778 struct rb_root root;
1779 struct rb_node *curr_node;
1780 struct fname *extra_fname;
1781 loff_t last_pos;
1782 __u32 curr_hash;
1783 __u32 curr_minor_hash;
1784 __u32 next_hash;
1785 };
1786
1787 /* calculate the first block number of the group */
1788 static inline ext4_fsblk_t
1789 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1790 {
1791 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1792 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1793 }
1794
1795 /*
1796 * Special error return code only used by dx_probe() and its callers.
1797 */
1798 #define ERR_BAD_DX_DIR -75000
1799
1800 /*
1801 * Timeout and state flag for lazy initialization inode thread.
1802 */
1803 #define EXT4_DEF_LI_WAIT_MULT 10
1804 #define EXT4_DEF_LI_MAX_START_DELAY 5
1805 #define EXT4_LAZYINIT_QUIT 0x0001
1806 #define EXT4_LAZYINIT_RUNNING 0x0002
1807
1808 /*
1809 * Lazy inode table initialization info
1810 */
1811 struct ext4_lazy_init {
1812 unsigned long li_state;
1813 struct list_head li_request_list;
1814 struct mutex li_list_mtx;
1815 };
1816
1817 struct ext4_li_request {
1818 struct super_block *lr_super;
1819 struct ext4_sb_info *lr_sbi;
1820 ext4_group_t lr_next_group;
1821 struct list_head lr_request;
1822 unsigned long lr_next_sched;
1823 unsigned long lr_timeout;
1824 };
1825
1826 struct ext4_features {
1827 struct kobject f_kobj;
1828 struct completion f_kobj_unregister;
1829 };
1830
1831 /*
1832 * This structure will be used for multiple mount protection. It will be
1833 * written into the block number saved in the s_mmp_block field in the
1834 * superblock. Programs that check MMP should assume that if
1835 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1836 * to use the filesystem, regardless of how old the timestamp is.
1837 */
1838 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1839 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1840 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1841 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1842
1843 struct mmp_struct {
1844 __le32 mmp_magic; /* Magic number for MMP */
1845 __le32 mmp_seq; /* Sequence no. updated periodically */
1846
1847 /*
1848 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1849 * purposes and do not affect the correctness of the algorithm
1850 */
1851 __le64 mmp_time; /* Time last updated */
1852 char mmp_nodename[64]; /* Node which last updated MMP block */
1853 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1854
1855 /*
1856 * mmp_check_interval is used to verify if the MMP block has been
1857 * updated on the block device. The value is updated based on the
1858 * maximum time to write the MMP block during an update cycle.
1859 */
1860 __le16 mmp_check_interval;
1861
1862 __le16 mmp_pad1;
1863 __le32 mmp_pad2[226];
1864 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1865 };
1866
1867 /* arguments passed to the mmp thread */
1868 struct mmpd_data {
1869 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1870 struct super_block *sb; /* super block of the fs */
1871 };
1872
1873 /*
1874 * Check interval multiplier
1875 * The MMP block is written every update interval and initially checked every
1876 * update interval x the multiplier (the value is then adapted based on the
1877 * write latency). The reason is that writes can be delayed under load and we
1878 * don't want readers to incorrectly assume that the filesystem is no longer
1879 * in use.
1880 */
1881 #define EXT4_MMP_CHECK_MULT 2UL
1882
1883 /*
1884 * Minimum interval for MMP checking in seconds.
1885 */
1886 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1887
1888 /*
1889 * Maximum interval for MMP checking in seconds.
1890 */
1891 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1892
1893 /*
1894 * Function prototypes
1895 */
1896
1897 /*
1898 * Ok, these declarations are also in <linux/kernel.h> but none of the
1899 * ext4 source programs needs to include it so they are duplicated here.
1900 */
1901 # define NORET_TYPE /**/
1902 # define ATTRIB_NORET __attribute__((noreturn))
1903 # define NORET_AND noreturn,
1904
1905 /* bitmap.c */
1906 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1907 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1908 struct ext4_group_desc *gdp,
1909 struct buffer_head *bh, int sz);
1910 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1911 struct ext4_group_desc *gdp,
1912 struct buffer_head *bh, int sz);
1913 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1914 struct ext4_group_desc *gdp,
1915 struct buffer_head *bh);
1916 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1917 struct ext4_group_desc *gdp,
1918 struct buffer_head *bh);
1919
1920 /* balloc.c */
1921 extern void ext4_get_group_no_and_offset(struct super_block *sb,
1922 ext4_fsblk_t blocknr,
1923 ext4_group_t *blockgrpp,
1924 ext4_grpblk_t *offsetp);
1925 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1926 ext4_fsblk_t block);
1927
1928 extern void ext4_validate_block_bitmap(struct super_block *sb,
1929 struct ext4_group_desc *desc,
1930 unsigned int block_group,
1931 struct buffer_head *bh);
1932 extern unsigned int ext4_block_group(struct super_block *sb,
1933 ext4_fsblk_t blocknr);
1934 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1935 ext4_fsblk_t blocknr);
1936 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1937 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1938 ext4_group_t group);
1939 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1940 ext4_fsblk_t goal,
1941 unsigned int flags,
1942 unsigned long *count,
1943 int *errp);
1944 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1945 s64 nclusters, unsigned int flags);
1946 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1947 extern void ext4_check_blocks_bitmap(struct super_block *);
1948 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1949 ext4_group_t block_group,
1950 struct buffer_head ** bh);
1951 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1952
1953 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1954 ext4_group_t block_group);
1955 extern int ext4_wait_block_bitmap(struct super_block *sb,
1956 ext4_group_t block_group,
1957 struct buffer_head *bh);
1958 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1959 ext4_group_t block_group);
1960 extern void ext4_init_block_bitmap(struct super_block *sb,
1961 struct buffer_head *bh,
1962 ext4_group_t group,
1963 struct ext4_group_desc *desc);
1964 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1965 ext4_group_t block_group,
1966 struct ext4_group_desc *gdp);
1967 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1968 ext4_group_t block_group,
1969 struct ext4_group_desc *gdp);
1970 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1971
1972 /* dir.c */
1973 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1974 struct file *,
1975 struct ext4_dir_entry_2 *,
1976 struct buffer_head *, char *, int,
1977 unsigned int);
1978 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1979 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1980 (de), (bh), (buf), (size), (offset)))
1981 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1982 __u32 minor_hash,
1983 struct ext4_dir_entry_2 *dirent);
1984 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1985 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1986 struct buffer_head *bh,
1987 void *buf, int buf_size,
1988 const char *name, int namelen,
1989 struct ext4_dir_entry_2 **dest_de);
1990 void ext4_insert_dentry(struct inode *inode,
1991 struct ext4_dir_entry_2 *de,
1992 int buf_size,
1993 const char *name, int namelen);
1994 static inline void ext4_update_dx_flag(struct inode *inode)
1995 {
1996 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1997 EXT4_FEATURE_COMPAT_DIR_INDEX))
1998 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1999 }
2000 static unsigned char ext4_filetype_table[] = {
2001 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2002 };
2003
2004 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2005 {
2006 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2007 (filetype >= EXT4_FT_MAX))
2008 return DT_UNKNOWN;
2009
2010 return ext4_filetype_table[filetype];
2011 }
2012
2013 /* fsync.c */
2014 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2015 extern int ext4_flush_unwritten_io(struct inode *);
2016
2017 /* hash.c */
2018 extern int ext4fs_dirhash(const char *name, int len, struct
2019 dx_hash_info *hinfo);
2020
2021 /* ialloc.c */
2022 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2023 const struct qstr *qstr, __u32 goal,
2024 uid_t *owner, int handle_type,
2025 unsigned int line_no, int nblocks);
2026
2027 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2028 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2029 0, 0, 0)
2030 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2031 type, nblocks) \
2032 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2033 (type), __LINE__, (nblocks))
2034
2035
2036 extern void ext4_free_inode(handle_t *, struct inode *);
2037 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2038 extern unsigned long ext4_count_free_inodes(struct super_block *);
2039 extern unsigned long ext4_count_dirs(struct super_block *);
2040 extern void ext4_check_inodes_bitmap(struct super_block *);
2041 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2042 extern int ext4_init_inode_table(struct super_block *sb,
2043 ext4_group_t group, int barrier);
2044 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2045
2046 /* mballoc.c */
2047 extern long ext4_mb_stats;
2048 extern long ext4_mb_max_to_scan;
2049 extern int ext4_mb_init(struct super_block *);
2050 extern int ext4_mb_release(struct super_block *);
2051 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2052 struct ext4_allocation_request *, int *);
2053 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2054 extern void ext4_discard_preallocations(struct inode *);
2055 extern int __init ext4_init_mballoc(void);
2056 extern void ext4_exit_mballoc(void);
2057 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2058 struct buffer_head *bh, ext4_fsblk_t block,
2059 unsigned long count, int flags);
2060 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2061 ext4_group_t ngroups);
2062 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2063 ext4_group_t i, struct ext4_group_desc *desc);
2064 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2065 ext4_fsblk_t block, unsigned long count);
2066 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2067
2068 /* inode.c */
2069 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2070 ext4_lblk_t, int, int *);
2071 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2072 ext4_lblk_t, int, int *);
2073 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2074 struct buffer_head *bh_result, int create);
2075 int ext4_get_block(struct inode *inode, sector_t iblock,
2076 struct buffer_head *bh_result, int create);
2077 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2078 struct buffer_head *bh, int create);
2079 int ext4_walk_page_buffers(handle_t *handle,
2080 struct buffer_head *head,
2081 unsigned from,
2082 unsigned to,
2083 int *partial,
2084 int (*fn)(handle_t *handle,
2085 struct buffer_head *bh));
2086 int do_journal_get_write_access(handle_t *handle,
2087 struct buffer_head *bh);
2088 #define FALL_BACK_TO_NONDELALLOC 1
2089 #define CONVERT_INLINE_DATA 2
2090
2091 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2092 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2093 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2094 extern int ext4_setattr(struct dentry *, struct iattr *);
2095 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2096 struct kstat *stat);
2097 extern void ext4_evict_inode(struct inode *);
2098 extern void ext4_clear_inode(struct inode *);
2099 extern int ext4_sync_inode(handle_t *, struct inode *);
2100 extern void ext4_dirty_inode(struct inode *, int);
2101 extern int ext4_change_inode_journal_flag(struct inode *, int);
2102 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2103 extern int ext4_can_truncate(struct inode *inode);
2104 extern void ext4_truncate(struct inode *);
2105 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2106 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2107 extern void ext4_set_inode_flags(struct inode *);
2108 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2109 extern int ext4_alloc_da_blocks(struct inode *inode);
2110 extern void ext4_set_aops(struct inode *inode);
2111 extern int ext4_writepage_trans_blocks(struct inode *);
2112 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2113 extern int ext4_discard_partial_page_buffers(handle_t *handle,
2114 struct address_space *mapping, loff_t from,
2115 loff_t length, int flags);
2116 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2117 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2118 extern void ext4_da_update_reserve_space(struct inode *inode,
2119 int used, int quota_claim);
2120
2121 /* indirect.c */
2122 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2123 struct ext4_map_blocks *map, int flags);
2124 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2125 const struct iovec *iov, loff_t offset,
2126 unsigned long nr_segs);
2127 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2128 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2129 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2130 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
2131 ext4_lblk_t first, ext4_lblk_t stop);
2132
2133 /* ioctl.c */
2134 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2135 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2136
2137 /* migrate.c */
2138 extern int ext4_ext_migrate(struct inode *);
2139 extern int ext4_ind_migrate(struct inode *inode);
2140
2141 /* namei.c */
2142 extern int ext4_dirent_csum_verify(struct inode *inode,
2143 struct ext4_dir_entry *dirent);
2144 extern int ext4_orphan_add(handle_t *, struct inode *);
2145 extern int ext4_orphan_del(handle_t *, struct inode *);
2146 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2147 __u32 start_minor_hash, __u32 *next_hash);
2148 extern int search_dir(struct buffer_head *bh,
2149 char *search_buf,
2150 int buf_size,
2151 struct inode *dir,
2152 const struct qstr *d_name,
2153 unsigned int offset,
2154 struct ext4_dir_entry_2 **res_dir);
2155 extern int ext4_generic_delete_entry(handle_t *handle,
2156 struct inode *dir,
2157 struct ext4_dir_entry_2 *de_del,
2158 struct buffer_head *bh,
2159 void *entry_buf,
2160 int buf_size,
2161 int csum_size);
2162
2163 /* resize.c */
2164 extern int ext4_group_add(struct super_block *sb,
2165 struct ext4_new_group_data *input);
2166 extern int ext4_group_extend(struct super_block *sb,
2167 struct ext4_super_block *es,
2168 ext4_fsblk_t n_blocks_count);
2169 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2170
2171 /* super.c */
2172 extern int ext4_calculate_overhead(struct super_block *sb);
2173 extern int ext4_superblock_csum_verify(struct super_block *sb,
2174 struct ext4_super_block *es);
2175 extern void ext4_superblock_csum_set(struct super_block *sb);
2176 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2177 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2178 extern void ext4_kvfree(void *ptr);
2179 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2180 ext4_group_t ngroup);
2181 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2182 char nbuf[16]);
2183 extern __printf(4, 5)
2184 void __ext4_error(struct super_block *, const char *, unsigned int,
2185 const char *, ...);
2186 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2187 __LINE__, ## message)
2188 extern __printf(5, 6)
2189 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2190 const char *, ...);
2191 extern __printf(5, 6)
2192 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2193 const char *, ...);
2194 extern void __ext4_std_error(struct super_block *, const char *,
2195 unsigned int, int);
2196 extern __printf(4, 5)
2197 void __ext4_abort(struct super_block *, const char *, unsigned int,
2198 const char *, ...);
2199 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2200 __LINE__, ## message)
2201 extern __printf(4, 5)
2202 void __ext4_warning(struct super_block *, const char *, unsigned int,
2203 const char *, ...);
2204 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2205 __LINE__, ## message)
2206 extern __printf(3, 4)
2207 void ext4_msg(struct super_block *, const char *, const char *, ...);
2208 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2209 const char *, unsigned int, const char *);
2210 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2211 __LINE__, msg)
2212 extern __printf(7, 8)
2213 void __ext4_grp_locked_error(const char *, unsigned int,
2214 struct super_block *, ext4_group_t,
2215 unsigned long, ext4_fsblk_t,
2216 const char *, ...);
2217 #define ext4_grp_locked_error(sb, grp, message...) \
2218 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2219 extern void ext4_update_dynamic_rev(struct super_block *sb);
2220 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2221 __u32 compat);
2222 extern int ext4_update_rocompat_feature(handle_t *handle,
2223 struct super_block *sb, __u32 rocompat);
2224 extern int ext4_update_incompat_feature(handle_t *handle,
2225 struct super_block *sb, __u32 incompat);
2226 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2227 struct ext4_group_desc *bg);
2228 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2229 struct ext4_group_desc *bg);
2230 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2231 struct ext4_group_desc *bg);
2232 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2233 struct ext4_group_desc *bg);
2234 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2235 struct ext4_group_desc *bg);
2236 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2237 struct ext4_group_desc *bg);
2238 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2239 struct ext4_group_desc *bg);
2240 extern void ext4_block_bitmap_set(struct super_block *sb,
2241 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2242 extern void ext4_inode_bitmap_set(struct super_block *sb,
2243 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2244 extern void ext4_inode_table_set(struct super_block *sb,
2245 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2246 extern void ext4_free_group_clusters_set(struct super_block *sb,
2247 struct ext4_group_desc *bg,
2248 __u32 count);
2249 extern void ext4_free_inodes_set(struct super_block *sb,
2250 struct ext4_group_desc *bg, __u32 count);
2251 extern void ext4_used_dirs_set(struct super_block *sb,
2252 struct ext4_group_desc *bg, __u32 count);
2253 extern void ext4_itable_unused_set(struct super_block *sb,
2254 struct ext4_group_desc *bg, __u32 count);
2255 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2256 struct ext4_group_desc *gdp);
2257 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2258 struct ext4_group_desc *gdp);
2259 extern int ext4_register_li_request(struct super_block *sb,
2260 ext4_group_t first_not_zeroed);
2261
2262 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2263 {
2264 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2265 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2266 (EXT4_SB(sb)->s_chksum_driver != NULL);
2267 }
2268
2269 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2270 {
2271 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2272 le32_to_cpu(es->s_blocks_count_lo);
2273 }
2274
2275 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2276 {
2277 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2278 le32_to_cpu(es->s_r_blocks_count_lo);
2279 }
2280
2281 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2282 {
2283 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2284 le32_to_cpu(es->s_free_blocks_count_lo);
2285 }
2286
2287 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2288 ext4_fsblk_t blk)
2289 {
2290 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2291 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2292 }
2293
2294 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2295 ext4_fsblk_t blk)
2296 {
2297 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2298 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2299 }
2300
2301 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2302 ext4_fsblk_t blk)
2303 {
2304 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2305 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2306 }
2307
2308 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2309 {
2310 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2311 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2312 le32_to_cpu(raw_inode->i_size_lo);
2313 else
2314 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2315 }
2316
2317 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2318 {
2319 raw_inode->i_size_lo = cpu_to_le32(i_size);
2320 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2321 }
2322
2323 static inline
2324 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2325 ext4_group_t group)
2326 {
2327 struct ext4_group_info ***grp_info;
2328 long indexv, indexh;
2329 grp_info = EXT4_SB(sb)->s_group_info;
2330 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2331 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2332 return grp_info[indexv][indexh];
2333 }
2334
2335 /*
2336 * Reading s_groups_count requires using smp_rmb() afterwards. See
2337 * the locking protocol documented in the comments of ext4_group_add()
2338 * in resize.c
2339 */
2340 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2341 {
2342 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2343
2344 smp_rmb();
2345 return ngroups;
2346 }
2347
2348 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2349 ext4_group_t block_group)
2350 {
2351 return block_group >> sbi->s_log_groups_per_flex;
2352 }
2353
2354 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2355 {
2356 return 1 << sbi->s_log_groups_per_flex;
2357 }
2358
2359 #define ext4_std_error(sb, errno) \
2360 do { \
2361 if ((errno)) \
2362 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2363 } while (0)
2364
2365 #ifdef CONFIG_SMP
2366 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2367 * counters. So we need to make sure we have free clusters more
2368 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2369 */
2370 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2371 #else
2372 #define EXT4_FREECLUSTERS_WATERMARK 0
2373 #endif
2374
2375 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2376 {
2377 /*
2378 * XXX: replace with spinlock if seen contended -bzzz
2379 */
2380 down_write(&EXT4_I(inode)->i_data_sem);
2381 if (newsize > EXT4_I(inode)->i_disksize)
2382 EXT4_I(inode)->i_disksize = newsize;
2383 up_write(&EXT4_I(inode)->i_data_sem);
2384 return ;
2385 }
2386
2387 struct ext4_group_info {
2388 unsigned long bb_state;
2389 struct rb_root bb_free_root;
2390 ext4_grpblk_t bb_first_free; /* first free block */
2391 ext4_grpblk_t bb_free; /* total free blocks */
2392 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2393 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2394 struct list_head bb_prealloc_list;
2395 #ifdef DOUBLE_CHECK
2396 void *bb_bitmap;
2397 #endif
2398 struct rw_semaphore alloc_sem;
2399 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2400 * regions, index is order.
2401 * bb_counters[3] = 5 means
2402 * 5 free 8-block regions. */
2403 };
2404
2405 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2406 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2407
2408 #define EXT4_MB_GRP_NEED_INIT(grp) \
2409 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2410
2411 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2412 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2413 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2414 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2415 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2416 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2417
2418 #define EXT4_MAX_CONTENTION 8
2419 #define EXT4_CONTENTION_THRESHOLD 2
2420
2421 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2422 ext4_group_t group)
2423 {
2424 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2425 }
2426
2427 /*
2428 * Returns true if the filesystem is busy enough that attempts to
2429 * access the block group locks has run into contention.
2430 */
2431 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2432 {
2433 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2434 }
2435
2436 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2437 {
2438 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2439 if (spin_trylock(lock))
2440 /*
2441 * We're able to grab the lock right away, so drop the
2442 * lock contention counter.
2443 */
2444 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2445 else {
2446 /*
2447 * The lock is busy, so bump the contention counter,
2448 * and then wait on the spin lock.
2449 */
2450 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2451 EXT4_MAX_CONTENTION);
2452 spin_lock(lock);
2453 }
2454 }
2455
2456 static inline void ext4_unlock_group(struct super_block *sb,
2457 ext4_group_t group)
2458 {
2459 spin_unlock(ext4_group_lock_ptr(sb, group));
2460 }
2461
2462 /*
2463 * Block validity checking
2464 */
2465 #define ext4_check_indirect_blockref(inode, bh) \
2466 ext4_check_blockref(__func__, __LINE__, inode, \
2467 (__le32 *)(bh)->b_data, \
2468 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2469
2470 #define ext4_ind_check_inode(inode) \
2471 ext4_check_blockref(__func__, __LINE__, inode, \
2472 EXT4_I(inode)->i_data, \
2473 EXT4_NDIR_BLOCKS)
2474
2475 /*
2476 * Inodes and files operations
2477 */
2478
2479 /* dir.c */
2480 extern const struct file_operations ext4_dir_operations;
2481
2482 /* file.c */
2483 extern const struct inode_operations ext4_file_inode_operations;
2484 extern const struct file_operations ext4_file_operations;
2485 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2486 extern void ext4_unwritten_wait(struct inode *inode);
2487
2488 /* inline.c */
2489 extern int ext4_has_inline_data(struct inode *inode);
2490 extern int ext4_get_inline_size(struct inode *inode);
2491 extern int ext4_get_max_inline_size(struct inode *inode);
2492 extern int ext4_find_inline_data_nolock(struct inode *inode);
2493 extern void ext4_write_inline_data(struct inode *inode,
2494 struct ext4_iloc *iloc,
2495 void *buffer, loff_t pos,
2496 unsigned int len);
2497 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2498 unsigned int len);
2499 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2500 unsigned int len);
2501 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2502
2503 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2504 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2505 struct inode *inode,
2506 loff_t pos, unsigned len,
2507 unsigned flags,
2508 struct page **pagep);
2509 extern int ext4_write_inline_data_end(struct inode *inode,
2510 loff_t pos, unsigned len,
2511 unsigned copied,
2512 struct page *page);
2513 extern struct buffer_head *
2514 ext4_journalled_write_inline_data(struct inode *inode,
2515 unsigned len,
2516 struct page *page);
2517 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2518 struct inode *inode,
2519 loff_t pos, unsigned len,
2520 unsigned flags,
2521 struct page **pagep,
2522 void **fsdata);
2523 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2524 unsigned len, unsigned copied,
2525 struct page *page);
2526 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2527 struct inode *inode);
2528 extern int ext4_try_create_inline_dir(handle_t *handle,
2529 struct inode *parent,
2530 struct inode *inode);
2531 extern int ext4_read_inline_dir(struct file *filp,
2532 void *dirent, filldir_t filldir,
2533 int *has_inline_data);
2534 extern int htree_inlinedir_to_tree(struct file *dir_file,
2535 struct inode *dir, ext4_lblk_t block,
2536 struct dx_hash_info *hinfo,
2537 __u32 start_hash, __u32 start_minor_hash,
2538 int *has_inline_data);
2539 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2540 const struct qstr *d_name,
2541 struct ext4_dir_entry_2 **res_dir,
2542 int *has_inline_data);
2543 extern int ext4_delete_inline_entry(handle_t *handle,
2544 struct inode *dir,
2545 struct ext4_dir_entry_2 *de_del,
2546 struct buffer_head *bh,
2547 int *has_inline_data);
2548 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2549 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2550 struct ext4_dir_entry_2 **parent_de,
2551 int *retval);
2552 extern int ext4_inline_data_fiemap(struct inode *inode,
2553 struct fiemap_extent_info *fieinfo,
2554 int *has_inline);
2555 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2556 struct inode *inode,
2557 int needed);
2558 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2559
2560 extern int ext4_convert_inline_data(struct inode *inode);
2561
2562 /* namei.c */
2563 extern const struct inode_operations ext4_dir_inode_operations;
2564 extern const struct inode_operations ext4_special_inode_operations;
2565 extern struct dentry *ext4_get_parent(struct dentry *child);
2566 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2567 struct ext4_dir_entry_2 *de,
2568 int blocksize, int csum_size,
2569 unsigned int parent_ino, int dotdot_real_len);
2570 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2571 unsigned int blocksize);
2572 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2573 struct inode *inode,
2574 struct buffer_head *bh);
2575 #define S_SHIFT 12
2576 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2577 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2578 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2579 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2580 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2581 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2582 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2583 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2584 };
2585
2586 static inline void ext4_set_de_type(struct super_block *sb,
2587 struct ext4_dir_entry_2 *de,
2588 umode_t mode) {
2589 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2590 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2591 }
2592
2593
2594 /* symlink.c */
2595 extern const struct inode_operations ext4_symlink_inode_operations;
2596 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2597
2598 /* block_validity */
2599 extern void ext4_release_system_zone(struct super_block *sb);
2600 extern int ext4_setup_system_zone(struct super_block *sb);
2601 extern int __init ext4_init_system_zone(void);
2602 extern void ext4_exit_system_zone(void);
2603 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2604 ext4_fsblk_t start_blk,
2605 unsigned int count);
2606 extern int ext4_check_blockref(const char *, unsigned int,
2607 struct inode *, __le32 *, unsigned int);
2608
2609 /* extents.c */
2610 struct ext4_ext_path;
2611 struct ext4_extent;
2612
2613 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2614 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2615 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2616 int chunk);
2617 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2618 struct ext4_map_blocks *map, int flags);
2619 extern void ext4_ext_truncate(handle_t *, struct inode *);
2620 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2621 ext4_lblk_t end);
2622 extern void ext4_ext_init(struct super_block *);
2623 extern void ext4_ext_release(struct super_block *);
2624 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2625 loff_t len);
2626 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2627 ssize_t len);
2628 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2629 struct ext4_map_blocks *map, int flags);
2630 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2631 ext4_lblk_t lblocks);
2632 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2633 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2634 int num,
2635 struct ext4_ext_path *path);
2636 extern int ext4_can_extents_be_merged(struct inode *inode,
2637 struct ext4_extent *ex1,
2638 struct ext4_extent *ex2);
2639 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2640 struct ext4_ext_path *,
2641 struct ext4_extent *, int);
2642 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2643 struct ext4_ext_path *);
2644 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2645 extern int ext4_ext_check_inode(struct inode *inode);
2646 extern int ext4_find_delalloc_range(struct inode *inode,
2647 ext4_lblk_t lblk_start,
2648 ext4_lblk_t lblk_end);
2649 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2650 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2651 __u64 start, __u64 len);
2652
2653
2654 /* move_extent.c */
2655 extern void ext4_double_down_write_data_sem(struct inode *first,
2656 struct inode *second);
2657 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2658 struct inode *donor_inode);
2659 void ext4_inode_double_lock(struct inode *inode1, struct inode *inode2);
2660 void ext4_inode_double_unlock(struct inode *inode1, struct inode *inode2);
2661 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2662 __u64 start_orig, __u64 start_donor,
2663 __u64 len, __u64 *moved_len);
2664
2665 /* page-io.c */
2666 extern int __init ext4_init_pageio(void);
2667 extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2668 extern void ext4_exit_pageio(void);
2669 extern void ext4_ioend_shutdown(struct inode *);
2670 extern void ext4_free_io_end(ext4_io_end_t *io);
2671 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2672 extern void ext4_end_io_work(struct work_struct *work);
2673 extern void ext4_io_submit(struct ext4_io_submit *io);
2674 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2675 struct page *page,
2676 int len,
2677 struct writeback_control *wbc);
2678
2679 /* mmp.c */
2680 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2681 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2682 extern int ext4_mmp_csum_verify(struct super_block *sb,
2683 struct mmp_struct *mmp);
2684
2685 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2686 enum ext4_state_bits {
2687 BH_Uninit /* blocks are allocated but uninitialized on disk */
2688 = BH_JBDPrivateStart,
2689 BH_AllocFromCluster, /* allocated blocks were part of already
2690 * allocated cluster. Note that this flag will
2691 * never, ever appear in a buffer_head's state
2692 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2693 * this is used. */
2694 };
2695
2696 BUFFER_FNS(Uninit, uninit)
2697 TAS_BUFFER_FNS(Uninit, uninit)
2698
2699 /*
2700 * Add new method to test whether block and inode bitmaps are properly
2701 * initialized. With uninit_bg reading the block from disk is not enough
2702 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2703 */
2704 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2705
2706 static inline int bitmap_uptodate(struct buffer_head *bh)
2707 {
2708 return (buffer_uptodate(bh) &&
2709 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2710 }
2711 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2712 {
2713 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2714 }
2715
2716 /*
2717 * Disable DIO read nolock optimization, so new dioreaders will be forced
2718 * to grab i_mutex
2719 */
2720 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2721 {
2722 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2723 smp_mb();
2724 }
2725 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2726 {
2727 smp_mb();
2728 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2729 }
2730
2731 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2732
2733 /* For ioend & aio unwritten conversion wait queues */
2734 #define EXT4_WQ_HASH_SZ 37
2735 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2736 EXT4_WQ_HASH_SZ])
2737 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2738 EXT4_WQ_HASH_SZ])
2739 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2740 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2741
2742 #define EXT4_RESIZING 0
2743 extern int ext4_resize_begin(struct super_block *sb);
2744 extern void ext4_resize_end(struct super_block *sb);
2745
2746 #endif /* __KERNEL__ */
2747
2748 #endif /* _EXT4_H */
Linux with added FPC-III support