projects / GitHub / mt8127 / android_kernel_alcatel_ttab.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge tag 'v3.10.97' into update
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / fs / ext4 / ext4.h
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <crypto/hash.h>
34 #ifdef __KERNEL__
35 #include <linux/compat.h>
36 #endif
37
38 /*
39 * The fourth extended filesystem constants/structures
40 */
41
42 /*
43 * Define EXT4FS_DEBUG to produce debug messages
44 */
45 #undef EXT4FS_DEBUG
46
47 /*
48 * Debug code
49 */
50 #ifdef EXT4FS_DEBUG
51 #define ext4_debug(f, a...) \
52 do { \
53 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
54 __FILE__, __LINE__, __func__); \
55 printk(KERN_DEBUG f, ## a); \
56 } while (0)
57 #else
58 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
59 #endif
60
61 /*
62 * Turn on EXT_DEBUG to get lots of info about extents operations.
63 */
64 #define EXT_DEBUG__
65 #ifdef EXT_DEBUG
66 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
67 #else
68 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
69 #endif
70
71 #define EXT4_ERROR_INODE(inode, fmt, a...) \
72 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
73
74 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
75 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
76
77 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
78 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
79
80 /* data type for block offset of block group */
81 typedef int ext4_grpblk_t;
82
83 /* data type for filesystem-wide blocks number */
84 typedef unsigned long long ext4_fsblk_t;
85
86 /* data type for file logical block number */
87 typedef __u32 ext4_lblk_t;
88
89 /* data type for block group number */
90 typedef unsigned int ext4_group_t;
91
92 /*
93 * Flags used in mballoc's allocation_context flags field.
94 *
95 * Also used to show what's going on for debugging purposes when the
96 * flag field is exported via the traceport interface
97 */
98
99 /* prefer goal again. length */
100 #define EXT4_MB_HINT_MERGE 0x0001
101 /* blocks already reserved */
102 #define EXT4_MB_HINT_RESERVED 0x0002
103 /* metadata is being allocated */
104 #define EXT4_MB_HINT_METADATA 0x0004
105 /* first blocks in the file */
106 #define EXT4_MB_HINT_FIRST 0x0008
107 /* search for the best chunk */
108 #define EXT4_MB_HINT_BEST 0x0010
109 /* data is being allocated */
110 #define EXT4_MB_HINT_DATA 0x0020
111 /* don't preallocate (for tails) */
112 #define EXT4_MB_HINT_NOPREALLOC 0x0040
113 /* allocate for locality group */
114 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
115 /* allocate goal blocks or none */
116 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
117 /* goal is meaningful */
118 #define EXT4_MB_HINT_TRY_GOAL 0x0200
119 /* blocks already pre-reserved by delayed allocation */
120 #define EXT4_MB_DELALLOC_RESERVED 0x0400
121 /* We are doing stream allocation */
122 #define EXT4_MB_STREAM_ALLOC 0x0800
123 /* Use reserved root blocks if needed */
124 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
125 /* Use blocks from reserved pool */
126 #define EXT4_MB_USE_RESERVED 0x2000
127
128 struct ext4_allocation_request {
129 /* target inode for block we're allocating */
130 struct inode *inode;
131 /* how many blocks we want to allocate */
132 unsigned int len;
133 /* logical block in target inode */
134 ext4_lblk_t logical;
135 /* the closest logical allocated block to the left */
136 ext4_lblk_t lleft;
137 /* the closest logical allocated block to the right */
138 ext4_lblk_t lright;
139 /* phys. target (a hint) */
140 ext4_fsblk_t goal;
141 /* phys. block for the closest logical allocated block to the left */
142 ext4_fsblk_t pleft;
143 /* phys. block for the closest logical allocated block to the right */
144 ext4_fsblk_t pright;
145 /* flags. see above EXT4_MB_HINT_* */
146 unsigned int flags;
147 };
148
149 /*
150 * Logical to physical block mapping, used by ext4_map_blocks()
151 *
152 * This structure is used to pass requests into ext4_map_blocks() as
153 * well as to store the information returned by ext4_map_blocks(). It
154 * takes less room on the stack than a struct buffer_head.
155 */
156 #define EXT4_MAP_NEW (1 << BH_New)
157 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
158 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
159 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
160 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
161 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
162 * ext4_map_blocks wants to know whether or not the underlying cluster has
163 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
164 * the requested mapping was from previously mapped (or delayed allocated)
165 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
166 * should never appear on buffer_head's state flags.
167 */
168 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
169 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
170 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
171 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
172
173 struct ext4_map_blocks {
174 ext4_fsblk_t m_pblk;
175 ext4_lblk_t m_lblk;
176 unsigned int m_len;
177 unsigned int m_flags;
178 };
179
180 /*
181 * For delayed allocation tracking
182 */
183 struct mpage_da_data {
184 struct inode *inode;
185 sector_t b_blocknr; /* start block number of extent */
186 size_t b_size; /* size of extent */
187 unsigned long b_state; /* state of the extent */
188 unsigned long first_page, next_page; /* extent of pages */
189 struct writeback_control *wbc;
190 int io_done;
191 int pages_written;
192 int retval;
193 };
194
195 /*
196 * Flags for ext4_io_end->flags
197 */
198 #define EXT4_IO_END_UNWRITTEN 0x0001
199 #define EXT4_IO_END_ERROR 0x0002
200 #define EXT4_IO_END_DIRECT 0x0004
201
202 /*
203 * For converting uninitialized extents on a work queue.
204 */
205 typedef struct ext4_io_end {
206 struct list_head list; /* per-file finished IO list */
207 struct inode *inode; /* file being written to */
208 unsigned int flag; /* unwritten or not */
209 loff_t offset; /* offset in the file */
210 ssize_t size; /* size of the extent */
211 struct kiocb *iocb; /* iocb struct for AIO */
212 int result; /* error value for AIO */
213 } ext4_io_end_t;
214
215 struct ext4_io_submit {
216 int io_op;
217 struct bio *io_bio;
218 ext4_io_end_t *io_end;
219 sector_t io_next_block;
220 };
221
222 /*
223 * Special inodes numbers
224 */
225 #define EXT4_BAD_INO 1 /* Bad blocks inode */
226 #define EXT4_ROOT_INO 2 /* Root inode */
227 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
228 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
229 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
230 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
231 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
232 #define EXT4_JOURNAL_INO 8 /* Journal inode */
233
234 /* First non-reserved inode for old ext4 filesystems */
235 #define EXT4_GOOD_OLD_FIRST_INO 11
236
237 /*
238 * Maximal count of links to a file
239 */
240 #define EXT4_LINK_MAX 65000
241
242 /*
243 * Macro-instructions used to manage several block sizes
244 */
245 #define EXT4_MIN_BLOCK_SIZE 1024
246 #define EXT4_MAX_BLOCK_SIZE 65536
247 #define EXT4_MIN_BLOCK_LOG_SIZE 10
248 #define EXT4_MAX_BLOCK_LOG_SIZE 16
249 #ifdef __KERNEL__
250 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
251 #else
252 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
253 #endif
254 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
255 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
256 EXT4_SB(s)->s_cluster_bits)
257 #ifdef __KERNEL__
258 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
259 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
260 #else
261 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
262 #endif
263 #ifdef __KERNEL__
264 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
265 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
266 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
267 #else
268 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
269 EXT4_GOOD_OLD_INODE_SIZE : \
270 (s)->s_inode_size)
271 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
272 EXT4_GOOD_OLD_FIRST_INO : \
273 (s)->s_first_ino)
274 #endif
275 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
276
277 /* Translate a block number to a cluster number */
278 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
279 /* Translate a cluster number to a block number */
280 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
281 /* Translate # of blks to # of clusters */
282 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
283 (sbi)->s_cluster_bits)
284 /* Mask out the low bits to get the starting block of the cluster */
285 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
286 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
287 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
288 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
289 /* Get the cluster offset */
290 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
291 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
292 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
293 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
294
295 /*
296 * Structure of a blocks group descriptor
297 */
298 struct ext4_group_desc
299 {
300 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
301 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
302 __le32 bg_inode_table_lo; /* Inodes table block */
303 __le16 bg_free_blocks_count_lo;/* Free blocks count */
304 __le16 bg_free_inodes_count_lo;/* Free inodes count */
305 __le16 bg_used_dirs_count_lo; /* Directories count */
306 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
307 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
308 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
309 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
310 __le16 bg_itable_unused_lo; /* Unused inodes count */
311 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
312 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
313 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
314 __le32 bg_inode_table_hi; /* Inodes table block MSB */
315 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
316 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
317 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
318 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
319 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
320 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
321 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
322 __u32 bg_reserved;
323 };
324
325 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
326 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
327 sizeof(__le16))
328 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
329 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
330 sizeof(__le16))
331
332 /*
333 * Structure of a flex block group info
334 */
335
336 struct flex_groups {
337 atomic64_t free_clusters;
338 atomic_t free_inodes;
339 atomic_t used_dirs;
340 };
341
342 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
343 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
344 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
345
346 /*
347 * Macro-instructions used to manage group descriptors
348 */
349 #define EXT4_MIN_DESC_SIZE 32
350 #define EXT4_MIN_DESC_SIZE_64BIT 64
351 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
352 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
353 #ifdef __KERNEL__
354 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
355 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
356 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
357 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
358 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
359 #else
360 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
361 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
362 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
363 #endif
364
365 /*
366 * Constants relative to the data blocks
367 */
368 #define EXT4_NDIR_BLOCKS 12
369 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
370 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
371 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
372 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
373
374 /*
375 * Inode flags
376 */
377 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
378 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
379 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
380 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
381 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
382 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
383 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
384 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
385 /* Reserved for compression usage... */
386 #define EXT4_DIRTY_FL 0x00000100
387 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
388 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
389 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
390 /* End compression flags --- maybe not all used */
391 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
392 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
393 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
394 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
395 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
396 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
397 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
398 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
399 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
400 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
401 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
402 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
403
404 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
405 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
406
407 /* Flags that should be inherited by new inodes from their parent. */
408 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
409 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
410 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
411 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
412
413 /* Flags that are appropriate for regular files (all but dir-specific ones). */
414 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
415
416 /* Flags that are appropriate for non-directories/regular files. */
417 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
418
419 /* Mask out flags that are inappropriate for the given type of inode. */
420 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
421 {
422 if (S_ISDIR(mode))
423 return flags;
424 else if (S_ISREG(mode))
425 return flags & EXT4_REG_FLMASK;
426 else
427 return flags & EXT4_OTHER_FLMASK;
428 }
429
430 /*
431 * Inode flags used for atomic set/get
432 */
433 enum {
434 EXT4_INODE_SECRM = 0, /* Secure deletion */
435 EXT4_INODE_UNRM = 1, /* Undelete */
436 EXT4_INODE_COMPR = 2, /* Compress file */
437 EXT4_INODE_SYNC = 3, /* Synchronous updates */
438 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
439 EXT4_INODE_APPEND = 5, /* writes to file may only append */
440 EXT4_INODE_NODUMP = 6, /* do not dump file */
441 EXT4_INODE_NOATIME = 7, /* do not update atime */
442 /* Reserved for compression usage... */
443 EXT4_INODE_DIRTY = 8,
444 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
445 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
446 EXT4_INODE_ECOMPR = 11, /* Compression error */
447 /* End compression flags --- maybe not all used */
448 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
449 EXT4_INODE_IMAGIC = 13, /* AFS directory */
450 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
451 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
452 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
453 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
454 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
455 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
456 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
457 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
458 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
459 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
460 };
461
462 /*
463 * Since it's pretty easy to mix up bit numbers and hex values, we use a
464 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
465 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
466 * any extra space in the compiled kernel image, otherwise, the build will fail.
467 * It's important that these values are the same, since we are using
468 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
469 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
470 * values found in ext2, ext3 and ext4 filesystems, and of course the values
471 * defined in e2fsprogs.
472 *
473 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
474 */
475 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
476 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
477
478 static inline void ext4_check_flag_values(void)
479 {
480 CHECK_FLAG_VALUE(SECRM);
481 CHECK_FLAG_VALUE(UNRM);
482 CHECK_FLAG_VALUE(COMPR);
483 CHECK_FLAG_VALUE(SYNC);
484 CHECK_FLAG_VALUE(IMMUTABLE);
485 CHECK_FLAG_VALUE(APPEND);
486 CHECK_FLAG_VALUE(NODUMP);
487 CHECK_FLAG_VALUE(NOATIME);
488 CHECK_FLAG_VALUE(DIRTY);
489 CHECK_FLAG_VALUE(COMPRBLK);
490 CHECK_FLAG_VALUE(NOCOMPR);
491 CHECK_FLAG_VALUE(ECOMPR);
492 CHECK_FLAG_VALUE(INDEX);
493 CHECK_FLAG_VALUE(IMAGIC);
494 CHECK_FLAG_VALUE(JOURNAL_DATA);
495 CHECK_FLAG_VALUE(NOTAIL);
496 CHECK_FLAG_VALUE(DIRSYNC);
497 CHECK_FLAG_VALUE(TOPDIR);
498 CHECK_FLAG_VALUE(HUGE_FILE);
499 CHECK_FLAG_VALUE(EXTENTS);
500 CHECK_FLAG_VALUE(EA_INODE);
501 CHECK_FLAG_VALUE(EOFBLOCKS);
502 CHECK_FLAG_VALUE(INLINE_DATA);
503 CHECK_FLAG_VALUE(RESERVED);
504 }
505
506 /* Used to pass group descriptor data when online resize is done */
507 struct ext4_new_group_input {
508 __u32 group; /* Group number for this data */
509 __u64 block_bitmap; /* Absolute block number of block bitmap */
510 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
511 __u64 inode_table; /* Absolute block number of inode table start */
512 __u32 blocks_count; /* Total number of blocks in this group */
513 __u16 reserved_blocks; /* Number of reserved blocks in this group */
514 __u16 unused;
515 };
516
517 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
518 struct compat_ext4_new_group_input {
519 u32 group;
520 compat_u64 block_bitmap;
521 compat_u64 inode_bitmap;
522 compat_u64 inode_table;
523 u32 blocks_count;
524 u16 reserved_blocks;
525 u16 unused;
526 };
527 #endif
528
529 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
530 struct ext4_new_group_data {
531 __u32 group;
532 __u64 block_bitmap;
533 __u64 inode_bitmap;
534 __u64 inode_table;
535 __u32 blocks_count;
536 __u16 reserved_blocks;
537 __u16 unused;
538 __u32 free_blocks_count;
539 };
540
541 /* Indexes used to index group tables in ext4_new_group_data */
542 enum {
543 BLOCK_BITMAP = 0, /* block bitmap */
544 INODE_BITMAP, /* inode bitmap */
545 INODE_TABLE, /* inode tables */
546 GROUP_TABLE_COUNT,
547 };
548
549 /*
550 * Flags used by ext4_map_blocks()
551 */
552 /* Allocate any needed blocks and/or convert an unitialized
553 extent to be an initialized ext4 */
554 #define EXT4_GET_BLOCKS_CREATE 0x0001
555 /* Request the creation of an unitialized extent */
556 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
557 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
558 EXT4_GET_BLOCKS_CREATE)
559 /* Caller is from the delayed allocation writeout path
560 * finally doing the actual allocation of delayed blocks */
561 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
562 /* caller is from the direct IO path, request to creation of an
563 unitialized extents if not allocated, split the uninitialized
564 extent if blocks has been preallocated already*/
565 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
566 #define EXT4_GET_BLOCKS_CONVERT 0x0010
567 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
568 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
569 /* Convert extent to initialized after IO complete */
570 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
571 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
572 /* Eventual metadata allocation (due to growing extent tree)
573 * should not fail, so try to use reserved blocks for that.*/
574 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
575 /* Don't normalize allocation size (used for fallocate) */
576 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
577 /* Request will not result in inode size update (user for fallocate) */
578 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
579 /* Do not take i_data_sem locking in ext4_map_blocks */
580 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
581 /* Do not put hole in extent cache */
582 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
583
584 /*
585 * Flags used by ext4_free_blocks
586 */
587 #define EXT4_FREE_BLOCKS_METADATA 0x0001
588 #define EXT4_FREE_BLOCKS_FORGET 0x0002
589 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
590 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
591 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
592 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
593 #define EXT4_FREE_BLOCKS_RESERVE 0x0040
594
595 /*
596 * Flags used by ext4_discard_partial_page_buffers
597 */
598 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
599
600 /*
601 * ioctl commands
602 */
603 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
604 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
605 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
606 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
607 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
608 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
609 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
610 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
611 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
612 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
613 #define EXT4_IOC_MIGRATE _IO('f', 9)
614 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
615 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
616 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
617 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
618 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
619 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
620
621 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
622 /*
623 * ioctl commands in 32 bit emulation
624 */
625 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
626 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
627 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
628 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
629 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
630 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
631 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
632 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
633 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
634 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
635 #endif
636
637 /* Max physical block we can address w/o extents */
638 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
639
640 /*
641 * Structure of an inode on the disk
642 */
643 struct ext4_inode {
644 __le16 i_mode; /* File mode */
645 __le16 i_uid; /* Low 16 bits of Owner Uid */
646 __le32 i_size_lo; /* Size in bytes */
647 __le32 i_atime; /* Access time */
648 __le32 i_ctime; /* Inode Change time */
649 __le32 i_mtime; /* Modification time */
650 __le32 i_dtime; /* Deletion Time */
651 __le16 i_gid; /* Low 16 bits of Group Id */
652 __le16 i_links_count; /* Links count */
653 __le32 i_blocks_lo; /* Blocks count */
654 __le32 i_flags; /* File flags */
655 union {
656 struct {
657 __le32 l_i_version;
658 } linux1;
659 struct {
660 __u32 h_i_translator;
661 } hurd1;
662 struct {
663 __u32 m_i_reserved1;
664 } masix1;
665 } osd1; /* OS dependent 1 */
666 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
667 __le32 i_generation; /* File version (for NFS) */
668 __le32 i_file_acl_lo; /* File ACL */
669 __le32 i_size_high;
670 __le32 i_obso_faddr; /* Obsoleted fragment address */
671 union {
672 struct {
673 __le16 l_i_blocks_high; /* were l_i_reserved1 */
674 __le16 l_i_file_acl_high;
675 __le16 l_i_uid_high; /* these 2 fields */
676 __le16 l_i_gid_high; /* were reserved2[0] */
677 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
678 __le16 l_i_reserved;
679 } linux2;
680 struct {
681 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
682 __u16 h_i_mode_high;
683 __u16 h_i_uid_high;
684 __u16 h_i_gid_high;
685 __u32 h_i_author;
686 } hurd2;
687 struct {
688 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
689 __le16 m_i_file_acl_high;
690 __u32 m_i_reserved2[2];
691 } masix2;
692 } osd2; /* OS dependent 2 */
693 __le16 i_extra_isize;
694 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
695 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
696 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
697 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
698 __le32 i_crtime; /* File Creation time */
699 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
700 __le32 i_version_hi; /* high 32 bits for 64-bit version */
701 };
702
703 struct move_extent {
704 __u32 reserved; /* should be zero */
705 __u32 donor_fd; /* donor file descriptor */
706 __u64 orig_start; /* logical start offset in block for orig */
707 __u64 donor_start; /* logical start offset in block for donor */
708 __u64 len; /* block length to be moved */
709 __u64 moved_len; /* moved block length */
710 };
711
712 #define EXT4_EPOCH_BITS 2
713 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
714 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
715
716 /*
717 * Extended fields will fit into an inode if the filesystem was formatted
718 * with large inodes (-I 256 or larger) and there are not currently any EAs
719 * consuming all of the available space. For new inodes we always reserve
720 * enough space for the kernel's known extended fields, but for inodes
721 * created with an old kernel this might not have been the case. None of
722 * the extended inode fields is critical for correct filesystem operation.
723 * This macro checks if a certain field fits in the inode. Note that
724 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
725 */
726 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
727 ((offsetof(typeof(*ext4_inode), field) + \
728 sizeof((ext4_inode)->field)) \
729 <= (EXT4_GOOD_OLD_INODE_SIZE + \
730 (einode)->i_extra_isize)) \
731
732 /*
733 * We use an encoding that preserves the times for extra epoch "00":
734 *
735 * extra msb of adjust for signed
736 * epoch 32-bit 32-bit tv_sec to
737 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
738 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
739 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
740 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
741 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
742 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
743 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
744 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
745 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
746 *
747 * Note that previous versions of the kernel on 64-bit systems would
748 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
749 * 1970. e2fsck will correct this, assuming that it is run on the
750 * affected filesystem before 2242.
751 */
752
753 static inline __le32 ext4_encode_extra_time(struct timespec *time)
754 {
755 u32 extra = sizeof(time->tv_sec) > 4 ?
756 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
757 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
758 }
759
760 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
761 {
762 if (unlikely(sizeof(time->tv_sec) > 4 &&
763 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
764 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
765 /* Handle legacy encoding of pre-1970 dates with epoch
766 * bits 1,1. We assume that by kernel version 4.20,
767 * everyone will have run fsck over the affected
768 * filesystems to correct the problem. (This
769 * backwards compatibility may be removed before this
770 * time, at the discretion of the ext4 developers.)
771 */
772 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
773 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
774 extra_bits = 0;
775 time->tv_sec += extra_bits << 32;
776 #else
777 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
778 #endif
779 }
780 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
781 }
782
783 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
784 do { \
785 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
786 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
787 (raw_inode)->xtime ## _extra = \
788 ext4_encode_extra_time(&(inode)->xtime); \
789 } while (0)
790
791 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
792 do { \
793 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
794 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
795 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
796 (raw_inode)->xtime ## _extra = \
797 ext4_encode_extra_time(&(einode)->xtime); \
798 } while (0)
799
800 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
801 do { \
802 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
803 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
804 ext4_decode_extra_time(&(inode)->xtime, \
805 raw_inode->xtime ## _extra); \
806 else \
807 (inode)->xtime.tv_nsec = 0; \
808 } while (0)
809
810 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
811 do { \
812 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
813 (einode)->xtime.tv_sec = \
814 (signed)le32_to_cpu((raw_inode)->xtime); \
815 else \
816 (einode)->xtime.tv_sec = 0; \
817 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
818 ext4_decode_extra_time(&(einode)->xtime, \
819 raw_inode->xtime ## _extra); \
820 else \
821 (einode)->xtime.tv_nsec = 0; \
822 } while (0)
823
824 #define i_disk_version osd1.linux1.l_i_version
825
826 #if defined(__KERNEL__) || defined(__linux__)
827 #define i_reserved1 osd1.linux1.l_i_reserved1
828 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
829 #define i_blocks_high osd2.linux2.l_i_blocks_high
830 #define i_uid_low i_uid
831 #define i_gid_low i_gid
832 #define i_uid_high osd2.linux2.l_i_uid_high
833 #define i_gid_high osd2.linux2.l_i_gid_high
834 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
835
836 #elif defined(__GNU__)
837
838 #define i_translator osd1.hurd1.h_i_translator
839 #define i_uid_high osd2.hurd2.h_i_uid_high
840 #define i_gid_high osd2.hurd2.h_i_gid_high
841 #define i_author osd2.hurd2.h_i_author
842
843 #elif defined(__masix__)
844
845 #define i_reserved1 osd1.masix1.m_i_reserved1
846 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
847 #define i_reserved2 osd2.masix2.m_i_reserved2
848
849 #endif /* defined(__KERNEL__) || defined(__linux__) */
850
851 #include "extents_status.h"
852
853 /*
854 * fourth extended file system inode data in memory
855 */
856 struct ext4_inode_info {
857 __le32 i_data[15]; /* unconverted */
858 __u32 i_dtime;
859 ext4_fsblk_t i_file_acl;
860
861 /*
862 * i_block_group is the number of the block group which contains
863 * this file's inode. Constant across the lifetime of the inode,
864 * it is ued for making block allocation decisions - we try to
865 * place a file's data blocks near its inode block, and new inodes
866 * near to their parent directory's inode.
867 */
868 ext4_group_t i_block_group;
869 ext4_lblk_t i_dir_start_lookup;
870 #if (BITS_PER_LONG < 64)
871 unsigned long i_state_flags; /* Dynamic state flags */
872 #endif
873 unsigned long i_flags;
874
875 /*
876 * Extended attributes can be read independently of the main file
877 * data. Taking i_mutex even when reading would cause contention
878 * between readers of EAs and writers of regular file data, so
879 * instead we synchronize on xattr_sem when reading or changing
880 * EAs.
881 */
882 struct rw_semaphore xattr_sem;
883
884 struct list_head i_orphan; /* unlinked but open inodes */
885
886 /*
887 * i_disksize keeps track of what the inode size is ON DISK, not
888 * in memory. During truncate, i_size is set to the new size by
889 * the VFS prior to calling ext4_truncate(), but the filesystem won't
890 * set i_disksize to 0 until the truncate is actually under way.
891 *
892 * The intent is that i_disksize always represents the blocks which
893 * are used by this file. This allows recovery to restart truncate
894 * on orphans if we crash during truncate. We actually write i_disksize
895 * into the on-disk inode when writing inodes out, instead of i_size.
896 *
897 * The only time when i_disksize and i_size may be different is when
898 * a truncate is in progress. The only things which change i_disksize
899 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
900 */
901 loff_t i_disksize;
902
903 /*
904 * i_data_sem is for serialising ext4_truncate() against
905 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
906 * data tree are chopped off during truncate. We can't do that in
907 * ext4 because whenever we perform intermediate commits during
908 * truncate, the inode and all the metadata blocks *must* be in a
909 * consistent state which allows truncation of the orphans to restart
910 * during recovery. Hence we must fix the get_block-vs-truncate race
911 * by other means, so we have i_data_sem.
912 */
913 struct rw_semaphore i_data_sem;
914 struct inode vfs_inode;
915 struct jbd2_inode *jinode;
916
917 /*
918 * File creation time. Its function is same as that of
919 * struct timespec i_{a,c,m}time in the generic inode.
920 */
921 struct timespec i_crtime;
922
923 /* mballoc */
924 struct list_head i_prealloc_list;
925 spinlock_t i_prealloc_lock;
926
927 /* extents status tree */
928 struct ext4_es_tree i_es_tree;
929 rwlock_t i_es_lock;
930 struct list_head i_es_lru;
931 unsigned int i_es_lru_nr; /* protected by i_es_lock */
932
933 /* ialloc */
934 ext4_group_t i_last_alloc_group;
935
936 /* allocation reservation info for delalloc */
937 /* In case of bigalloc, these refer to clusters rather than blocks */
938 unsigned int i_reserved_data_blocks;
939 unsigned int i_reserved_meta_blocks;
940 unsigned int i_allocated_meta_blocks;
941 ext4_lblk_t i_da_metadata_calc_last_lblock;
942 int i_da_metadata_calc_len;
943
944 /* on-disk additional length */
945 __u16 i_extra_isize;
946
947 /* Indicate the inline data space. */
948 u16 i_inline_off;
949 u16 i_inline_size;
950
951 #ifdef CONFIG_QUOTA
952 /* quota space reservation, managed internally by quota code */
953 qsize_t i_reserved_quota;
954 #endif
955
956 /* completed IOs that might need unwritten extents handling */
957 struct list_head i_completed_io_list;
958 spinlock_t i_completed_io_lock;
959 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
960 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
961 struct work_struct i_unwritten_work; /* deferred extent conversion */
962
963 spinlock_t i_block_reservation_lock;
964
965 /*
966 * Transactions that contain inode's metadata needed to complete
967 * fsync and fdatasync, respectively.
968 */
969 tid_t i_sync_tid;
970 tid_t i_datasync_tid;
971
972 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
973 __u32 i_csum_seed;
974 };
975
976 /*
977 * File system states
978 */
979 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
980 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
981 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
982
983 /*
984 * Misc. filesystem flags
985 */
986 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
987 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
988 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
989
990 /*
991 * Mount flags set via mount options or defaults
992 */
993 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
994 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
995 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
996 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
997 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
998 #define EXT4_MOUNT_ERRORS_MASK 0x00070
999 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1000 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1001 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1002 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1003 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1004 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1005 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1006 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1007 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1008 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1009 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1010 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1011 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1012 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1013 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1014 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1015 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1016 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1017 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1018 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1019 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1020 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1021 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1022
1023 /*
1024 * Mount flags set either automatically (could not be set by mount option)
1025 * based on per file system feature or property or in special cases such as
1026 * distinguishing between explicit mount option definition and default.
1027 */
1028 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1029 specified delalloc */
1030 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1031 size of blocksize * 8
1032 blocks */
1033
1034 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1035 ~EXT4_MOUNT_##opt
1036 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1037 EXT4_MOUNT_##opt
1038 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1039 EXT4_MOUNT_##opt)
1040
1041 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1042 ~EXT4_MOUNT2_##opt
1043 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1044 EXT4_MOUNT2_##opt
1045 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1046 EXT4_MOUNT2_##opt)
1047
1048 #define ext4_test_and_set_bit __test_and_set_bit_le
1049 #define ext4_set_bit __set_bit_le
1050 #define ext4_set_bit_atomic ext2_set_bit_atomic
1051 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1052 #define ext4_clear_bit __clear_bit_le
1053 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1054 #define ext4_test_bit test_bit_le
1055 #define ext4_find_next_zero_bit find_next_zero_bit_le
1056 #define ext4_find_next_bit find_next_bit_le
1057
1058 extern void ext4_set_bits(void *bm, int cur, int len);
1059
1060 /*
1061 * Maximal mount counts between two filesystem checks
1062 */
1063 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1064 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1065
1066 /*
1067 * Behaviour when detecting errors
1068 */
1069 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1070 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1071 #define EXT4_ERRORS_PANIC 3 /* Panic */
1072 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1073
1074 /* Metadata checksum algorithm codes */
1075 #define EXT4_CRC32C_CHKSUM 1
1076
1077 /*
1078 * Structure of the super block
1079 */
1080 struct ext4_super_block {
1081 /*00*/ __le32 s_inodes_count; /* Inodes count */
1082 __le32 s_blocks_count_lo; /* Blocks count */
1083 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1084 __le32 s_free_blocks_count_lo; /* Free blocks count */
1085 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1086 __le32 s_first_data_block; /* First Data Block */
1087 __le32 s_log_block_size; /* Block size */
1088 __le32 s_log_cluster_size; /* Allocation cluster size */
1089 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1090 __le32 s_clusters_per_group; /* # Clusters per group */
1091 __le32 s_inodes_per_group; /* # Inodes per group */
1092 __le32 s_mtime; /* Mount time */
1093 /*30*/ __le32 s_wtime; /* Write time */
1094 __le16 s_mnt_count; /* Mount count */
1095 __le16 s_max_mnt_count; /* Maximal mount count */
1096 __le16 s_magic; /* Magic signature */
1097 __le16 s_state; /* File system state */
1098 __le16 s_errors; /* Behaviour when detecting errors */
1099 __le16 s_minor_rev_level; /* minor revision level */
1100 /*40*/ __le32 s_lastcheck; /* time of last check */
1101 __le32 s_checkinterval; /* max. time between checks */
1102 __le32 s_creator_os; /* OS */
1103 __le32 s_rev_level; /* Revision level */
1104 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1105 __le16 s_def_resgid; /* Default gid for reserved blocks */
1106 /*
1107 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1108 *
1109 * Note: the difference between the compatible feature set and
1110 * the incompatible feature set is that if there is a bit set
1111 * in the incompatible feature set that the kernel doesn't
1112 * know about, it should refuse to mount the filesystem.
1113 *
1114 * e2fsck's requirements are more strict; if it doesn't know
1115 * about a feature in either the compatible or incompatible
1116 * feature set, it must abort and not try to meddle with
1117 * things it doesn't understand...
1118 */
1119 __le32 s_first_ino; /* First non-reserved inode */
1120 __le16 s_inode_size; /* size of inode structure */
1121 __le16 s_block_group_nr; /* block group # of this superblock */
1122 __le32 s_feature_compat; /* compatible feature set */
1123 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1124 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1125 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1126 /*78*/ char s_volume_name[16]; /* volume name */
1127 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1128 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1129 /*
1130 * Performance hints. Directory preallocation should only
1131 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1132 */
1133 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1134 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1135 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1136 /*
1137 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1138 */
1139 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1140 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1141 __le32 s_journal_dev; /* device number of journal file */
1142 __le32 s_last_orphan; /* start of list of inodes to delete */
1143 __le32 s_hash_seed[4]; /* HTREE hash seed */
1144 __u8 s_def_hash_version; /* Default hash version to use */
1145 __u8 s_jnl_backup_type;
1146 __le16 s_desc_size; /* size of group descriptor */
1147 /*100*/ __le32 s_default_mount_opts;
1148 __le32 s_first_meta_bg; /* First metablock block group */
1149 __le32 s_mkfs_time; /* When the filesystem was created */
1150 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1151 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1152 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1153 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1154 __le32 s_free_blocks_count_hi; /* Free blocks count */
1155 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1156 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1157 __le32 s_flags; /* Miscellaneous flags */
1158 __le16 s_raid_stride; /* RAID stride */
1159 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1160 __le64 s_mmp_block; /* Block for multi-mount protection */
1161 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1162 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1163 __u8 s_checksum_type; /* metadata checksum algorithm used */
1164 __le16 s_reserved_pad;
1165 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1166 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1167 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1168 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1169 snapshot's future use */
1170 __le32 s_snapshot_list; /* inode number of the head of the
1171 on-disk snapshot list */
1172 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1173 __le32 s_error_count; /* number of fs errors */
1174 __le32 s_first_error_time; /* first time an error happened */
1175 __le32 s_first_error_ino; /* inode involved in first error */
1176 __le64 s_first_error_block; /* block involved of first error */
1177 __u8 s_first_error_func[32]; /* function where the error happened */
1178 __le32 s_first_error_line; /* line number where error happened */
1179 __le32 s_last_error_time; /* most recent time of an error */
1180 __le32 s_last_error_ino; /* inode involved in last error */
1181 __le32 s_last_error_line; /* line number where error happened */
1182 __le64 s_last_error_block; /* block involved of last error */
1183 __u8 s_last_error_func[32]; /* function where the error happened */
1184 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1185 __u8 s_mount_opts[64];
1186 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1187 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1188 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1189 __le32 s_reserved[108]; /* Padding to the end of the block */
1190 __le32 s_checksum; /* crc32c(superblock) */
1191 };
1192
1193 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1194
1195 #ifdef __KERNEL__
1196
1197 /*
1198 * run-time mount flags
1199 */
1200 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1201 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1202
1203 /*
1204 * fourth extended-fs super-block data in memory
1205 */
1206 struct ext4_sb_info {
1207 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1208 unsigned long s_inodes_per_block;/* Number of inodes per block */
1209 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1210 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1211 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1212 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1213 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1214 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1215 ext4_group_t s_groups_count; /* Number of groups in the fs */
1216 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1217 unsigned long s_overhead; /* # of fs overhead clusters */
1218 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1219 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1220 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1221 struct buffer_head * s_sbh; /* Buffer containing the super block */
1222 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1223 struct buffer_head **s_group_desc;
1224 unsigned int s_mount_opt;
1225 unsigned int s_mount_opt2;
1226 unsigned int s_mount_flags;
1227 unsigned int s_def_mount_opt;
1228 ext4_fsblk_t s_sb_block;
1229 atomic64_t s_resv_clusters;
1230 kuid_t s_resuid;
1231 kgid_t s_resgid;
1232 unsigned short s_mount_state;
1233 unsigned short s_pad;
1234 int s_addr_per_block_bits;
1235 int s_desc_per_block_bits;
1236 int s_inode_size;
1237 int s_first_ino;
1238 unsigned int s_inode_readahead_blks;
1239 unsigned int s_inode_goal;
1240 spinlock_t s_next_gen_lock;
1241 u32 s_next_generation;
1242 u32 s_hash_seed[4];
1243 int s_def_hash_version;
1244 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1245 struct percpu_counter s_freeclusters_counter;
1246 struct percpu_counter s_freeinodes_counter;
1247 struct percpu_counter s_dirs_counter;
1248 struct percpu_counter s_dirtyclusters_counter;
1249 struct blockgroup_lock *s_blockgroup_lock;
1250 struct proc_dir_entry *s_proc;
1251 struct kobject s_kobj;
1252 struct completion s_kobj_unregister;
1253 struct super_block *s_sb;
1254
1255 /* Journaling */
1256 struct journal_s *s_journal;
1257 struct list_head s_orphan;
1258 struct mutex s_orphan_lock;
1259 unsigned long s_resize_flags; /* Flags indicating if there
1260 is a resizer */
1261 unsigned long s_commit_interval;
1262 u32 s_max_batch_time;
1263 u32 s_min_batch_time;
1264 struct block_device *journal_bdev;
1265 #ifdef CONFIG_QUOTA
1266 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1267 int s_jquota_fmt; /* Format of quota to use */
1268 #endif
1269 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1270 struct rb_root system_blks;
1271
1272 #ifdef EXTENTS_STATS
1273 /* ext4 extents stats */
1274 unsigned long s_ext_min;
1275 unsigned long s_ext_max;
1276 unsigned long s_depth_max;
1277 spinlock_t s_ext_stats_lock;
1278 unsigned long s_ext_blocks;
1279 unsigned long s_ext_extents;
1280 #endif
1281
1282 /* for buddy allocator */
1283 struct ext4_group_info ***s_group_info;
1284 struct inode *s_buddy_cache;
1285 spinlock_t s_md_lock;
1286 unsigned short *s_mb_offsets;
1287 unsigned int *s_mb_maxs;
1288 unsigned int s_group_info_size;
1289
1290 /* tunables */
1291 unsigned long s_stripe;
1292 unsigned int s_mb_stream_request;
1293 unsigned int s_mb_max_to_scan;
1294 unsigned int s_mb_min_to_scan;
1295 unsigned int s_mb_stats;
1296 unsigned int s_mb_order2_reqs;
1297 unsigned int s_mb_group_prealloc;
1298 unsigned int s_max_writeback_mb_bump;
1299 unsigned int s_max_dir_size_kb;
1300 /* where last allocation was done - for stream allocation */
1301 unsigned long s_mb_last_group;
1302 unsigned long s_mb_last_start;
1303
1304 /* stats for buddy allocator */
1305 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1306 atomic_t s_bal_success; /* we found long enough chunks */
1307 atomic_t s_bal_allocated; /* in blocks */
1308 atomic_t s_bal_ex_scanned; /* total extents scanned */
1309 atomic_t s_bal_goals; /* goal hits */
1310 atomic_t s_bal_breaks; /* too long searches */
1311 atomic_t s_bal_2orders; /* 2^order hits */
1312 spinlock_t s_bal_lock;
1313 unsigned long s_mb_buddies_generated;
1314 unsigned long long s_mb_generation_time;
1315 atomic_t s_mb_lost_chunks;
1316 atomic_t s_mb_preallocated;
1317 atomic_t s_mb_discarded;
1318 atomic_t s_lock_busy;
1319
1320 /* locality groups */
1321 struct ext4_locality_group __percpu *s_locality_groups;
1322
1323 /* for write statistics */
1324 unsigned long s_sectors_written_start;
1325 u64 s_kbytes_written;
1326
1327 /* the size of zero-out chunk */
1328 unsigned int s_extent_max_zeroout_kb;
1329
1330 unsigned int s_log_groups_per_flex;
1331 struct flex_groups *s_flex_groups;
1332 ext4_group_t s_flex_groups_allocated;
1333
1334 /* workqueue for dio unwritten */
1335 struct workqueue_struct *dio_unwritten_wq;
1336
1337 /* timer for periodic error stats printing */
1338 struct timer_list s_err_report;
1339
1340 /* Lazy inode table initialization info */
1341 struct ext4_li_request *s_li_request;
1342 /* Wait multiplier for lazy initialization thread */
1343 unsigned int s_li_wait_mult;
1344
1345 /* Kernel thread for multiple mount protection */
1346 struct task_struct *s_mmp_tsk;
1347
1348 /* record the last minlen when FITRIM is called. */
1349 atomic_t s_last_trim_minblks;
1350
1351 /* Reference to checksum algorithm driver via cryptoapi */
1352 struct crypto_shash *s_chksum_driver;
1353
1354 /* Precomputed FS UUID checksum for seeding other checksums */
1355 __u32 s_csum_seed;
1356
1357 /* Reclaim extents from extent status tree */
1358 struct shrinker s_es_shrinker;
1359 struct list_head s_es_lru;
1360 struct percpu_counter s_extent_cache_cnt;
1361 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1362 };
1363
1364 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1365 {
1366 return sb->s_fs_info;
1367 }
1368 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1369 {
1370 return container_of(inode, struct ext4_inode_info, vfs_inode);
1371 }
1372
1373 static inline struct timespec ext4_current_time(struct inode *inode)
1374 {
1375 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1376 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1377 }
1378
1379 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1380 {
1381 return ino == EXT4_ROOT_INO ||
1382 ino == EXT4_USR_QUOTA_INO ||
1383 ino == EXT4_GRP_QUOTA_INO ||
1384 ino == EXT4_BOOT_LOADER_INO ||
1385 ino == EXT4_JOURNAL_INO ||
1386 ino == EXT4_RESIZE_INO ||
1387 (ino >= EXT4_FIRST_INO(sb) &&
1388 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1389 }
1390
1391 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1392 struct ext4_io_end *io_end)
1393 {
1394 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1395 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1396 atomic_inc(&EXT4_I(inode)->i_unwritten);
1397 }
1398 }
1399
1400 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1401 {
1402 return inode->i_private;
1403 }
1404
1405 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1406 {
1407 inode->i_private = io;
1408 }
1409
1410 /*
1411 * Inode dynamic state flags
1412 */
1413 enum {
1414 EXT4_STATE_JDATA, /* journaled data exists */
1415 EXT4_STATE_NEW, /* inode is newly created */
1416 EXT4_STATE_XATTR, /* has in-inode xattrs */
1417 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1418 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1419 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1420 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1421 EXT4_STATE_NEWENTRY, /* File just added to dir */
1422 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1423 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1424 nolocking */
1425 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1426 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1427 };
1428
1429 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1430 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1431 { \
1432 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1433 } \
1434 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1435 { \
1436 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1437 } \
1438 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1439 { \
1440 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1441 }
1442
1443 EXT4_INODE_BIT_FNS(flag, flags, 0)
1444 #if (BITS_PER_LONG < 64)
1445 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1446
1447 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1448 {
1449 (ei)->i_state_flags = 0;
1450 }
1451 #else
1452 EXT4_INODE_BIT_FNS(state, flags, 32)
1453
1454 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1455 {
1456 /* We depend on the fact that callers will set i_flags */
1457 }
1458 #endif
1459 #else
1460 /* Assume that user mode programs are passing in an ext4fs superblock, not
1461 * a kernel struct super_block. This will allow us to call the feature-test
1462 * macros from user land. */
1463 #define EXT4_SB(sb) (sb)
1464 #endif
1465
1466 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1467
1468 /*
1469 * Codes for operating systems
1470 */
1471 #define EXT4_OS_LINUX 0
1472 #define EXT4_OS_HURD 1
1473 #define EXT4_OS_MASIX 2
1474 #define EXT4_OS_FREEBSD 3
1475 #define EXT4_OS_LITES 4
1476
1477 /*
1478 * Revision levels
1479 */
1480 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1481 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1482
1483 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1484 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1485
1486 #define EXT4_GOOD_OLD_INODE_SIZE 128
1487
1488 /*
1489 * Feature set definitions
1490 */
1491
1492 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1493 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1494 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1495 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1496 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1497 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1498 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1499 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1500 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1501 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1502 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1503 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1504 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1505 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1506 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1507 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1508 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1509 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1510
1511 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1512 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1513 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1514 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1515 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1516 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1517
1518 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1519 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1520 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1521 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1522 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1523 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1524 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1525 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1526 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1527 /*
1528 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1529 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1530 * all other data structures' checksums. However, the METADATA_CSUM and
1531 * GDT_CSUM bits are mutually exclusive.
1532 */
1533 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1534
1535 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1536 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1537 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1538 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1539 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1540 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1541 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1542 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1543 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1544 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1545 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1546 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1547 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1548 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1549
1550 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1551 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1552 EXT4_FEATURE_INCOMPAT_META_BG)
1553 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1554 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1555 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1556
1557 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1558 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1559 EXT4_FEATURE_INCOMPAT_RECOVER| \
1560 EXT4_FEATURE_INCOMPAT_META_BG)
1561 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1562 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1563 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1564
1565 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1566 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1567 EXT4_FEATURE_INCOMPAT_RECOVER| \
1568 EXT4_FEATURE_INCOMPAT_META_BG| \
1569 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1570 EXT4_FEATURE_INCOMPAT_64BIT| \
1571 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1572 EXT4_FEATURE_INCOMPAT_MMP | \
1573 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1574 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1575 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1576 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1577 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1578 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1579 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1580 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1581 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1582 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1583 EXT4_FEATURE_RO_COMPAT_QUOTA)
1584
1585 /*
1586 * Default values for user and/or group using reserved blocks
1587 */
1588 #define EXT4_DEF_RESUID 0
1589 #define EXT4_DEF_RESGID 0
1590
1591 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1592
1593 /*
1594 * Default mount options
1595 */
1596 #define EXT4_DEFM_DEBUG 0x0001
1597 #define EXT4_DEFM_BSDGROUPS 0x0002
1598 #define EXT4_DEFM_XATTR_USER 0x0004
1599 #define EXT4_DEFM_ACL 0x0008
1600 #define EXT4_DEFM_UID16 0x0010
1601 #define EXT4_DEFM_JMODE 0x0060
1602 #define EXT4_DEFM_JMODE_DATA 0x0020
1603 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1604 #define EXT4_DEFM_JMODE_WBACK 0x0060
1605 #define EXT4_DEFM_NOBARRIER 0x0100
1606 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1607 #define EXT4_DEFM_DISCARD 0x0400
1608 #define EXT4_DEFM_NODELALLOC 0x0800
1609
1610 /*
1611 * Default journal batch times
1612 */
1613 #define EXT4_DEF_MIN_BATCH_TIME 0
1614 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1615
1616 /*
1617 * Minimum number of groups in a flexgroup before we separate out
1618 * directories into the first block group of a flexgroup
1619 */
1620 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1621
1622 /*
1623 * Structure of a directory entry
1624 */
1625 #define EXT4_NAME_LEN 255
1626
1627 struct ext4_dir_entry {
1628 __le32 inode; /* Inode number */
1629 __le16 rec_len; /* Directory entry length */
1630 __le16 name_len; /* Name length */
1631 char name[EXT4_NAME_LEN]; /* File name */
1632 };
1633
1634 /*
1635 * The new version of the directory entry. Since EXT4 structures are
1636 * stored in intel byte order, and the name_len field could never be
1637 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1638 * file_type field.
1639 */
1640 struct ext4_dir_entry_2 {
1641 __le32 inode; /* Inode number */
1642 __le16 rec_len; /* Directory entry length */
1643 __u8 name_len; /* Name length */
1644 __u8 file_type;
1645 char name[EXT4_NAME_LEN]; /* File name */
1646 };
1647
1648 /*
1649 * This is a bogus directory entry at the end of each leaf block that
1650 * records checksums.
1651 */
1652 struct ext4_dir_entry_tail {
1653 __le32 det_reserved_zero1; /* Pretend to be unused */
1654 __le16 det_rec_len; /* 12 */
1655 __u8 det_reserved_zero2; /* Zero name length */
1656 __u8 det_reserved_ft; /* 0xDE, fake file type */
1657 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1658 };
1659
1660 #define EXT4_DIRENT_TAIL(block, blocksize) \
1661 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1662 ((blocksize) - \
1663 sizeof(struct ext4_dir_entry_tail))))
1664
1665 /*
1666 * Ext4 directory file types. Only the low 3 bits are used. The
1667 * other bits are reserved for now.
1668 */
1669 #define EXT4_FT_UNKNOWN 0
1670 #define EXT4_FT_REG_FILE 1
1671 #define EXT4_FT_DIR 2
1672 #define EXT4_FT_CHRDEV 3
1673 #define EXT4_FT_BLKDEV 4
1674 #define EXT4_FT_FIFO 5
1675 #define EXT4_FT_SOCK 6
1676 #define EXT4_FT_SYMLINK 7
1677
1678 #define EXT4_FT_MAX 8
1679
1680 #define EXT4_FT_DIR_CSUM 0xDE
1681
1682 /*
1683 * EXT4_DIR_PAD defines the directory entries boundaries
1684 *
1685 * NOTE: It must be a multiple of 4
1686 */
1687 #define EXT4_DIR_PAD 4
1688 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1689 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1690 ~EXT4_DIR_ROUND)
1691 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1692
1693 /*
1694 * If we ever get support for fs block sizes > page_size, we'll need
1695 * to remove the #if statements in the next two functions...
1696 */
1697 static inline unsigned int
1698 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1699 {
1700 unsigned len = le16_to_cpu(dlen);
1701
1702 #if (PAGE_CACHE_SIZE >= 65536)
1703 if (len == EXT4_MAX_REC_LEN || len == 0)
1704 return blocksize;
1705 return (len & 65532) | ((len & 3) << 16);
1706 #else
1707 return len;
1708 #endif
1709 }
1710
1711 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1712 {
1713 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1714 BUG();
1715 #if (PAGE_CACHE_SIZE >= 65536)
1716 if (len < 65536)
1717 return cpu_to_le16(len);
1718 if (len == blocksize) {
1719 if (blocksize == 65536)
1720 return cpu_to_le16(EXT4_MAX_REC_LEN);
1721 else
1722 return cpu_to_le16(0);
1723 }
1724 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1725 #else
1726 return cpu_to_le16(len);
1727 #endif
1728 }
1729
1730 /*
1731 * Hash Tree Directory indexing
1732 * (c) Daniel Phillips, 2001
1733 */
1734
1735 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1736 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1737 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1738 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1739 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1740
1741 /* Legal values for the dx_root hash_version field: */
1742
1743 #define DX_HASH_LEGACY 0
1744 #define DX_HASH_HALF_MD4 1
1745 #define DX_HASH_TEA 2
1746 #define DX_HASH_LEGACY_UNSIGNED 3
1747 #define DX_HASH_HALF_MD4_UNSIGNED 4
1748 #define DX_HASH_TEA_UNSIGNED 5
1749
1750 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1751 const void *address, unsigned int length)
1752 {
1753 struct {
1754 struct shash_desc shash;
1755 char ctx[4];
1756 } desc;
1757 int err;
1758
1759 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1760
1761 desc.shash.tfm = sbi->s_chksum_driver;
1762 desc.shash.flags = 0;
1763 *(u32 *)desc.ctx = crc;
1764
1765 err = crypto_shash_update(&desc.shash, address, length);
1766 BUG_ON(err);
1767
1768 return *(u32 *)desc.ctx;
1769 }
1770
1771 #ifdef __KERNEL__
1772
1773 /* hash info structure used by the directory hash */
1774 struct dx_hash_info
1775 {
1776 u32 hash;
1777 u32 minor_hash;
1778 int hash_version;
1779 u32 *seed;
1780 };
1781
1782
1783 /* 32 and 64 bit signed EOF for dx directories */
1784 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1785 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1786
1787
1788 /*
1789 * Control parameters used by ext4_htree_next_block
1790 */
1791 #define HASH_NB_ALWAYS 1
1792
1793
1794 /*
1795 * Describe an inode's exact location on disk and in memory
1796 */
1797 struct ext4_iloc
1798 {
1799 struct buffer_head *bh;
1800 unsigned long offset;
1801 ext4_group_t block_group;
1802 };
1803
1804 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1805 {
1806 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1807 }
1808
1809 /*
1810 * This structure is stuffed into the struct file's private_data field
1811 * for directories. It is where we put information so that we can do
1812 * readdir operations in hash tree order.
1813 */
1814 struct dir_private_info {
1815 struct rb_root root;
1816 struct rb_node *curr_node;
1817 struct fname *extra_fname;
1818 loff_t last_pos;
1819 __u32 curr_hash;
1820 __u32 curr_minor_hash;
1821 __u32 next_hash;
1822 };
1823
1824 /* calculate the first block number of the group */
1825 static inline ext4_fsblk_t
1826 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1827 {
1828 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1829 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1830 }
1831
1832 /*
1833 * Special error return code only used by dx_probe() and its callers.
1834 */
1835 #define ERR_BAD_DX_DIR -75000
1836
1837 /*
1838 * Timeout and state flag for lazy initialization inode thread.
1839 */
1840 #define EXT4_DEF_LI_WAIT_MULT 10
1841 #define EXT4_DEF_LI_MAX_START_DELAY 5
1842 #define EXT4_LAZYINIT_QUIT 0x0001
1843 #define EXT4_LAZYINIT_RUNNING 0x0002
1844
1845 /*
1846 * Lazy inode table initialization info
1847 */
1848 struct ext4_lazy_init {
1849 unsigned long li_state;
1850 struct list_head li_request_list;
1851 struct mutex li_list_mtx;
1852 };
1853
1854 struct ext4_li_request {
1855 struct super_block *lr_super;
1856 struct ext4_sb_info *lr_sbi;
1857 ext4_group_t lr_next_group;
1858 struct list_head lr_request;
1859 unsigned long lr_next_sched;
1860 unsigned long lr_timeout;
1861 };
1862
1863 struct ext4_features {
1864 struct kobject f_kobj;
1865 struct completion f_kobj_unregister;
1866 };
1867
1868 /*
1869 * This structure will be used for multiple mount protection. It will be
1870 * written into the block number saved in the s_mmp_block field in the
1871 * superblock. Programs that check MMP should assume that if
1872 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1873 * to use the filesystem, regardless of how old the timestamp is.
1874 */
1875 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1876 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1877 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1878 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1879
1880 struct mmp_struct {
1881 __le32 mmp_magic; /* Magic number for MMP */
1882 __le32 mmp_seq; /* Sequence no. updated periodically */
1883
1884 /*
1885 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1886 * purposes and do not affect the correctness of the algorithm
1887 */
1888 __le64 mmp_time; /* Time last updated */
1889 char mmp_nodename[64]; /* Node which last updated MMP block */
1890 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1891
1892 /*
1893 * mmp_check_interval is used to verify if the MMP block has been
1894 * updated on the block device. The value is updated based on the
1895 * maximum time to write the MMP block during an update cycle.
1896 */
1897 __le16 mmp_check_interval;
1898
1899 __le16 mmp_pad1;
1900 __le32 mmp_pad2[226];
1901 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1902 };
1903
1904 /* arguments passed to the mmp thread */
1905 struct mmpd_data {
1906 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1907 struct super_block *sb; /* super block of the fs */
1908 };
1909
1910 /*
1911 * Check interval multiplier
1912 * The MMP block is written every update interval and initially checked every
1913 * update interval x the multiplier (the value is then adapted based on the
1914 * write latency). The reason is that writes can be delayed under load and we
1915 * don't want readers to incorrectly assume that the filesystem is no longer
1916 * in use.
1917 */
1918 #define EXT4_MMP_CHECK_MULT 2UL
1919
1920 /*
1921 * Minimum interval for MMP checking in seconds.
1922 */
1923 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1924
1925 /*
1926 * Maximum interval for MMP checking in seconds.
1927 */
1928 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1929
1930 /*
1931 * Function prototypes
1932 */
1933
1934 /*
1935 * Ok, these declarations are also in <linux/kernel.h> but none of the
1936 * ext4 source programs needs to include it so they are duplicated here.
1937 */
1938 # define NORET_TYPE /**/
1939 # define ATTRIB_NORET __attribute__((noreturn))
1940 # define NORET_AND noreturn,
1941
1942 /* bitmap.c */
1943 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1944 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1945 struct ext4_group_desc *gdp,
1946 struct buffer_head *bh, int sz);
1947 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1948 struct ext4_group_desc *gdp,
1949 struct buffer_head *bh, int sz);
1950 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1951 struct ext4_group_desc *gdp,
1952 struct buffer_head *bh);
1953 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1954 struct ext4_group_desc *gdp,
1955 struct buffer_head *bh);
1956
1957 /* balloc.c */
1958 extern void ext4_get_group_no_and_offset(struct super_block *sb,
1959 ext4_fsblk_t blocknr,
1960 ext4_group_t *blockgrpp,
1961 ext4_grpblk_t *offsetp);
1962 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1963 ext4_fsblk_t block);
1964
1965 extern void ext4_validate_block_bitmap(struct super_block *sb,
1966 struct ext4_group_desc *desc,
1967 unsigned int block_group,
1968 struct buffer_head *bh);
1969 extern unsigned int ext4_block_group(struct super_block *sb,
1970 ext4_fsblk_t blocknr);
1971 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1972 ext4_fsblk_t blocknr);
1973 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1974 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1975 ext4_group_t group);
1976 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1977 ext4_fsblk_t goal,
1978 unsigned int flags,
1979 unsigned long *count,
1980 int *errp);
1981 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1982 s64 nclusters, unsigned int flags);
1983 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1984 extern void ext4_check_blocks_bitmap(struct super_block *);
1985 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1986 ext4_group_t block_group,
1987 struct buffer_head ** bh);
1988 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1989
1990 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1991 ext4_group_t block_group);
1992 extern int ext4_wait_block_bitmap(struct super_block *sb,
1993 ext4_group_t block_group,
1994 struct buffer_head *bh);
1995 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1996 ext4_group_t block_group);
1997 extern void ext4_init_block_bitmap(struct super_block *sb,
1998 struct buffer_head *bh,
1999 ext4_group_t group,
2000 struct ext4_group_desc *desc);
2001 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2002 ext4_group_t block_group,
2003 struct ext4_group_desc *gdp);
2004 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
2005 ext4_group_t block_group,
2006 struct ext4_group_desc *gdp);
2007 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2008
2009 /* dir.c */
2010 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2011 struct file *,
2012 struct ext4_dir_entry_2 *,
2013 struct buffer_head *, char *, int,
2014 unsigned int);
2015 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2016 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2017 (de), (bh), (buf), (size), (offset)))
2018 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2019 __u32 minor_hash,
2020 struct ext4_dir_entry_2 *dirent);
2021 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2022 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2023 struct buffer_head *bh,
2024 void *buf, int buf_size,
2025 const char *name, int namelen,
2026 struct ext4_dir_entry_2 **dest_de);
2027 void ext4_insert_dentry(struct inode *inode,
2028 struct ext4_dir_entry_2 *de,
2029 int buf_size,
2030 const char *name, int namelen);
2031 static inline void ext4_update_dx_flag(struct inode *inode)
2032 {
2033 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2034 EXT4_FEATURE_COMPAT_DIR_INDEX))
2035 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2036 }
2037 static unsigned char ext4_filetype_table[] = {
2038 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2039 };
2040
2041 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2042 {
2043 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2044 (filetype >= EXT4_FT_MAX))
2045 return DT_UNKNOWN;
2046
2047 return ext4_filetype_table[filetype];
2048 }
2049
2050 /* fsync.c */
2051 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2052 extern int ext4_flush_unwritten_io(struct inode *);
2053
2054 /* hash.c */
2055 extern int ext4fs_dirhash(const char *name, int len, struct
2056 dx_hash_info *hinfo);
2057
2058 /* ialloc.c */
2059 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2060 const struct qstr *qstr, __u32 goal,
2061 uid_t *owner, int handle_type,
2062 unsigned int line_no, int nblocks);
2063
2064 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2065 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2066 0, 0, 0)
2067 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2068 type, nblocks) \
2069 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2070 (type), __LINE__, (nblocks))
2071
2072
2073 extern void ext4_free_inode(handle_t *, struct inode *);
2074 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2075 extern unsigned long ext4_count_free_inodes(struct super_block *);
2076 extern unsigned long ext4_count_dirs(struct super_block *);
2077 extern void ext4_check_inodes_bitmap(struct super_block *);
2078 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2079 extern int ext4_init_inode_table(struct super_block *sb,
2080 ext4_group_t group, int barrier);
2081 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2082
2083 /* mballoc.c */
2084 extern long ext4_mb_stats;
2085 extern long ext4_mb_max_to_scan;
2086 extern int ext4_mb_init(struct super_block *);
2087 extern int ext4_mb_release(struct super_block *);
2088 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2089 struct ext4_allocation_request *, int *);
2090 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2091 extern void ext4_discard_preallocations(struct inode *);
2092 extern int __init ext4_init_mballoc(void);
2093 extern void ext4_exit_mballoc(void);
2094 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2095 struct buffer_head *bh, ext4_fsblk_t block,
2096 unsigned long count, int flags);
2097 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2098 ext4_group_t ngroups);
2099 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2100 ext4_group_t i, struct ext4_group_desc *desc);
2101 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2102 ext4_fsblk_t block, unsigned long count);
2103 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
2104 unsigned long blkdev_flags);
2105
2106 /* inode.c */
2107 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2108 ext4_lblk_t, int, int *);
2109 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2110 ext4_lblk_t, int, int *);
2111 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2112 struct buffer_head *bh_result, int create);
2113 int ext4_get_block(struct inode *inode, sector_t iblock,
2114 struct buffer_head *bh_result, int create);
2115 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2116 struct buffer_head *bh, int create);
2117 int ext4_walk_page_buffers(handle_t *handle,
2118 struct buffer_head *head,
2119 unsigned from,
2120 unsigned to,
2121 int *partial,
2122 int (*fn)(handle_t *handle,
2123 struct buffer_head *bh));
2124 int do_journal_get_write_access(handle_t *handle,
2125 struct buffer_head *bh);
2126 #define FALL_BACK_TO_NONDELALLOC 1
2127 #define CONVERT_INLINE_DATA 2
2128
2129 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2130 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2131 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2132 extern int ext4_setattr(struct dentry *, struct iattr *);
2133 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2134 struct kstat *stat);
2135 extern void ext4_evict_inode(struct inode *);
2136 extern void ext4_clear_inode(struct inode *);
2137 extern int ext4_sync_inode(handle_t *, struct inode *);
2138 extern void ext4_dirty_inode(struct inode *, int);
2139 extern int ext4_change_inode_journal_flag(struct inode *, int);
2140 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2141 extern int ext4_can_truncate(struct inode *inode);
2142 extern void ext4_truncate(struct inode *);
2143 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2144 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2145 extern void ext4_set_inode_flags(struct inode *);
2146 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2147 extern int ext4_alloc_da_blocks(struct inode *inode);
2148 extern void ext4_set_aops(struct inode *inode);
2149 extern int ext4_writepage_trans_blocks(struct inode *);
2150 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2151 extern int ext4_discard_partial_page_buffers(handle_t *handle,
2152 struct address_space *mapping, loff_t from,
2153 loff_t length, int flags);
2154 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2155 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2156 extern void ext4_da_update_reserve_space(struct inode *inode,
2157 int used, int quota_claim);
2158
2159 /* indirect.c */
2160 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2161 struct ext4_map_blocks *map, int flags);
2162 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2163 const struct iovec *iov, loff_t offset,
2164 unsigned long nr_segs);
2165 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2166 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2167 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2168 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
2169 ext4_lblk_t first, ext4_lblk_t stop);
2170
2171 /* ioctl.c */
2172 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2173 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2174
2175 /* migrate.c */
2176 extern int ext4_ext_migrate(struct inode *);
2177 extern int ext4_ind_migrate(struct inode *inode);
2178
2179 /* namei.c */
2180 extern int ext4_dirent_csum_verify(struct inode *inode,
2181 struct ext4_dir_entry *dirent);
2182 extern int ext4_orphan_add(handle_t *, struct inode *);
2183 extern int ext4_orphan_del(handle_t *, struct inode *);
2184 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2185 __u32 start_minor_hash, __u32 *next_hash);
2186 extern int search_dir(struct buffer_head *bh,
2187 char *search_buf,
2188 int buf_size,
2189 struct inode *dir,
2190 const struct qstr *d_name,
2191 unsigned int offset,
2192 struct ext4_dir_entry_2 **res_dir);
2193 extern int ext4_generic_delete_entry(handle_t *handle,
2194 struct inode *dir,
2195 struct ext4_dir_entry_2 *de_del,
2196 struct buffer_head *bh,
2197 void *entry_buf,
2198 int buf_size,
2199 int csum_size);
2200
2201 /* resize.c */
2202 extern int ext4_group_add(struct super_block *sb,
2203 struct ext4_new_group_data *input);
2204 extern int ext4_group_extend(struct super_block *sb,
2205 struct ext4_super_block *es,
2206 ext4_fsblk_t n_blocks_count);
2207 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2208
2209 /* super.c */
2210 extern int ext4_calculate_overhead(struct super_block *sb);
2211 extern int ext4_superblock_csum_verify(struct super_block *sb,
2212 struct ext4_super_block *es);
2213 extern void ext4_superblock_csum_set(struct super_block *sb);
2214 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2215 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2216 extern void ext4_kvfree(void *ptr);
2217 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2218 ext4_group_t ngroup);
2219 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2220 char nbuf[16]);
2221 extern __printf(4, 5)
2222 void __ext4_error(struct super_block *, const char *, unsigned int,
2223 const char *, ...);
2224 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2225 __LINE__, ## message)
2226 extern __printf(5, 6)
2227 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2228 const char *, ...);
2229 extern __printf(5, 6)
2230 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2231 const char *, ...);
2232 extern void __ext4_std_error(struct super_block *, const char *,
2233 unsigned int, int);
2234 extern __printf(4, 5)
2235 void __ext4_abort(struct super_block *, const char *, unsigned int,
2236 const char *, ...);
2237 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2238 __LINE__, ## message)
2239 extern __printf(4, 5)
2240 void __ext4_warning(struct super_block *, const char *, unsigned int,
2241 const char *, ...);
2242 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2243 __LINE__, ## message)
2244 extern __printf(3, 4)
2245 void ext4_msg(struct super_block *, const char *, const char *, ...);
2246 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2247 const char *, unsigned int, const char *);
2248 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2249 __LINE__, msg)
2250 extern __printf(7, 8)
2251 void __ext4_grp_locked_error(const char *, unsigned int,
2252 struct super_block *, ext4_group_t,
2253 unsigned long, ext4_fsblk_t,
2254 const char *, ...);
2255 #define ext4_grp_locked_error(sb, grp, message...) \
2256 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2257 extern void ext4_update_dynamic_rev(struct super_block *sb);
2258 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2259 __u32 compat);
2260 extern int ext4_update_rocompat_feature(handle_t *handle,
2261 struct super_block *sb, __u32 rocompat);
2262 extern int ext4_update_incompat_feature(handle_t *handle,
2263 struct super_block *sb, __u32 incompat);
2264 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2265 struct ext4_group_desc *bg);
2266 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2267 struct ext4_group_desc *bg);
2268 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2269 struct ext4_group_desc *bg);
2270 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2271 struct ext4_group_desc *bg);
2272 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2273 struct ext4_group_desc *bg);
2274 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2275 struct ext4_group_desc *bg);
2276 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2277 struct ext4_group_desc *bg);
2278 extern void ext4_block_bitmap_set(struct super_block *sb,
2279 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2280 extern void ext4_inode_bitmap_set(struct super_block *sb,
2281 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2282 extern void ext4_inode_table_set(struct super_block *sb,
2283 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2284 extern void ext4_free_group_clusters_set(struct super_block *sb,
2285 struct ext4_group_desc *bg,
2286 __u32 count);
2287 extern void ext4_free_inodes_set(struct super_block *sb,
2288 struct ext4_group_desc *bg, __u32 count);
2289 extern void ext4_used_dirs_set(struct super_block *sb,
2290 struct ext4_group_desc *bg, __u32 count);
2291 extern void ext4_itable_unused_set(struct super_block *sb,
2292 struct ext4_group_desc *bg, __u32 count);
2293 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2294 struct ext4_group_desc *gdp);
2295 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2296 struct ext4_group_desc *gdp);
2297 extern int ext4_register_li_request(struct super_block *sb,
2298 ext4_group_t first_not_zeroed);
2299
2300 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2301 {
2302 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2303 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2304 (EXT4_SB(sb)->s_chksum_driver != NULL);
2305 }
2306
2307 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2308 {
2309 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2310 le32_to_cpu(es->s_blocks_count_lo);
2311 }
2312
2313 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2314 {
2315 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2316 le32_to_cpu(es->s_r_blocks_count_lo);
2317 }
2318
2319 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2320 {
2321 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2322 le32_to_cpu(es->s_free_blocks_count_lo);
2323 }
2324
2325 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2326 ext4_fsblk_t blk)
2327 {
2328 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2329 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2330 }
2331
2332 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2333 ext4_fsblk_t blk)
2334 {
2335 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2336 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2337 }
2338
2339 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2340 ext4_fsblk_t blk)
2341 {
2342 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2343 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2344 }
2345
2346 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2347 {
2348 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2349 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2350 le32_to_cpu(raw_inode->i_size_lo);
2351 else
2352 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2353 }
2354
2355 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2356 {
2357 raw_inode->i_size_lo = cpu_to_le32(i_size);
2358 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2359 }
2360
2361 static inline
2362 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2363 ext4_group_t group)
2364 {
2365 struct ext4_group_info ***grp_info;
2366 long indexv, indexh;
2367 grp_info = EXT4_SB(sb)->s_group_info;
2368 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2369 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2370 return grp_info[indexv][indexh];
2371 }
2372
2373 /*
2374 * Reading s_groups_count requires using smp_rmb() afterwards. See
2375 * the locking protocol documented in the comments of ext4_group_add()
2376 * in resize.c
2377 */
2378 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2379 {
2380 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2381
2382 smp_rmb();
2383 return ngroups;
2384 }
2385
2386 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2387 ext4_group_t block_group)
2388 {
2389 return block_group >> sbi->s_log_groups_per_flex;
2390 }
2391
2392 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2393 {
2394 return 1 << sbi->s_log_groups_per_flex;
2395 }
2396
2397 #define ext4_std_error(sb, errno) \
2398 do { \
2399 if ((errno)) \
2400 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2401 } while (0)
2402
2403 #ifdef CONFIG_SMP
2404 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2405 * counters. So we need to make sure we have free clusters more
2406 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2407 */
2408 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2409 #else
2410 #define EXT4_FREECLUSTERS_WATERMARK 0
2411 #endif
2412
2413 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2414 {
2415 /*
2416 * XXX: replace with spinlock if seen contended -bzzz
2417 */
2418 down_write(&EXT4_I(inode)->i_data_sem);
2419 if (newsize > EXT4_I(inode)->i_disksize)
2420 EXT4_I(inode)->i_disksize = newsize;
2421 up_write(&EXT4_I(inode)->i_data_sem);
2422 return ;
2423 }
2424
2425 struct ext4_group_info {
2426 unsigned long bb_state;
2427 struct rb_root bb_free_root;
2428 ext4_grpblk_t bb_first_free; /* first free block */
2429 ext4_grpblk_t bb_free; /* total free blocks */
2430 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2431 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2432 struct list_head bb_prealloc_list;
2433 #ifdef DOUBLE_CHECK
2434 void *bb_bitmap;
2435 #endif
2436 struct rw_semaphore alloc_sem;
2437 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2438 * regions, index is order.
2439 * bb_counters[3] = 5 means
2440 * 5 free 8-block regions. */
2441 };
2442
2443 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2444 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2445
2446 #define EXT4_MB_GRP_NEED_INIT(grp) \
2447 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2448
2449 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2450 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2451 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2452 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2453 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2454 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2455
2456 #define EXT4_MAX_CONTENTION 8
2457 #define EXT4_CONTENTION_THRESHOLD 2
2458
2459 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2460 ext4_group_t group)
2461 {
2462 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2463 }
2464
2465 /*
2466 * Returns true if the filesystem is busy enough that attempts to
2467 * access the block group locks has run into contention.
2468 */
2469 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2470 {
2471 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2472 }
2473
2474 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2475 {
2476 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2477 if (spin_trylock(lock))
2478 /*
2479 * We're able to grab the lock right away, so drop the
2480 * lock contention counter.
2481 */
2482 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2483 else {
2484 /*
2485 * The lock is busy, so bump the contention counter,
2486 * and then wait on the spin lock.
2487 */
2488 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2489 EXT4_MAX_CONTENTION);
2490 spin_lock(lock);
2491 }
2492 }
2493
2494 static inline void ext4_unlock_group(struct super_block *sb,
2495 ext4_group_t group)
2496 {
2497 spin_unlock(ext4_group_lock_ptr(sb, group));
2498 }
2499
2500 /*
2501 * Block validity checking
2502 */
2503 #define ext4_check_indirect_blockref(inode, bh) \
2504 ext4_check_blockref(__func__, __LINE__, inode, \
2505 (__le32 *)(bh)->b_data, \
2506 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2507
2508 #define ext4_ind_check_inode(inode) \
2509 ext4_check_blockref(__func__, __LINE__, inode, \
2510 EXT4_I(inode)->i_data, \
2511 EXT4_NDIR_BLOCKS)
2512
2513 /*
2514 * Inodes and files operations
2515 */
2516
2517 /* dir.c */
2518 extern const struct file_operations ext4_dir_operations;
2519
2520 /* file.c */
2521 extern const struct inode_operations ext4_file_inode_operations;
2522 extern const struct file_operations ext4_file_operations;
2523 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2524 extern void ext4_unwritten_wait(struct inode *inode);
2525
2526 /* inline.c */
2527 extern int ext4_has_inline_data(struct inode *inode);
2528 extern int ext4_get_inline_size(struct inode *inode);
2529 extern int ext4_get_max_inline_size(struct inode *inode);
2530 extern int ext4_find_inline_data_nolock(struct inode *inode);
2531 extern void ext4_write_inline_data(struct inode *inode,
2532 struct ext4_iloc *iloc,
2533 void *buffer, loff_t pos,
2534 unsigned int len);
2535 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2536 unsigned int len);
2537 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2538 unsigned int len);
2539 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2540
2541 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2542 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2543 struct inode *inode,
2544 loff_t pos, unsigned len,
2545 unsigned flags,
2546 struct page **pagep);
2547 extern int ext4_write_inline_data_end(struct inode *inode,
2548 loff_t pos, unsigned len,
2549 unsigned copied,
2550 struct page *page);
2551 extern struct buffer_head *
2552 ext4_journalled_write_inline_data(struct inode *inode,
2553 unsigned len,
2554 struct page *page);
2555 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2556 struct inode *inode,
2557 loff_t pos, unsigned len,
2558 unsigned flags,
2559 struct page **pagep,
2560 void **fsdata);
2561 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2562 unsigned len, unsigned copied,
2563 struct page *page);
2564 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2565 struct inode *inode);
2566 extern int ext4_try_create_inline_dir(handle_t *handle,
2567 struct inode *parent,
2568 struct inode *inode);
2569 extern int ext4_read_inline_dir(struct file *filp,
2570 void *dirent, filldir_t filldir,
2571 int *has_inline_data);
2572 extern int htree_inlinedir_to_tree(struct file *dir_file,
2573 struct inode *dir, ext4_lblk_t block,
2574 struct dx_hash_info *hinfo,
2575 __u32 start_hash, __u32 start_minor_hash,
2576 int *has_inline_data);
2577 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2578 const struct qstr *d_name,
2579 struct ext4_dir_entry_2 **res_dir,
2580 int *has_inline_data);
2581 extern int ext4_delete_inline_entry(handle_t *handle,
2582 struct inode *dir,
2583 struct ext4_dir_entry_2 *de_del,
2584 struct buffer_head *bh,
2585 int *has_inline_data);
2586 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2587 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2588 struct ext4_dir_entry_2 **parent_de,
2589 int *retval);
2590 extern int ext4_inline_data_fiemap(struct inode *inode,
2591 struct fiemap_extent_info *fieinfo,
2592 int *has_inline);
2593 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2594 struct inode *inode,
2595 int needed);
2596 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2597
2598 extern int ext4_convert_inline_data(struct inode *inode);
2599
2600 /* namei.c */
2601 extern const struct inode_operations ext4_dir_inode_operations;
2602 extern const struct inode_operations ext4_special_inode_operations;
2603 extern struct dentry *ext4_get_parent(struct dentry *child);
2604 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2605 struct ext4_dir_entry_2 *de,
2606 int blocksize, int csum_size,
2607 unsigned int parent_ino, int dotdot_real_len);
2608 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2609 unsigned int blocksize);
2610 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2611 struct inode *inode,
2612 struct buffer_head *bh);
2613 #define S_SHIFT 12
2614 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2615 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2616 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2617 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2618 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2619 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2620 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2621 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2622 };
2623
2624 static inline void ext4_set_de_type(struct super_block *sb,
2625 struct ext4_dir_entry_2 *de,
2626 umode_t mode) {
2627 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2628 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2629 }
2630
2631
2632 /* symlink.c */
2633 extern const struct inode_operations ext4_symlink_inode_operations;
2634 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2635
2636 /* block_validity */
2637 extern void ext4_release_system_zone(struct super_block *sb);
2638 extern int ext4_setup_system_zone(struct super_block *sb);
2639 extern int __init ext4_init_system_zone(void);
2640 extern void ext4_exit_system_zone(void);
2641 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2642 ext4_fsblk_t start_blk,
2643 unsigned int count);
2644 extern int ext4_check_blockref(const char *, unsigned int,
2645 struct inode *, __le32 *, unsigned int);
2646
2647 /* extents.c */
2648 struct ext4_ext_path;
2649 struct ext4_extent;
2650
2651 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2652 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2653 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2654 int chunk);
2655 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2656 struct ext4_map_blocks *map, int flags);
2657 extern void ext4_ext_truncate(handle_t *, struct inode *);
2658 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2659 ext4_lblk_t end);
2660 extern void ext4_ext_init(struct super_block *);
2661 extern void ext4_ext_release(struct super_block *);
2662 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2663 loff_t len);
2664 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2665 ssize_t len);
2666 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2667 struct ext4_map_blocks *map, int flags);
2668 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2669 ext4_lblk_t lblocks);
2670 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2671 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2672 int num,
2673 struct ext4_ext_path *path);
2674 extern int ext4_can_extents_be_merged(struct inode *inode,
2675 struct ext4_extent *ex1,
2676 struct ext4_extent *ex2);
2677 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2678 struct ext4_ext_path *,
2679 struct ext4_extent *, int);
2680 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2681 struct ext4_ext_path *);
2682 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2683 extern int ext4_ext_check_inode(struct inode *inode);
2684 extern int ext4_find_delalloc_range(struct inode *inode,
2685 ext4_lblk_t lblk_start,
2686 ext4_lblk_t lblk_end);
2687 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2688 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2689 __u64 start, __u64 len);
2690
2691
2692 /* move_extent.c */
2693 extern void ext4_double_down_write_data_sem(struct inode *first,
2694 struct inode *second);
2695 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2696 struct inode *donor_inode);
2697 void ext4_inode_double_lock(struct inode *inode1, struct inode *inode2);
2698 void ext4_inode_double_unlock(struct inode *inode1, struct inode *inode2);
2699 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2700 __u64 start_orig, __u64 start_donor,
2701 __u64 len, __u64 *moved_len);
2702
2703 /* page-io.c */
2704 extern int __init ext4_init_pageio(void);
2705 extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2706 extern void ext4_exit_pageio(void);
2707 extern void ext4_ioend_shutdown(struct inode *);
2708 extern void ext4_free_io_end(ext4_io_end_t *io);
2709 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2710 extern void ext4_end_io_work(struct work_struct *work);
2711 extern void ext4_io_submit(struct ext4_io_submit *io);
2712 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2713 struct page *page,
2714 int len,
2715 struct writeback_control *wbc);
2716
2717 /* mmp.c */
2718 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2719 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2720 extern int ext4_mmp_csum_verify(struct super_block *sb,
2721 struct mmp_struct *mmp);
2722
2723 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2724 enum ext4_state_bits {
2725 BH_Uninit /* blocks are allocated but uninitialized on disk */
2726 = BH_JBDPrivateStart,
2727 BH_AllocFromCluster, /* allocated blocks were part of already
2728 * allocated cluster. Note that this flag will
2729 * never, ever appear in a buffer_head's state
2730 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2731 * this is used. */
2732 };
2733
2734 BUFFER_FNS(Uninit, uninit)
2735 TAS_BUFFER_FNS(Uninit, uninit)
2736
2737 /*
2738 * Add new method to test whether block and inode bitmaps are properly
2739 * initialized. With uninit_bg reading the block from disk is not enough
2740 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2741 */
2742 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2743
2744 static inline int bitmap_uptodate(struct buffer_head *bh)
2745 {
2746 return (buffer_uptodate(bh) &&
2747 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2748 }
2749 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2750 {
2751 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2752 }
2753
2754 /*
2755 * Disable DIO read nolock optimization, so new dioreaders will be forced
2756 * to grab i_mutex
2757 */
2758 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2759 {
2760 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2761 smp_mb();
2762 }
2763 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2764 {
2765 smp_mb();
2766 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2767 }
2768
2769 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2770
2771 /* For ioend & aio unwritten conversion wait queues */
2772 #define EXT4_WQ_HASH_SZ 37
2773 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2774 EXT4_WQ_HASH_SZ])
2775 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2776 EXT4_WQ_HASH_SZ])
2777 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2778 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2779
2780 #define EXT4_RESIZING 0
2781 extern int ext4_resize_begin(struct super_block *sb);
2782 extern void ext4_resize_end(struct super_block *sb);
2783
2784 #endif /* __KERNEL__ */
2785
2786 #endif /* _EXT4_H */
kernel source for telekom puls (alcatel/mediatek)