1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #ifdef __KERNEL__
33 #include <linux/compat.h>
34 #endif
35
36 /*
37 * The fourth extended filesystem constants/structures
38 */
39
40 /*
41 * Define EXT4FS_DEBUG to produce debug messages
42 */
43 #undef EXT4FS_DEBUG
44
45 /*
46 * Debug code
47 */
48 #ifdef EXT4FS_DEBUG
49 #define ext4_debug(f, a...) \
50 do { \
51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
52 __FILE__, __LINE__, __func__); \
53 printk(KERN_DEBUG f, ## a); \
54 } while (0)
55 #else
56 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
57 #endif
58
59 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
61
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
63 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
64
65 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
66 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
67
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t;
70
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t;
73
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t;
76
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t;
79
80 /*
81 * Flags used in mballoc's allocation_context flags field.
82 *
83 * Also used to show what's going on for debugging purposes when the
84 * flag field is exported via the traceport interface
85 */
86
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE 0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED 0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA 0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST 0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST 0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA 0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC 0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL 0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED 0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC 0x0800
111 /* Use reserved root blocks if needed */
112 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
113
114 struct ext4_allocation_request {
115 /* target inode for block we're allocating */
116 struct inode *inode;
117 /* how many blocks we want to allocate */
118 unsigned int len;
119 /* logical block in target inode */
120 ext4_lblk_t logical;
121 /* the closest logical allocated block to the left */
122 ext4_lblk_t lleft;
123 /* the closest logical allocated block to the right */
124 ext4_lblk_t lright;
125 /* phys. target (a hint) */
126 ext4_fsblk_t goal;
127 /* phys. block for the closest logical allocated block to the left */
128 ext4_fsblk_t pleft;
129 /* phys. block for the closest logical allocated block to the right */
130 ext4_fsblk_t pright;
131 /* flags. see above EXT4_MB_HINT_* */
132 unsigned int flags;
133 };
134
135 /*
136 * Logical to physical block mapping, used by ext4_map_blocks()
137 *
138 * This structure is used to pass requests into ext4_map_blocks() as
139 * well as to store the information returned by ext4_map_blocks(). It
140 * takes less room on the stack than a struct buffer_head.
141 */
142 #define EXT4_MAP_NEW (1 << BH_New)
143 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
144 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
145 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
146 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
147 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
148 * ext4_map_blocks wants to know whether or not the underlying cluster has
149 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
150 * the requested mapping was from previously mapped (or delayed allocated)
151 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
152 * should never appear on buffer_head's state flags.
153 */
154 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
155 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
156 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
157 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
158
159 struct ext4_map_blocks {
160 ext4_fsblk_t m_pblk;
161 ext4_lblk_t m_lblk;
162 unsigned int m_len;
163 unsigned int m_flags;
164 };
165
166 /*
167 * For delayed allocation tracking
168 */
169 struct mpage_da_data {
170 struct inode *inode;
171 sector_t b_blocknr; /* start block number of extent */
172 size_t b_size; /* size of extent */
173 unsigned long b_state; /* state of the extent */
174 unsigned long first_page, next_page; /* extent of pages */
175 struct writeback_control *wbc;
176 int io_done;
177 int pages_written;
178 int retval;
179 };
180
181 /*
182 * Flags for ext4_io_end->flags
183 */
184 #define EXT4_IO_END_UNWRITTEN 0x0001
185 #define EXT4_IO_END_ERROR 0x0002
186 #define EXT4_IO_END_QUEUED 0x0004
187 #define EXT4_IO_END_DIRECT 0x0008
188 #define EXT4_IO_END_IN_FSYNC 0x0010
189
190 struct ext4_io_page {
191 struct page *p_page;
192 atomic_t p_count;
193 };
194
195 #define MAX_IO_PAGES 128
196
197 /*
198 * For converting uninitialized extents on a work queue.
199 *
200 * 'page' is only used from the writepage() path; 'pages' is only used for
201 * buffered writes; they are used to keep page references until conversion
202 * takes place. For AIO/DIO, neither field is filled in.
203 */
204 typedef struct ext4_io_end {
205 struct list_head list; /* per-file finished IO list */
206 struct inode *inode; /* file being written to */
207 unsigned int flag; /* unwritten or not */
208 struct page *page; /* for writepage() path */
209 loff_t offset; /* offset in the file */
210 ssize_t size; /* size of the extent */
211 struct work_struct work; /* data work queue */
212 struct kiocb *iocb; /* iocb struct for AIO */
213 int result; /* error value for AIO */
214 int num_io_pages; /* for writepages() */
215 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
216 } ext4_io_end_t;
217
218 struct ext4_io_submit {
219 int io_op;
220 struct bio *io_bio;
221 ext4_io_end_t *io_end;
222 struct ext4_io_page *io_page;
223 sector_t io_next_block;
224 };
225
226 /*
227 * Special inodes numbers
228 */
229 #define EXT4_BAD_INO 1 /* Bad blocks inode */
230 #define EXT4_ROOT_INO 2 /* Root inode */
231 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
232 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
233 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
234 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
235 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
236 #define EXT4_JOURNAL_INO 8 /* Journal inode */
237
238 /* First non-reserved inode for old ext4 filesystems */
239 #define EXT4_GOOD_OLD_FIRST_INO 11
240
241 /*
242 * Maximal count of links to a file
243 */
244 #define EXT4_LINK_MAX 65000
245
246 /*
247 * Macro-instructions used to manage several block sizes
248 */
249 #define EXT4_MIN_BLOCK_SIZE 1024
250 #define EXT4_MAX_BLOCK_SIZE 65536
251 #define EXT4_MIN_BLOCK_LOG_SIZE 10
252 #define EXT4_MAX_BLOCK_LOG_SIZE 16
253 #ifdef __KERNEL__
254 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
255 #else
256 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
257 #endif
258 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
259 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
260 EXT4_SB(s)->s_cluster_bits)
261 #ifdef __KERNEL__
262 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
263 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
264 #else
265 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
266 #endif
267 #ifdef __KERNEL__
268 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
269 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
270 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
271 #else
272 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
273 EXT4_GOOD_OLD_INODE_SIZE : \
274 (s)->s_inode_size)
275 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
276 EXT4_GOOD_OLD_FIRST_INO : \
277 (s)->s_first_ino)
278 #endif
279 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
280
281 /* Translate a block number to a cluster number */
282 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
283 /* Translate a cluster number to a block number */
284 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
285 /* Translate # of blks to # of clusters */
286 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
287 (sbi)->s_cluster_bits)
288
289 /*
290 * Structure of a blocks group descriptor
291 */
292 struct ext4_group_desc
293 {
294 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
295 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
296 __le32 bg_inode_table_lo; /* Inodes table block */
297 __le16 bg_free_blocks_count_lo;/* Free blocks count */
298 __le16 bg_free_inodes_count_lo;/* Free inodes count */
299 __le16 bg_used_dirs_count_lo; /* Directories count */
300 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
301 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
302 __le16 bg_itable_unused_lo; /* Unused inodes count */
303 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
304 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
305 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
306 __le32 bg_inode_table_hi; /* Inodes table block MSB */
307 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
308 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
309 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
310 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
311 __u32 bg_reserved2[3];
312 };
313
314 /*
315 * Structure of a flex block group info
316 */
317
318 struct flex_groups {
319 atomic64_t free_clusters;
320 atomic_t free_inodes;
321 atomic_t used_dirs;
322 };
323
324 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
325 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
326 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
327
328 /*
329 * Macro-instructions used to manage group descriptors
330 */
331 #define EXT4_MIN_DESC_SIZE 32
332 #define EXT4_MIN_DESC_SIZE_64BIT 64
333 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
334 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
335 #ifdef __KERNEL__
336 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
337 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
338 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
339 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
340 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
341 #else
342 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
343 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
344 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
345 #endif
346
347 /*
348 * Constants relative to the data blocks
349 */
350 #define EXT4_NDIR_BLOCKS 12
351 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
352 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
353 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
354 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
355
356 /*
357 * Inode flags
358 */
359 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
360 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
361 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
362 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
363 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
364 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
365 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
366 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
367 /* Reserved for compression usage... */
368 #define EXT4_DIRTY_FL 0x00000100
369 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
370 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
371 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
372 /* End compression flags --- maybe not all used */
373 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
374 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
375 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
376 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
377 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
378 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
379 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
380 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
381 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
382 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
383 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
384
385 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
386 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
387
388 /* Flags that should be inherited by new inodes from their parent. */
389 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
390 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
391 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
392 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
393
394 /* Flags that are appropriate for regular files (all but dir-specific ones). */
395 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
396
397 /* Flags that are appropriate for non-directories/regular files. */
398 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
399
400 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)401 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
402 {
403 if (S_ISDIR(mode))
404 return flags;
405 else if (S_ISREG(mode))
406 return flags & EXT4_REG_FLMASK;
407 else
408 return flags & EXT4_OTHER_FLMASK;
409 }
410
411 /*
412 * Inode flags used for atomic set/get
413 */
414 enum {
415 EXT4_INODE_SECRM = 0, /* Secure deletion */
416 EXT4_INODE_UNRM = 1, /* Undelete */
417 EXT4_INODE_COMPR = 2, /* Compress file */
418 EXT4_INODE_SYNC = 3, /* Synchronous updates */
419 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
420 EXT4_INODE_APPEND = 5, /* writes to file may only append */
421 EXT4_INODE_NODUMP = 6, /* do not dump file */
422 EXT4_INODE_NOATIME = 7, /* do not update atime */
423 /* Reserved for compression usage... */
424 EXT4_INODE_DIRTY = 8,
425 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
426 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
427 EXT4_INODE_ECOMPR = 11, /* Compression error */
428 /* End compression flags --- maybe not all used */
429 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
430 EXT4_INODE_IMAGIC = 13, /* AFS directory */
431 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
432 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
433 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
434 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
435 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
436 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
437 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
438 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
439 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
440 };
441
442 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
443 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
444 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
445 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
446
447 /*
448 * Since it's pretty easy to mix up bit numbers and hex values, and we
449 * can't do a compile-time test for ENUM values, we use a run-time
450 * test to make sure that EXT4_XXX_FL is consistent with respect to
451 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
452 * out so it won't cost any extra space in the compiled kernel image.
453 * But it's important that these values are the same, since we are
454 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
455 * must be consistent with the values of FS_XXX_FL defined in
456 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
457 * ext4 filesystems, and of course the values defined in e2fsprogs.
458 *
459 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
460 */
ext4_check_flag_values(void)461 static inline void ext4_check_flag_values(void)
462 {
463 CHECK_FLAG_VALUE(SECRM);
464 CHECK_FLAG_VALUE(UNRM);
465 CHECK_FLAG_VALUE(COMPR);
466 CHECK_FLAG_VALUE(SYNC);
467 CHECK_FLAG_VALUE(IMMUTABLE);
468 CHECK_FLAG_VALUE(APPEND);
469 CHECK_FLAG_VALUE(NODUMP);
470 CHECK_FLAG_VALUE(NOATIME);
471 CHECK_FLAG_VALUE(DIRTY);
472 CHECK_FLAG_VALUE(COMPRBLK);
473 CHECK_FLAG_VALUE(NOCOMPR);
474 CHECK_FLAG_VALUE(ECOMPR);
475 CHECK_FLAG_VALUE(INDEX);
476 CHECK_FLAG_VALUE(IMAGIC);
477 CHECK_FLAG_VALUE(JOURNAL_DATA);
478 CHECK_FLAG_VALUE(NOTAIL);
479 CHECK_FLAG_VALUE(DIRSYNC);
480 CHECK_FLAG_VALUE(TOPDIR);
481 CHECK_FLAG_VALUE(HUGE_FILE);
482 CHECK_FLAG_VALUE(EXTENTS);
483 CHECK_FLAG_VALUE(EA_INODE);
484 CHECK_FLAG_VALUE(EOFBLOCKS);
485 CHECK_FLAG_VALUE(RESERVED);
486 }
487
488 /* Used to pass group descriptor data when online resize is done */
489 struct ext4_new_group_input {
490 __u32 group; /* Group number for this data */
491 __u64 block_bitmap; /* Absolute block number of block bitmap */
492 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
493 __u64 inode_table; /* Absolute block number of inode table start */
494 __u32 blocks_count; /* Total number of blocks in this group */
495 __u16 reserved_blocks; /* Number of reserved blocks in this group */
496 __u16 unused;
497 };
498
499 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
500 struct compat_ext4_new_group_input {
501 u32 group;
502 compat_u64 block_bitmap;
503 compat_u64 inode_bitmap;
504 compat_u64 inode_table;
505 u32 blocks_count;
506 u16 reserved_blocks;
507 u16 unused;
508 };
509 #endif
510
511 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
512 struct ext4_new_group_data {
513 __u32 group;
514 __u64 block_bitmap;
515 __u64 inode_bitmap;
516 __u64 inode_table;
517 __u32 blocks_count;
518 __u16 reserved_blocks;
519 __u16 unused;
520 __u32 free_blocks_count;
521 };
522
523 /* Indexes used to index group tables in ext4_new_group_data */
524 enum {
525 BLOCK_BITMAP = 0, /* block bitmap */
526 INODE_BITMAP, /* inode bitmap */
527 INODE_TABLE, /* inode tables */
528 GROUP_TABLE_COUNT,
529 };
530
531 /*
532 * Flags used by ext4_map_blocks()
533 */
534 /* Allocate any needed blocks and/or convert an unitialized
535 extent to be an initialized ext4 */
536 #define EXT4_GET_BLOCKS_CREATE 0x0001
537 /* Request the creation of an unitialized extent */
538 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
539 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
540 EXT4_GET_BLOCKS_CREATE)
541 /* Caller is from the delayed allocation writeout path,
542 so set the magic i_delalloc_reserve_flag after taking the
543 inode allocation semaphore for */
544 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
545 /* caller is from the direct IO path, request to creation of an
546 unitialized extents if not allocated, split the uninitialized
547 extent if blocks has been preallocated already*/
548 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
549 #define EXT4_GET_BLOCKS_CONVERT 0x0010
550 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
551 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
552 /* Convert extent to initialized after IO complete */
553 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
554 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
555 /* Punch out blocks of an extent */
556 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
557 /* Don't normalize allocation size (used for fallocate) */
558 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
559 /* Request will not result in inode size update (user for fallocate) */
560 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
561
562 /*
563 * Flags used by ext4_free_blocks
564 */
565 #define EXT4_FREE_BLOCKS_METADATA 0x0001
566 #define EXT4_FREE_BLOCKS_FORGET 0x0002
567 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
568 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
569 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
570 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
571
572 /*
573 * Flags used by ext4_discard_partial_page_buffers
574 */
575 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
576
577 /*
578 * ioctl commands
579 */
580 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
581 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
582 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
583 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
584 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
585 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
586 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
587 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
588 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
589 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
590 #define EXT4_IOC_MIGRATE _IO('f', 9)
591 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
592 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
593 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
594 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
595 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
596
597 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
598 /*
599 * ioctl commands in 32 bit emulation
600 */
601 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
602 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
603 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
604 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
605 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
606 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
607 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
608 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
609 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
610 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
611 #endif
612
613 /* Max physical block we can address w/o extents */
614 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
615
616 /*
617 * Structure of an inode on the disk
618 */
619 struct ext4_inode {
620 __le16 i_mode; /* File mode */
621 __le16 i_uid; /* Low 16 bits of Owner Uid */
622 __le32 i_size_lo; /* Size in bytes */
623 __le32 i_atime; /* Access time */
624 __le32 i_ctime; /* Inode Change time */
625 __le32 i_mtime; /* Modification time */
626 __le32 i_dtime; /* Deletion Time */
627 __le16 i_gid; /* Low 16 bits of Group Id */
628 __le16 i_links_count; /* Links count */
629 __le32 i_blocks_lo; /* Blocks count */
630 __le32 i_flags; /* File flags */
631 union {
632 struct {
633 __le32 l_i_version;
634 } linux1;
635 struct {
636 __u32 h_i_translator;
637 } hurd1;
638 struct {
639 __u32 m_i_reserved1;
640 } masix1;
641 } osd1; /* OS dependent 1 */
642 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
643 __le32 i_generation; /* File version (for NFS) */
644 __le32 i_file_acl_lo; /* File ACL */
645 __le32 i_size_high;
646 __le32 i_obso_faddr; /* Obsoleted fragment address */
647 union {
648 struct {
649 __le16 l_i_blocks_high; /* were l_i_reserved1 */
650 __le16 l_i_file_acl_high;
651 __le16 l_i_uid_high; /* these 2 fields */
652 __le16 l_i_gid_high; /* were reserved2[0] */
653 __u32 l_i_reserved2;
654 } linux2;
655 struct {
656 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
657 __u16 h_i_mode_high;
658 __u16 h_i_uid_high;
659 __u16 h_i_gid_high;
660 __u32 h_i_author;
661 } hurd2;
662 struct {
663 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
664 __le16 m_i_file_acl_high;
665 __u32 m_i_reserved2[2];
666 } masix2;
667 } osd2; /* OS dependent 2 */
668 __le16 i_extra_isize;
669 __le16 i_pad1;
670 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
671 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
672 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
673 __le32 i_crtime; /* File Creation time */
674 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
675 __le32 i_version_hi; /* high 32 bits for 64-bit version */
676 };
677
678 struct move_extent {
679 __u32 reserved; /* should be zero */
680 __u32 donor_fd; /* donor file descriptor */
681 __u64 orig_start; /* logical start offset in block for orig */
682 __u64 donor_start; /* logical start offset in block for donor */
683 __u64 len; /* block length to be moved */
684 __u64 moved_len; /* moved block length */
685 };
686
687 #define EXT4_EPOCH_BITS 2
688 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
689 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
690
691 /*
692 * Extended fields will fit into an inode if the filesystem was formatted
693 * with large inodes (-I 256 or larger) and there are not currently any EAs
694 * consuming all of the available space. For new inodes we always reserve
695 * enough space for the kernel's known extended fields, but for inodes
696 * created with an old kernel this might not have been the case. None of
697 * the extended inode fields is critical for correct filesystem operation.
698 * This macro checks if a certain field fits in the inode. Note that
699 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
700 */
701 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
702 ((offsetof(typeof(*ext4_inode), field) + \
703 sizeof((ext4_inode)->field)) \
704 <= (EXT4_GOOD_OLD_INODE_SIZE + \
705 (einode)->i_extra_isize)) \
706
ext4_encode_extra_time(struct timespec * time)707 static inline __le32 ext4_encode_extra_time(struct timespec *time)
708 {
709 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
710 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
711 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
712 }
713
ext4_decode_extra_time(struct timespec * time,__le32 extra)714 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
715 {
716 if (sizeof(time->tv_sec) > 4)
717 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
718 << 32;
719 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
720 }
721
722 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
723 do { \
724 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
725 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
726 (raw_inode)->xtime ## _extra = \
727 ext4_encode_extra_time(&(inode)->xtime); \
728 } while (0)
729
730 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
731 do { \
732 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
733 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
734 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
735 (raw_inode)->xtime ## _extra = \
736 ext4_encode_extra_time(&(einode)->xtime); \
737 } while (0)
738
739 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
740 do { \
741 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
742 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
743 ext4_decode_extra_time(&(inode)->xtime, \
744 raw_inode->xtime ## _extra); \
745 else \
746 (inode)->xtime.tv_nsec = 0; \
747 } while (0)
748
749 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
750 do { \
751 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
752 (einode)->xtime.tv_sec = \
753 (signed)le32_to_cpu((raw_inode)->xtime); \
754 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
755 ext4_decode_extra_time(&(einode)->xtime, \
756 raw_inode->xtime ## _extra); \
757 else \
758 (einode)->xtime.tv_nsec = 0; \
759 } while (0)
760
761 #define i_disk_version osd1.linux1.l_i_version
762
763 #if defined(__KERNEL__) || defined(__linux__)
764 #define i_reserved1 osd1.linux1.l_i_reserved1
765 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
766 #define i_blocks_high osd2.linux2.l_i_blocks_high
767 #define i_uid_low i_uid
768 #define i_gid_low i_gid
769 #define i_uid_high osd2.linux2.l_i_uid_high
770 #define i_gid_high osd2.linux2.l_i_gid_high
771 #define i_reserved2 osd2.linux2.l_i_reserved2
772
773 #elif defined(__GNU__)
774
775 #define i_translator osd1.hurd1.h_i_translator
776 #define i_uid_high osd2.hurd2.h_i_uid_high
777 #define i_gid_high osd2.hurd2.h_i_gid_high
778 #define i_author osd2.hurd2.h_i_author
779
780 #elif defined(__masix__)
781
782 #define i_reserved1 osd1.masix1.m_i_reserved1
783 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
784 #define i_reserved2 osd2.masix2.m_i_reserved2
785
786 #endif /* defined(__KERNEL__) || defined(__linux__) */
787
788 /*
789 * storage for cached extent
790 * If ec_len == 0, then the cache is invalid.
791 * If ec_start == 0, then the cache represents a gap (null mapping)
792 */
793 struct ext4_ext_cache {
794 ext4_fsblk_t ec_start;
795 ext4_lblk_t ec_block;
796 __u32 ec_len; /* must be 32bit to return holes */
797 };
798
799 /*
800 * fourth extended file system inode data in memory
801 */
802 struct ext4_inode_info {
803 __le32 i_data[15]; /* unconverted */
804 __u32 i_dtime;
805 ext4_fsblk_t i_file_acl;
806
807 /*
808 * i_block_group is the number of the block group which contains
809 * this file's inode. Constant across the lifetime of the inode,
810 * it is ued for making block allocation decisions - we try to
811 * place a file's data blocks near its inode block, and new inodes
812 * near to their parent directory's inode.
813 */
814 ext4_group_t i_block_group;
815 ext4_lblk_t i_dir_start_lookup;
816 #if (BITS_PER_LONG < 64)
817 unsigned long i_state_flags; /* Dynamic state flags */
818 #endif
819 unsigned long i_flags;
820
821 #ifdef CONFIG_EXT4_FS_XATTR
822 /*
823 * Extended attributes can be read independently of the main file
824 * data. Taking i_mutex even when reading would cause contention
825 * between readers of EAs and writers of regular file data, so
826 * instead we synchronize on xattr_sem when reading or changing
827 * EAs.
828 */
829 struct rw_semaphore xattr_sem;
830 #endif
831
832 struct list_head i_orphan; /* unlinked but open inodes */
833
834 /*
835 * i_disksize keeps track of what the inode size is ON DISK, not
836 * in memory. During truncate, i_size is set to the new size by
837 * the VFS prior to calling ext4_truncate(), but the filesystem won't
838 * set i_disksize to 0 until the truncate is actually under way.
839 *
840 * The intent is that i_disksize always represents the blocks which
841 * are used by this file. This allows recovery to restart truncate
842 * on orphans if we crash during truncate. We actually write i_disksize
843 * into the on-disk inode when writing inodes out, instead of i_size.
844 *
845 * The only time when i_disksize and i_size may be different is when
846 * a truncate is in progress. The only things which change i_disksize
847 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
848 */
849 loff_t i_disksize;
850
851 /*
852 * i_data_sem is for serialising ext4_truncate() against
853 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
854 * data tree are chopped off during truncate. We can't do that in
855 * ext4 because whenever we perform intermediate commits during
856 * truncate, the inode and all the metadata blocks *must* be in a
857 * consistent state which allows truncation of the orphans to restart
858 * during recovery. Hence we must fix the get_block-vs-truncate race
859 * by other means, so we have i_data_sem.
860 */
861 struct rw_semaphore i_data_sem;
862 struct inode vfs_inode;
863 struct jbd2_inode *jinode;
864
865 struct ext4_ext_cache i_cached_extent;
866 /*
867 * File creation time. Its function is same as that of
868 * struct timespec i_{a,c,m}time in the generic inode.
869 */
870 struct timespec i_crtime;
871
872 /* mballoc */
873 struct list_head i_prealloc_list;
874 spinlock_t i_prealloc_lock;
875
876 /* ialloc */
877 ext4_group_t i_last_alloc_group;
878
879 /* allocation reservation info for delalloc */
880 /* In case of bigalloc, these refer to clusters rather than blocks */
881 unsigned int i_reserved_data_blocks;
882 unsigned int i_reserved_meta_blocks;
883 unsigned int i_allocated_meta_blocks;
884 ext4_lblk_t i_da_metadata_calc_last_lblock;
885 int i_da_metadata_calc_len;
886
887 /* on-disk additional length */
888 __u16 i_extra_isize;
889
890 #ifdef CONFIG_QUOTA
891 /* quota space reservation, managed internally by quota code */
892 qsize_t i_reserved_quota;
893 #endif
894
895 /* completed IOs that might need unwritten extents handling */
896 struct list_head i_completed_io_list;
897 spinlock_t i_completed_io_lock;
898 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
899 /* current io_end structure for async DIO write*/
900 ext4_io_end_t *cur_aio_dio;
901 atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
902
903 spinlock_t i_block_reservation_lock;
904
905 /*
906 * Transactions that contain inode's metadata needed to complete
907 * fsync and fdatasync, respectively.
908 */
909 tid_t i_sync_tid;
910 tid_t i_datasync_tid;
911 };
912
913 /*
914 * File system states
915 */
916 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
917 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
918 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
919
920 /*
921 * Misc. filesystem flags
922 */
923 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
924 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
925 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
926
927 /*
928 * Mount flags
929 */
930 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
931 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
932 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
933 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
934 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
935 #define EXT4_MOUNT_ERRORS_MASK 0x00070
936 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
937 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
938 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
939 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
940 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
941 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
942 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
943 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
944 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
945 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
946 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
947 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
948 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
949 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
950 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
951 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
952 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
953 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
954 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
955 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
956 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
957 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
958 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
959 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
960
961 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
962 specified delalloc */
963
964 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
965 ~EXT4_MOUNT_##opt
966 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
967 EXT4_MOUNT_##opt
968 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
969 EXT4_MOUNT_##opt)
970
971 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
972 ~EXT4_MOUNT2_##opt
973 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
974 EXT4_MOUNT2_##opt
975 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
976 EXT4_MOUNT2_##opt)
977
978 #define ext4_test_and_set_bit __test_and_set_bit_le
979 #define ext4_set_bit __set_bit_le
980 #define ext4_set_bit_atomic ext2_set_bit_atomic
981 #define ext4_test_and_clear_bit __test_and_clear_bit_le
982 #define ext4_clear_bit __clear_bit_le
983 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
984 #define ext4_test_bit test_bit_le
985 #define ext4_find_next_zero_bit find_next_zero_bit_le
986 #define ext4_find_next_bit find_next_bit_le
987
988 extern void ext4_set_bits(void *bm, int cur, int len);
989
990 /*
991 * Maximal mount counts between two filesystem checks
992 */
993 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
994 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
995
996 /*
997 * Behaviour when detecting errors
998 */
999 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1000 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1001 #define EXT4_ERRORS_PANIC 3 /* Panic */
1002 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1003
1004 /*
1005 * Structure of the super block
1006 */
1007 struct ext4_super_block {
1008 /*00*/ __le32 s_inodes_count; /* Inodes count */
1009 __le32 s_blocks_count_lo; /* Blocks count */
1010 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1011 __le32 s_free_blocks_count_lo; /* Free blocks count */
1012 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1013 __le32 s_first_data_block; /* First Data Block */
1014 __le32 s_log_block_size; /* Block size */
1015 __le32 s_log_cluster_size; /* Allocation cluster size */
1016 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1017 __le32 s_clusters_per_group; /* # Clusters per group */
1018 __le32 s_inodes_per_group; /* # Inodes per group */
1019 __le32 s_mtime; /* Mount time */
1020 /*30*/ __le32 s_wtime; /* Write time */
1021 __le16 s_mnt_count; /* Mount count */
1022 __le16 s_max_mnt_count; /* Maximal mount count */
1023 __le16 s_magic; /* Magic signature */
1024 __le16 s_state; /* File system state */
1025 __le16 s_errors; /* Behaviour when detecting errors */
1026 __le16 s_minor_rev_level; /* minor revision level */
1027 /*40*/ __le32 s_lastcheck; /* time of last check */
1028 __le32 s_checkinterval; /* max. time between checks */
1029 __le32 s_creator_os; /* OS */
1030 __le32 s_rev_level; /* Revision level */
1031 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1032 __le16 s_def_resgid; /* Default gid for reserved blocks */
1033 /*
1034 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1035 *
1036 * Note: the difference between the compatible feature set and
1037 * the incompatible feature set is that if there is a bit set
1038 * in the incompatible feature set that the kernel doesn't
1039 * know about, it should refuse to mount the filesystem.
1040 *
1041 * e2fsck's requirements are more strict; if it doesn't know
1042 * about a feature in either the compatible or incompatible
1043 * feature set, it must abort and not try to meddle with
1044 * things it doesn't understand...
1045 */
1046 __le32 s_first_ino; /* First non-reserved inode */
1047 __le16 s_inode_size; /* size of inode structure */
1048 __le16 s_block_group_nr; /* block group # of this superblock */
1049 __le32 s_feature_compat; /* compatible feature set */
1050 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1051 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1052 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1053 /*78*/ char s_volume_name[16]; /* volume name */
1054 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1055 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1056 /*
1057 * Performance hints. Directory preallocation should only
1058 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1059 */
1060 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1061 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1062 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1063 /*
1064 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1065 */
1066 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1067 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1068 __le32 s_journal_dev; /* device number of journal file */
1069 __le32 s_last_orphan; /* start of list of inodes to delete */
1070 __le32 s_hash_seed[4]; /* HTREE hash seed */
1071 __u8 s_def_hash_version; /* Default hash version to use */
1072 __u8 s_jnl_backup_type;
1073 __le16 s_desc_size; /* size of group descriptor */
1074 /*100*/ __le32 s_default_mount_opts;
1075 __le32 s_first_meta_bg; /* First metablock block group */
1076 __le32 s_mkfs_time; /* When the filesystem was created */
1077 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1078 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1079 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1080 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1081 __le32 s_free_blocks_count_hi; /* Free blocks count */
1082 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1083 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1084 __le32 s_flags; /* Miscellaneous flags */
1085 __le16 s_raid_stride; /* RAID stride */
1086 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1087 __le64 s_mmp_block; /* Block for multi-mount protection */
1088 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1089 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1090 __u8 s_reserved_char_pad;
1091 __le16 s_reserved_pad;
1092 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1093 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1094 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1095 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1096 snapshot's future use */
1097 __le32 s_snapshot_list; /* inode number of the head of the
1098 on-disk snapshot list */
1099 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1100 __le32 s_error_count; /* number of fs errors */
1101 __le32 s_first_error_time; /* first time an error happened */
1102 __le32 s_first_error_ino; /* inode involved in first error */
1103 __le64 s_first_error_block; /* block involved of first error */
1104 __u8 s_first_error_func[32]; /* function where the error happened */
1105 __le32 s_first_error_line; /* line number where error happened */
1106 __le32 s_last_error_time; /* most recent time of an error */
1107 __le32 s_last_error_ino; /* inode involved in last error */
1108 __le32 s_last_error_line; /* line number where error happened */
1109 __le64 s_last_error_block; /* block involved of last error */
1110 __u8 s_last_error_func[32]; /* function where the error happened */
1111 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1112 __u8 s_mount_opts[64];
1113 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1114 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1115 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1116 __le32 s_reserved[109]; /* Padding to the end of the block */
1117 };
1118
1119 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1120
1121 #ifdef __KERNEL__
1122
1123 /*
1124 * run-time mount flags
1125 */
1126 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1127 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1128
1129 /*
1130 * fourth extended-fs super-block data in memory
1131 */
1132 struct ext4_sb_info {
1133 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1134 unsigned long s_inodes_per_block;/* Number of inodes per block */
1135 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1136 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1137 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1138 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1139 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1140 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1141 ext4_group_t s_groups_count; /* Number of groups in the fs */
1142 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1143 unsigned long s_overhead; /* # of fs overhead clusters */
1144 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1145 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1146 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1147 struct buffer_head * s_sbh; /* Buffer containing the super block */
1148 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1149 struct buffer_head **s_group_desc;
1150 unsigned int s_mount_opt;
1151 unsigned int s_mount_opt2;
1152 unsigned int s_mount_flags;
1153 unsigned int s_def_mount_opt;
1154 ext4_fsblk_t s_sb_block;
1155 uid_t s_resuid;
1156 gid_t s_resgid;
1157 unsigned short s_mount_state;
1158 unsigned short s_pad;
1159 int s_addr_per_block_bits;
1160 int s_desc_per_block_bits;
1161 int s_inode_size;
1162 int s_first_ino;
1163 unsigned int s_inode_readahead_blks;
1164 unsigned int s_inode_goal;
1165 spinlock_t s_next_gen_lock;
1166 u32 s_next_generation;
1167 u32 s_hash_seed[4];
1168 int s_def_hash_version;
1169 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1170 struct percpu_counter s_freeclusters_counter;
1171 struct percpu_counter s_freeinodes_counter;
1172 struct percpu_counter s_dirs_counter;
1173 struct percpu_counter s_dirtyclusters_counter;
1174 struct blockgroup_lock *s_blockgroup_lock;
1175 struct proc_dir_entry *s_proc;
1176 struct kobject s_kobj;
1177 struct completion s_kobj_unregister;
1178
1179 /* Journaling */
1180 struct journal_s *s_journal;
1181 struct list_head s_orphan;
1182 struct mutex s_orphan_lock;
1183 unsigned long s_resize_flags; /* Flags indicating if there
1184 is a resizer */
1185 unsigned long s_commit_interval;
1186 u32 s_max_batch_time;
1187 u32 s_min_batch_time;
1188 struct block_device *journal_bdev;
1189 #ifdef CONFIG_QUOTA
1190 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1191 int s_jquota_fmt; /* Format of quota to use */
1192 #endif
1193 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1194 struct rb_root system_blks;
1195
1196 #ifdef EXTENTS_STATS
1197 /* ext4 extents stats */
1198 unsigned long s_ext_min;
1199 unsigned long s_ext_max;
1200 unsigned long s_depth_max;
1201 spinlock_t s_ext_stats_lock;
1202 unsigned long s_ext_blocks;
1203 unsigned long s_ext_extents;
1204 #endif
1205
1206 /* for buddy allocator */
1207 struct ext4_group_info ***s_group_info;
1208 struct inode *s_buddy_cache;
1209 spinlock_t s_md_lock;
1210 unsigned short *s_mb_offsets;
1211 unsigned int *s_mb_maxs;
1212
1213 /* tunables */
1214 unsigned long s_stripe;
1215 unsigned int s_mb_stream_request;
1216 unsigned int s_mb_max_to_scan;
1217 unsigned int s_mb_min_to_scan;
1218 unsigned int s_mb_stats;
1219 unsigned int s_mb_order2_reqs;
1220 unsigned int s_mb_group_prealloc;
1221 unsigned int s_max_writeback_mb_bump;
1222 /* where last allocation was done - for stream allocation */
1223 unsigned long s_mb_last_group;
1224 unsigned long s_mb_last_start;
1225
1226 /* stats for buddy allocator */
1227 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1228 atomic_t s_bal_success; /* we found long enough chunks */
1229 atomic_t s_bal_allocated; /* in blocks */
1230 atomic_t s_bal_ex_scanned; /* total extents scanned */
1231 atomic_t s_bal_goals; /* goal hits */
1232 atomic_t s_bal_breaks; /* too long searches */
1233 atomic_t s_bal_2orders; /* 2^order hits */
1234 spinlock_t s_bal_lock;
1235 unsigned long s_mb_buddies_generated;
1236 unsigned long long s_mb_generation_time;
1237 atomic_t s_mb_lost_chunks;
1238 atomic_t s_mb_preallocated;
1239 atomic_t s_mb_discarded;
1240 atomic_t s_lock_busy;
1241
1242 /* locality groups */
1243 struct ext4_locality_group __percpu *s_locality_groups;
1244
1245 /* for write statistics */
1246 unsigned long s_sectors_written_start;
1247 u64 s_kbytes_written;
1248
1249 unsigned int s_log_groups_per_flex;
1250 struct flex_groups *s_flex_groups;
1251
1252 /* workqueue for dio unwritten */
1253 struct workqueue_struct *dio_unwritten_wq;
1254
1255 /* timer for periodic error stats printing */
1256 struct timer_list s_err_report;
1257
1258 /* Lazy inode table initialization info */
1259 struct ext4_li_request *s_li_request;
1260 /* Wait multiplier for lazy initialization thread */
1261 unsigned int s_li_wait_mult;
1262
1263 /* Kernel thread for multiple mount protection */
1264 struct task_struct *s_mmp_tsk;
1265
1266 /* record the last minlen when FITRIM is called. */
1267 atomic_t s_last_trim_minblks;
1268 };
1269
EXT4_SB(struct super_block * sb)1270 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1271 {
1272 return sb->s_fs_info;
1273 }
EXT4_I(struct inode * inode)1274 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1275 {
1276 return container_of(inode, struct ext4_inode_info, vfs_inode);
1277 }
1278
ext4_current_time(struct inode * inode)1279 static inline struct timespec ext4_current_time(struct inode *inode)
1280 {
1281 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1282 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1283 }
1284
ext4_valid_inum(struct super_block * sb,unsigned long ino)1285 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1286 {
1287 return ino == EXT4_ROOT_INO ||
1288 ino == EXT4_JOURNAL_INO ||
1289 ino == EXT4_RESIZE_INO ||
1290 (ino >= EXT4_FIRST_INO(sb) &&
1291 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1292 }
1293
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)1294 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1295 struct ext4_io_end *io_end)
1296 {
1297 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1298 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1299 atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
1300 }
1301 }
1302
1303 /*
1304 * Inode dynamic state flags
1305 */
1306 enum {
1307 EXT4_STATE_JDATA, /* journaled data exists */
1308 EXT4_STATE_NEW, /* inode is newly created */
1309 EXT4_STATE_XATTR, /* has in-inode xattrs */
1310 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1311 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1312 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1313 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1314 EXT4_STATE_NEWENTRY, /* File just added to dir */
1315 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1316 };
1317
1318 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1319 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1320 { \
1321 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1322 } \
1323 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1324 { \
1325 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1326 } \
1327 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1328 { \
1329 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1330 }
1331
1332 EXT4_INODE_BIT_FNS(flag, flags, 0)
1333 #if (BITS_PER_LONG < 64)
1334 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1335
ext4_clear_state_flags(struct ext4_inode_info * ei)1336 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1337 {
1338 (ei)->i_state_flags = 0;
1339 }
1340 #else
1341 EXT4_INODE_BIT_FNS(state, flags, 32)
1342
1343 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1344 {
1345 /* We depend on the fact that callers will set i_flags */
1346 }
1347 #endif
1348 #else
1349 /* Assume that user mode programs are passing in an ext4fs superblock, not
1350 * a kernel struct super_block. This will allow us to call the feature-test
1351 * macros from user land. */
1352 #define EXT4_SB(sb) (sb)
1353 #endif
1354
1355 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1356
1357 /*
1358 * Codes for operating systems
1359 */
1360 #define EXT4_OS_LINUX 0
1361 #define EXT4_OS_HURD 1
1362 #define EXT4_OS_MASIX 2
1363 #define EXT4_OS_FREEBSD 3
1364 #define EXT4_OS_LITES 4
1365
1366 /*
1367 * Revision levels
1368 */
1369 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1370 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1371
1372 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1373 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1374
1375 #define EXT4_GOOD_OLD_INODE_SIZE 128
1376
1377 /*
1378 * Feature set definitions
1379 */
1380
1381 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1382 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1383 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1384 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1385 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1386 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1387 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1388 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1389 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1390 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1391 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1392 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1393 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1394 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1395 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1396 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1397 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1398 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1399
1400 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1401 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1402 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1403 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1404 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1405 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1406
1407 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1408 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1409 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1410 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1411 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1412 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1413 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1414 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1415 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1416 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1417
1418 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1419 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1420 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1421 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1422 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1423 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1424 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1425 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1426 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1427 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1428 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1429 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1430 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1431 #define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */
1432
1433 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1434 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1435 EXT4_FEATURE_INCOMPAT_META_BG)
1436 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1437 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1438 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1439
1440 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1441 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1442 EXT4_FEATURE_INCOMPAT_RECOVER| \
1443 EXT4_FEATURE_INCOMPAT_META_BG)
1444 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1445 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1446 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1447
1448 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1449 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1450 EXT4_FEATURE_INCOMPAT_RECOVER| \
1451 EXT4_FEATURE_INCOMPAT_META_BG| \
1452 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1453 EXT4_FEATURE_INCOMPAT_64BIT| \
1454 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1455 EXT4_FEATURE_INCOMPAT_MMP)
1456 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1457 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1458 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1459 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1460 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1461 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1462 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1463 EXT4_FEATURE_RO_COMPAT_BIGALLOC)
1464
1465 /*
1466 * Default values for user and/or group using reserved blocks
1467 */
1468 #define EXT4_DEF_RESUID 0
1469 #define EXT4_DEF_RESGID 0
1470
1471 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1472
1473 /*
1474 * Default mount options
1475 */
1476 #define EXT4_DEFM_DEBUG 0x0001
1477 #define EXT4_DEFM_BSDGROUPS 0x0002
1478 #define EXT4_DEFM_XATTR_USER 0x0004
1479 #define EXT4_DEFM_ACL 0x0008
1480 #define EXT4_DEFM_UID16 0x0010
1481 #define EXT4_DEFM_JMODE 0x0060
1482 #define EXT4_DEFM_JMODE_DATA 0x0020
1483 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1484 #define EXT4_DEFM_JMODE_WBACK 0x0060
1485 #define EXT4_DEFM_NOBARRIER 0x0100
1486 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1487 #define EXT4_DEFM_DISCARD 0x0400
1488 #define EXT4_DEFM_NODELALLOC 0x0800
1489
1490 /*
1491 * Default journal batch times
1492 */
1493 #define EXT4_DEF_MIN_BATCH_TIME 0
1494 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1495
1496 /*
1497 * Minimum number of groups in a flexgroup before we separate out
1498 * directories into the first block group of a flexgroup
1499 */
1500 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1501
1502 /*
1503 * Structure of a directory entry
1504 */
1505 #define EXT4_NAME_LEN 255
1506
1507 struct ext4_dir_entry {
1508 __le32 inode; /* Inode number */
1509 __le16 rec_len; /* Directory entry length */
1510 __le16 name_len; /* Name length */
1511 char name[EXT4_NAME_LEN]; /* File name */
1512 };
1513
1514 /*
1515 * The new version of the directory entry. Since EXT4 structures are
1516 * stored in intel byte order, and the name_len field could never be
1517 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1518 * file_type field.
1519 */
1520 struct ext4_dir_entry_2 {
1521 __le32 inode; /* Inode number */
1522 __le16 rec_len; /* Directory entry length */
1523 __u8 name_len; /* Name length */
1524 __u8 file_type;
1525 char name[EXT4_NAME_LEN]; /* File name */
1526 };
1527
1528 /*
1529 * Ext4 directory file types. Only the low 3 bits are used. The
1530 * other bits are reserved for now.
1531 */
1532 #define EXT4_FT_UNKNOWN 0
1533 #define EXT4_FT_REG_FILE 1
1534 #define EXT4_FT_DIR 2
1535 #define EXT4_FT_CHRDEV 3
1536 #define EXT4_FT_BLKDEV 4
1537 #define EXT4_FT_FIFO 5
1538 #define EXT4_FT_SOCK 6
1539 #define EXT4_FT_SYMLINK 7
1540
1541 #define EXT4_FT_MAX 8
1542
1543 /*
1544 * EXT4_DIR_PAD defines the directory entries boundaries
1545 *
1546 * NOTE: It must be a multiple of 4
1547 */
1548 #define EXT4_DIR_PAD 4
1549 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1550 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1551 ~EXT4_DIR_ROUND)
1552 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1553
1554 /*
1555 * If we ever get support for fs block sizes > page_size, we'll need
1556 * to remove the #if statements in the next two functions...
1557 */
1558 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1559 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1560 {
1561 unsigned len = le16_to_cpu(dlen);
1562
1563 #if (PAGE_CACHE_SIZE >= 65536)
1564 if (len == EXT4_MAX_REC_LEN || len == 0)
1565 return blocksize;
1566 return (len & 65532) | ((len & 3) << 16);
1567 #else
1568 return len;
1569 #endif
1570 }
1571
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1572 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1573 {
1574 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1575 BUG();
1576 #if (PAGE_CACHE_SIZE >= 65536)
1577 if (len < 65536)
1578 return cpu_to_le16(len);
1579 if (len == blocksize) {
1580 if (blocksize == 65536)
1581 return cpu_to_le16(EXT4_MAX_REC_LEN);
1582 else
1583 return cpu_to_le16(0);
1584 }
1585 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1586 #else
1587 return cpu_to_le16(len);
1588 #endif
1589 }
1590
1591 /*
1592 * Hash Tree Directory indexing
1593 * (c) Daniel Phillips, 2001
1594 */
1595
1596 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1597 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1598 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1599 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1600 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1601
1602 /* Legal values for the dx_root hash_version field: */
1603
1604 #define DX_HASH_LEGACY 0
1605 #define DX_HASH_HALF_MD4 1
1606 #define DX_HASH_TEA 2
1607 #define DX_HASH_LEGACY_UNSIGNED 3
1608 #define DX_HASH_HALF_MD4_UNSIGNED 4
1609 #define DX_HASH_TEA_UNSIGNED 5
1610
1611 #ifdef __KERNEL__
1612
1613 /* hash info structure used by the directory hash */
1614 struct dx_hash_info
1615 {
1616 u32 hash;
1617 u32 minor_hash;
1618 int hash_version;
1619 u32 *seed;
1620 };
1621
1622
1623 /* 32 and 64 bit signed EOF for dx directories */
1624 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1625 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1626
1627
1628 /*
1629 * Control parameters used by ext4_htree_next_block
1630 */
1631 #define HASH_NB_ALWAYS 1
1632
1633
1634 /*
1635 * Describe an inode's exact location on disk and in memory
1636 */
1637 struct ext4_iloc
1638 {
1639 struct buffer_head *bh;
1640 unsigned long offset;
1641 ext4_group_t block_group;
1642 };
1643
ext4_raw_inode(struct ext4_iloc * iloc)1644 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1645 {
1646 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1647 }
1648
1649 /*
1650 * This structure is stuffed into the struct file's private_data field
1651 * for directories. It is where we put information so that we can do
1652 * readdir operations in hash tree order.
1653 */
1654 struct dir_private_info {
1655 struct rb_root root;
1656 struct rb_node *curr_node;
1657 struct fname *extra_fname;
1658 loff_t last_pos;
1659 __u32 curr_hash;
1660 __u32 curr_minor_hash;
1661 __u32 next_hash;
1662 };
1663
1664 /* calculate the first block number of the group */
1665 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)1666 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1667 {
1668 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1669 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1670 }
1671
1672 /*
1673 * Special error return code only used by dx_probe() and its callers.
1674 */
1675 #define ERR_BAD_DX_DIR -75000
1676
1677 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1678 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1679
1680 /*
1681 * Timeout and state flag for lazy initialization inode thread.
1682 */
1683 #define EXT4_DEF_LI_WAIT_MULT 10
1684 #define EXT4_DEF_LI_MAX_START_DELAY 5
1685 #define EXT4_LAZYINIT_QUIT 0x0001
1686 #define EXT4_LAZYINIT_RUNNING 0x0002
1687
1688 /*
1689 * Lazy inode table initialization info
1690 */
1691 struct ext4_lazy_init {
1692 unsigned long li_state;
1693 struct list_head li_request_list;
1694 struct mutex li_list_mtx;
1695 };
1696
1697 struct ext4_li_request {
1698 struct super_block *lr_super;
1699 struct ext4_sb_info *lr_sbi;
1700 ext4_group_t lr_next_group;
1701 struct list_head lr_request;
1702 unsigned long lr_next_sched;
1703 unsigned long lr_timeout;
1704 };
1705
1706 struct ext4_features {
1707 struct kobject f_kobj;
1708 struct completion f_kobj_unregister;
1709 };
1710
1711 /*
1712 * This structure will be used for multiple mount protection. It will be
1713 * written into the block number saved in the s_mmp_block field in the
1714 * superblock. Programs that check MMP should assume that if
1715 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1716 * to use the filesystem, regardless of how old the timestamp is.
1717 */
1718 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1719 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1720 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1721 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1722
1723 struct mmp_struct {
1724 __le32 mmp_magic; /* Magic number for MMP */
1725 __le32 mmp_seq; /* Sequence no. updated periodically */
1726
1727 /*
1728 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1729 * purposes and do not affect the correctness of the algorithm
1730 */
1731 __le64 mmp_time; /* Time last updated */
1732 char mmp_nodename[64]; /* Node which last updated MMP block */
1733 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1734
1735 /*
1736 * mmp_check_interval is used to verify if the MMP block has been
1737 * updated on the block device. The value is updated based on the
1738 * maximum time to write the MMP block during an update cycle.
1739 */
1740 __le16 mmp_check_interval;
1741
1742 __le16 mmp_pad1;
1743 __le32 mmp_pad2[227];
1744 };
1745
1746 /* arguments passed to the mmp thread */
1747 struct mmpd_data {
1748 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1749 struct super_block *sb; /* super block of the fs */
1750 };
1751
1752 /*
1753 * Check interval multiplier
1754 * The MMP block is written every update interval and initially checked every
1755 * update interval x the multiplier (the value is then adapted based on the
1756 * write latency). The reason is that writes can be delayed under load and we
1757 * don't want readers to incorrectly assume that the filesystem is no longer
1758 * in use.
1759 */
1760 #define EXT4_MMP_CHECK_MULT 2UL
1761
1762 /*
1763 * Minimum interval for MMP checking in seconds.
1764 */
1765 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1766
1767 /*
1768 * Maximum interval for MMP checking in seconds.
1769 */
1770 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1771
1772 /*
1773 * Function prototypes
1774 */
1775
1776 /*
1777 * Ok, these declarations are also in <linux/kernel.h> but none of the
1778 * ext4 source programs needs to include it so they are duplicated here.
1779 */
1780 # define NORET_TYPE /**/
1781 # define ATTRIB_NORET __attribute__((noreturn))
1782 # define NORET_AND noreturn,
1783
1784 /* bitmap.c */
1785 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1786
1787 /* balloc.c */
1788 extern unsigned int ext4_block_group(struct super_block *sb,
1789 ext4_fsblk_t blocknr);
1790 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1791 ext4_fsblk_t blocknr);
1792 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1793 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1794 ext4_group_t group);
1795 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1796 ext4_fsblk_t goal,
1797 unsigned int flags,
1798 unsigned long *count,
1799 int *errp);
1800 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1801 s64 nclusters, unsigned int flags);
1802 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1803 extern void ext4_check_blocks_bitmap(struct super_block *);
1804 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1805 ext4_group_t block_group,
1806 struct buffer_head ** bh);
1807 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1808
1809 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1810 ext4_group_t block_group);
1811 extern int ext4_wait_block_bitmap(struct super_block *sb,
1812 ext4_group_t block_group,
1813 struct buffer_head *bh);
1814 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1815 ext4_group_t block_group);
1816 extern void ext4_init_block_bitmap(struct super_block *sb,
1817 struct buffer_head *bh,
1818 ext4_group_t group,
1819 struct ext4_group_desc *desc);
1820 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1821 ext4_group_t block_group,
1822 struct ext4_group_desc *gdp);
1823 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1824 ext4_group_t block_group,
1825 struct ext4_group_desc *gdp);
1826 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1827
1828 /* dir.c */
1829 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1830 struct file *,
1831 struct ext4_dir_entry_2 *,
1832 struct buffer_head *, unsigned int);
1833 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1834 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1835 (de), (bh), (offset)))
1836 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1837 __u32 minor_hash,
1838 struct ext4_dir_entry_2 *dirent);
1839 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1840
1841 /* fsync.c */
1842 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1843 extern int ext4_flush_completed_IO(struct inode *);
1844
1845 /* hash.c */
1846 extern int ext4fs_dirhash(const char *name, int len, struct
1847 dx_hash_info *hinfo);
1848
1849 /* ialloc.c */
1850 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
1851 const struct qstr *qstr, __u32 goal,
1852 uid_t *owner);
1853 extern void ext4_free_inode(handle_t *, struct inode *);
1854 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1855 extern unsigned long ext4_count_free_inodes(struct super_block *);
1856 extern unsigned long ext4_count_dirs(struct super_block *);
1857 extern void ext4_check_inodes_bitmap(struct super_block *);
1858 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1859 extern int ext4_init_inode_table(struct super_block *sb,
1860 ext4_group_t group, int barrier);
1861 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
1862
1863 /* mballoc.c */
1864 extern long ext4_mb_stats;
1865 extern long ext4_mb_max_to_scan;
1866 extern int ext4_mb_init(struct super_block *, int);
1867 extern int ext4_mb_release(struct super_block *);
1868 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1869 struct ext4_allocation_request *, int *);
1870 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1871 extern void ext4_discard_preallocations(struct inode *);
1872 extern int __init ext4_init_mballoc(void);
1873 extern void ext4_exit_mballoc(void);
1874 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1875 struct buffer_head *bh, ext4_fsblk_t block,
1876 unsigned long count, int flags);
1877 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1878 ext4_group_t i, struct ext4_group_desc *desc);
1879 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1880 ext4_fsblk_t block, unsigned long count);
1881 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
1882 unsigned long blkdev_flags);
1883
1884 /* inode.c */
1885 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1886 ext4_lblk_t, int, int *);
1887 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1888 ext4_lblk_t, int, int *);
1889 int ext4_get_block(struct inode *inode, sector_t iblock,
1890 struct buffer_head *bh_result, int create);
1891
1892 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1893 extern int ext4_write_inode(struct inode *, struct writeback_control *);
1894 extern int ext4_setattr(struct dentry *, struct iattr *);
1895 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1896 struct kstat *stat);
1897 extern void ext4_evict_inode(struct inode *);
1898 extern void ext4_clear_inode(struct inode *);
1899 extern int ext4_sync_inode(handle_t *, struct inode *);
1900 extern void ext4_dirty_inode(struct inode *, int);
1901 extern int ext4_change_inode_journal_flag(struct inode *, int);
1902 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1903 extern int ext4_can_truncate(struct inode *inode);
1904 extern void ext4_truncate(struct inode *);
1905 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1906 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1907 extern void ext4_set_inode_flags(struct inode *);
1908 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1909 extern int ext4_alloc_da_blocks(struct inode *inode);
1910 extern void ext4_set_aops(struct inode *inode);
1911 extern int ext4_writepage_trans_blocks(struct inode *);
1912 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1913 extern int ext4_discard_partial_page_buffers(handle_t *handle,
1914 struct address_space *mapping, loff_t from,
1915 loff_t length, int flags);
1916 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1917 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1918 extern void ext4_da_update_reserve_space(struct inode *inode,
1919 int used, int quota_claim);
1920
1921 /* indirect.c */
1922 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
1923 struct ext4_map_blocks *map, int flags);
1924 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
1925 const struct iovec *iov, loff_t offset,
1926 unsigned long nr_segs);
1927 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
1928 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
1929 extern void ext4_ind_truncate(struct inode *inode);
1930
1931 /* ioctl.c */
1932 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1933 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1934
1935 /* migrate.c */
1936 extern int ext4_ext_migrate(struct inode *);
1937
1938 /* namei.c */
1939 extern int ext4_orphan_add(handle_t *, struct inode *);
1940 extern int ext4_orphan_del(handle_t *, struct inode *);
1941 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1942 __u32 start_minor_hash, __u32 *next_hash);
1943
1944 /* resize.c */
1945 extern int ext4_group_add(struct super_block *sb,
1946 struct ext4_new_group_data *input);
1947 extern int ext4_group_extend(struct super_block *sb,
1948 struct ext4_super_block *es,
1949 ext4_fsblk_t n_blocks_count);
1950 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
1951
1952 /* super.c */
1953 extern int ext4_calculate_overhead(struct super_block *sb);
1954 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
1955 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
1956 extern void ext4_kvfree(void *ptr);
1957 extern __printf(4, 5)
1958 void __ext4_error(struct super_block *, const char *, unsigned int,
1959 const char *, ...);
1960 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1961 __LINE__, ## message)
1962 extern __printf(5, 6)
1963 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
1964 const char *, ...);
1965 extern __printf(5, 6)
1966 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
1967 const char *, ...);
1968 extern void __ext4_std_error(struct super_block *, const char *,
1969 unsigned int, int);
1970 extern __printf(4, 5)
1971 void __ext4_abort(struct super_block *, const char *, unsigned int,
1972 const char *, ...);
1973 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1974 __LINE__, ## message)
1975 extern __printf(4, 5)
1976 void __ext4_warning(struct super_block *, const char *, unsigned int,
1977 const char *, ...);
1978 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1979 __LINE__, ## message)
1980 extern __printf(3, 4)
1981 void ext4_msg(struct super_block *, const char *, const char *, ...);
1982 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
1983 const char *, unsigned int, const char *);
1984 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
1985 __LINE__, msg)
1986 extern __printf(7, 8)
1987 void __ext4_grp_locked_error(const char *, unsigned int,
1988 struct super_block *, ext4_group_t,
1989 unsigned long, ext4_fsblk_t,
1990 const char *, ...);
1991 #define ext4_grp_locked_error(sb, grp, message...) \
1992 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1993 extern void ext4_update_dynamic_rev(struct super_block *sb);
1994 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1995 __u32 compat);
1996 extern int ext4_update_rocompat_feature(handle_t *handle,
1997 struct super_block *sb, __u32 rocompat);
1998 extern int ext4_update_incompat_feature(handle_t *handle,
1999 struct super_block *sb, __u32 incompat);
2000 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2001 struct ext4_group_desc *bg);
2002 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2003 struct ext4_group_desc *bg);
2004 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2005 struct ext4_group_desc *bg);
2006 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2007 struct ext4_group_desc *bg);
2008 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2009 struct ext4_group_desc *bg);
2010 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2011 struct ext4_group_desc *bg);
2012 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2013 struct ext4_group_desc *bg);
2014 extern void ext4_block_bitmap_set(struct super_block *sb,
2015 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2016 extern void ext4_inode_bitmap_set(struct super_block *sb,
2017 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2018 extern void ext4_inode_table_set(struct super_block *sb,
2019 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2020 extern void ext4_free_group_clusters_set(struct super_block *sb,
2021 struct ext4_group_desc *bg,
2022 __u32 count);
2023 extern void ext4_free_inodes_set(struct super_block *sb,
2024 struct ext4_group_desc *bg, __u32 count);
2025 extern void ext4_used_dirs_set(struct super_block *sb,
2026 struct ext4_group_desc *bg, __u32 count);
2027 extern void ext4_itable_unused_set(struct super_block *sb,
2028 struct ext4_group_desc *bg, __u32 count);
2029 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
2030 struct ext4_group_desc *gdp);
2031 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
2032 struct ext4_group_desc *gdp);
2033
ext4_blocks_count(struct ext4_super_block * es)2034 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2035 {
2036 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2037 le32_to_cpu(es->s_blocks_count_lo);
2038 }
2039
ext4_r_blocks_count(struct ext4_super_block * es)2040 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2041 {
2042 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2043 le32_to_cpu(es->s_r_blocks_count_lo);
2044 }
2045
ext4_free_blocks_count(struct ext4_super_block * es)2046 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2047 {
2048 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2049 le32_to_cpu(es->s_free_blocks_count_lo);
2050 }
2051
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2052 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2053 ext4_fsblk_t blk)
2054 {
2055 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2056 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2057 }
2058
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2059 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2060 ext4_fsblk_t blk)
2061 {
2062 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2063 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2064 }
2065
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2066 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2067 ext4_fsblk_t blk)
2068 {
2069 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2070 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2071 }
2072
ext4_isize(struct ext4_inode * raw_inode)2073 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2074 {
2075 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2076 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2077 le32_to_cpu(raw_inode->i_size_lo);
2078 else
2079 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2080 }
2081
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2082 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2083 {
2084 raw_inode->i_size_lo = cpu_to_le32(i_size);
2085 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2086 }
2087
2088 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2089 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2090 ext4_group_t group)
2091 {
2092 struct ext4_group_info ***grp_info;
2093 long indexv, indexh;
2094 grp_info = EXT4_SB(sb)->s_group_info;
2095 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2096 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2097 return grp_info[indexv][indexh];
2098 }
2099
2100 /*
2101 * Reading s_groups_count requires using smp_rmb() afterwards. See
2102 * the locking protocol documented in the comments of ext4_group_add()
2103 * in resize.c
2104 */
ext4_get_groups_count(struct super_block * sb)2105 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2106 {
2107 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2108
2109 smp_rmb();
2110 return ngroups;
2111 }
2112
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2113 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2114 ext4_group_t block_group)
2115 {
2116 return block_group >> sbi->s_log_groups_per_flex;
2117 }
2118
ext4_flex_bg_size(struct ext4_sb_info * sbi)2119 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2120 {
2121 return 1 << sbi->s_log_groups_per_flex;
2122 }
2123
2124 #define ext4_std_error(sb, errno) \
2125 do { \
2126 if ((errno)) \
2127 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2128 } while (0)
2129
2130 #ifdef CONFIG_SMP
2131 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2132 * counters. So we need to make sure we have free clusters more
2133 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2134 */
2135 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2136 #else
2137 #define EXT4_FREECLUSTERS_WATERMARK 0
2138 #endif
2139
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2140 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2141 {
2142 /*
2143 * XXX: replace with spinlock if seen contended -bzzz
2144 */
2145 down_write(&EXT4_I(inode)->i_data_sem);
2146 if (newsize > EXT4_I(inode)->i_disksize)
2147 EXT4_I(inode)->i_disksize = newsize;
2148 up_write(&EXT4_I(inode)->i_data_sem);
2149 return ;
2150 }
2151
2152 struct ext4_group_info {
2153 unsigned long bb_state;
2154 struct rb_root bb_free_root;
2155 ext4_grpblk_t bb_first_free; /* first free block */
2156 ext4_grpblk_t bb_free; /* total free blocks */
2157 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2158 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2159 struct list_head bb_prealloc_list;
2160 #ifdef DOUBLE_CHECK
2161 void *bb_bitmap;
2162 #endif
2163 struct rw_semaphore alloc_sem;
2164 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2165 * regions, index is order.
2166 * bb_counters[3] = 5 means
2167 * 5 free 8-block regions. */
2168 };
2169
2170 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2171 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2172
2173 #define EXT4_MB_GRP_NEED_INIT(grp) \
2174 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2175
2176 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2177 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2178 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2179 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2180 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2181 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2182
2183 #define EXT4_MAX_CONTENTION 8
2184 #define EXT4_CONTENTION_THRESHOLD 2
2185
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2186 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2187 ext4_group_t group)
2188 {
2189 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2190 }
2191
2192 /*
2193 * Returns true if the filesystem is busy enough that attempts to
2194 * access the block group locks has run into contention.
2195 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2196 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2197 {
2198 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2199 }
2200
ext4_lock_group(struct super_block * sb,ext4_group_t group)2201 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2202 {
2203 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2204 if (spin_trylock(lock))
2205 /*
2206 * We're able to grab the lock right away, so drop the
2207 * lock contention counter.
2208 */
2209 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2210 else {
2211 /*
2212 * The lock is busy, so bump the contention counter,
2213 * and then wait on the spin lock.
2214 */
2215 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2216 EXT4_MAX_CONTENTION);
2217 spin_lock(lock);
2218 }
2219 }
2220
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2221 static inline void ext4_unlock_group(struct super_block *sb,
2222 ext4_group_t group)
2223 {
2224 spin_unlock(ext4_group_lock_ptr(sb, group));
2225 }
2226
ext4_mark_super_dirty(struct super_block * sb)2227 static inline void ext4_mark_super_dirty(struct super_block *sb)
2228 {
2229 if (EXT4_SB(sb)->s_journal == NULL)
2230 sb->s_dirt =1;
2231 }
2232
2233 /*
2234 * Block validity checking
2235 */
2236 #define ext4_check_indirect_blockref(inode, bh) \
2237 ext4_check_blockref(__func__, __LINE__, inode, \
2238 (__le32 *)(bh)->b_data, \
2239 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2240
2241 #define ext4_ind_check_inode(inode) \
2242 ext4_check_blockref(__func__, __LINE__, inode, \
2243 EXT4_I(inode)->i_data, \
2244 EXT4_NDIR_BLOCKS)
2245
2246 /*
2247 * Inodes and files operations
2248 */
2249
2250 /* dir.c */
2251 extern const struct file_operations ext4_dir_operations;
2252
2253 /* file.c */
2254 extern const struct inode_operations ext4_file_inode_operations;
2255 extern const struct file_operations ext4_file_operations;
2256 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2257
2258 /* namei.c */
2259 extern const struct inode_operations ext4_dir_inode_operations;
2260 extern const struct inode_operations ext4_special_inode_operations;
2261 extern struct dentry *ext4_get_parent(struct dentry *child);
2262
2263 /* symlink.c */
2264 extern const struct inode_operations ext4_symlink_inode_operations;
2265 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2266
2267 /* block_validity */
2268 extern void ext4_release_system_zone(struct super_block *sb);
2269 extern int ext4_setup_system_zone(struct super_block *sb);
2270 extern int __init ext4_init_system_zone(void);
2271 extern void ext4_exit_system_zone(void);
2272 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2273 ext4_fsblk_t start_blk,
2274 unsigned int count);
2275 extern int ext4_check_blockref(const char *, unsigned int,
2276 struct inode *, __le32 *, unsigned int);
2277
2278 /* extents.c */
2279 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2280 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2281 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2282 int chunk);
2283 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2284 struct ext4_map_blocks *map, int flags);
2285 extern void ext4_ext_truncate(struct inode *);
2286 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2287 loff_t length);
2288 extern void ext4_ext_init(struct super_block *);
2289 extern void ext4_ext_release(struct super_block *);
2290 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2291 loff_t len);
2292 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2293 ssize_t len);
2294 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2295 struct ext4_map_blocks *map, int flags);
2296 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2297 __u64 start, __u64 len);
2298 /* move_extent.c */
2299 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2300 __u64 start_orig, __u64 start_donor,
2301 __u64 len, __u64 *moved_len);
2302
2303 /* page-io.c */
2304 extern int __init ext4_init_pageio(void);
2305 extern void ext4_exit_pageio(void);
2306 extern void ext4_ioend_wait(struct inode *);
2307 extern void ext4_free_io_end(ext4_io_end_t *io);
2308 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2309 extern int ext4_end_io_nolock(ext4_io_end_t *io);
2310 extern void ext4_io_submit(struct ext4_io_submit *io);
2311 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2312 struct page *page,
2313 int len,
2314 struct writeback_control *wbc);
2315
2316 /* mmp.c */
2317 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2318
2319 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2320 enum ext4_state_bits {
2321 BH_Uninit /* blocks are allocated but uninitialized on disk */
2322 = BH_JBDPrivateStart,
2323 BH_AllocFromCluster, /* allocated blocks were part of already
2324 * allocated cluster. Note that this flag will
2325 * never, ever appear in a buffer_head's state
2326 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2327 * this is used. */
2328 BH_Da_Mapped, /* Delayed allocated block that now has a mapping. This
2329 * flag is set when ext4_map_blocks is called on a
2330 * delayed allocated block to get its real mapping. */
2331 };
2332
BUFFER_FNS(Uninit,uninit)2333 BUFFER_FNS(Uninit, uninit)
2334 TAS_BUFFER_FNS(Uninit, uninit)
2335 BUFFER_FNS(Da_Mapped, da_mapped)
2336
2337 /*
2338 * Add new method to test wether block and inode bitmaps are properly
2339 * initialized. With uninit_bg reading the block from disk is not enough
2340 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2341 */
2342 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2343
2344 static inline int bitmap_uptodate(struct buffer_head *bh)
2345 {
2346 return (buffer_uptodate(bh) &&
2347 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2348 }
set_bitmap_uptodate(struct buffer_head * bh)2349 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2350 {
2351 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2352 }
2353
2354 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2355
2356 /* For ioend & aio unwritten conversion wait queues */
2357 #define EXT4_WQ_HASH_SZ 37
2358 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2359 EXT4_WQ_HASH_SZ])
2360 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2361 EXT4_WQ_HASH_SZ])
2362 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2363 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2364
2365 #define EXT4_RESIZING 0
2366 extern int ext4_resize_begin(struct super_block *sb);
2367 extern void ext4_resize_end(struct super_block *sb);
2368
2369 #endif /* __KERNEL__ */
2370
2371 #include "ext4_extents.h"
2372
2373 #endif /* _EXT4_H */
2374