?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Add linux-next specific files for 20110716
[linux-2.6/next.git] / fs / ext4 / ext4.h
blobbd7ba373eed9ccadb0a1859f93715fef2762ced0
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(f, a...) do {} while (0)
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 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
148 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
149 EXT4_MAP_UNINIT)
150
151 struct ext4_map_blocks {
152 ext4_fsblk_t m_pblk;
153 ext4_lblk_t m_lblk;
154 unsigned int m_len;
155 unsigned int m_flags;
156 };
157
158 /*
159 * For delayed allocation tracking
160 */
161 struct mpage_da_data {
162 struct inode *inode;
163 sector_t b_blocknr; /* start block number of extent */
164 size_t b_size; /* size of extent */
165 unsigned long b_state; /* state of the extent */
166 unsigned long first_page, next_page; /* extent of pages */
167 struct writeback_control *wbc;
168 int io_done;
169 int pages_written;
170 int retval;
171 };
172
173 /*
174 * Flags for ext4_io_end->flags
175 */
176 #define EXT4_IO_END_UNWRITTEN 0x0001
177 #define EXT4_IO_END_ERROR 0x0002
178
179 struct ext4_io_page {
180 struct page *p_page;
181 atomic_t p_count;
182 };
183
184 #define MAX_IO_PAGES 128
185
186 typedef struct ext4_io_end {
187 struct list_head list; /* per-file finished IO list */
188 struct inode *inode; /* file being written to */
189 unsigned int flag; /* unwritten or not */
190 struct page *page; /* page struct for buffer write */
191 loff_t offset; /* offset in the file */
192 ssize_t size; /* size of the extent */
193 struct work_struct work; /* data work queue */
194 struct kiocb *iocb; /* iocb struct for AIO */
195 int result; /* error value for AIO */
196 int num_io_pages;
197 struct ext4_io_page *pages[MAX_IO_PAGES];
198 } ext4_io_end_t;
199
200 struct ext4_io_submit {
201 int io_op;
202 struct bio *io_bio;
203 ext4_io_end_t *io_end;
204 struct ext4_io_page *io_page;
205 sector_t io_next_block;
206 };
207
208 /*
209 * Special inodes numbers
210 */
211 #define EXT4_BAD_INO 1 /* Bad blocks inode */
212 #define EXT4_ROOT_INO 2 /* Root inode */
213 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
214 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
215 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
216 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
217 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
218 #define EXT4_JOURNAL_INO 8 /* Journal inode */
219
220 /* First non-reserved inode for old ext4 filesystems */
221 #define EXT4_GOOD_OLD_FIRST_INO 11
222
223 /*
224 * Maximal count of links to a file
225 */
226 #define EXT4_LINK_MAX 65000
227
228 /*
229 * Macro-instructions used to manage several block sizes
230 */
231 #define EXT4_MIN_BLOCK_SIZE 1024
232 #define EXT4_MAX_BLOCK_SIZE 65536
233 #define EXT4_MIN_BLOCK_LOG_SIZE 10
234 #define EXT4_MAX_BLOCK_LOG_SIZE 16
235 #ifdef __KERNEL__
236 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
237 #else
238 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
239 #endif
240 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
241 #ifdef __KERNEL__
242 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
243 #else
244 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
245 #endif
246 #ifdef __KERNEL__
247 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
248 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
249 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
250 #else
251 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
252 EXT4_GOOD_OLD_INODE_SIZE : \
253 (s)->s_inode_size)
254 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
255 EXT4_GOOD_OLD_FIRST_INO : \
256 (s)->s_first_ino)
257 #endif
258 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
259
260 /*
261 * Structure of a blocks group descriptor
262 */
263 struct ext4_group_desc
264 {
265 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
266 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
267 __le32 bg_inode_table_lo; /* Inodes table block */
268 __le16 bg_free_blocks_count_lo;/* Free blocks count */
269 __le16 bg_free_inodes_count_lo;/* Free inodes count */
270 __le16 bg_used_dirs_count_lo; /* Directories count */
271 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
272 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
273 __le16 bg_itable_unused_lo; /* Unused inodes count */
274 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
275 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
276 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
277 __le32 bg_inode_table_hi; /* Inodes table block MSB */
278 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
279 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
280 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
281 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
282 __u32 bg_reserved2[3];
283 };
284
285 /*
286 * Structure of a flex block group info
287 */
288
289 struct flex_groups {
290 atomic_t free_inodes;
291 atomic_t free_blocks;
292 atomic_t used_dirs;
293 };
294
295 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
296 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
297 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
298
299 /*
300 * Macro-instructions used to manage group descriptors
301 */
302 #define EXT4_MIN_DESC_SIZE 32
303 #define EXT4_MIN_DESC_SIZE_64BIT 64
304 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
305 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
306 #ifdef __KERNEL__
307 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
308 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
309 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
310 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
311 #else
312 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
313 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
314 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
315 #endif
316
317 /*
318 * Constants relative to the data blocks
319 */
320 #define EXT4_NDIR_BLOCKS 12
321 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
322 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
323 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
324 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
325
326 /*
327 * Inode flags
328 */
329 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
330 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
331 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
332 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
333 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
334 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
335 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
336 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
337 /* Reserved for compression usage... */
338 #define EXT4_DIRTY_FL 0x00000100
339 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
340 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
341 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
342 /* End compression flags --- maybe not all used */
343 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
344 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
345 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
346 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
347 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
348 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
349 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
350 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
351 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
352 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
353 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
354
355 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
356 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
357
358 /* Flags that should be inherited by new inodes from their parent. */
359 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
360 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
361 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
362 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
363 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
364
365 /* Flags that are appropriate for regular files (all but dir-specific ones). */
366 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
367
368 /* Flags that are appropriate for non-directories/regular files. */
369 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
370
371 /* Mask out flags that are inappropriate for the given type of inode. */
372 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
373 {
374 if (S_ISDIR(mode))
375 return flags;
376 else if (S_ISREG(mode))
377 return flags & EXT4_REG_FLMASK;
378 else
379 return flags & EXT4_OTHER_FLMASK;
380 }
381
382 /*
383 * Inode flags used for atomic set/get
384 */
385 enum {
386 EXT4_INODE_SECRM = 0, /* Secure deletion */
387 EXT4_INODE_UNRM = 1, /* Undelete */
388 EXT4_INODE_COMPR = 2, /* Compress file */
389 EXT4_INODE_SYNC = 3, /* Synchronous updates */
390 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
391 EXT4_INODE_APPEND = 5, /* writes to file may only append */
392 EXT4_INODE_NODUMP = 6, /* do not dump file */
393 EXT4_INODE_NOATIME = 7, /* do not update atime */
394 /* Reserved for compression usage... */
395 EXT4_INODE_DIRTY = 8,
396 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
397 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
398 EXT4_INODE_ECOMPR = 11, /* Compression error */
399 /* End compression flags --- maybe not all used */
400 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
401 EXT4_INODE_IMAGIC = 13, /* AFS directory */
402 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
403 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
404 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
405 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
406 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
407 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
408 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
409 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
410 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
411 };
412
413 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
414 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
415 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
416 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
417
418 /*
419 * Since it's pretty easy to mix up bit numbers and hex values, and we
420 * can't do a compile-time test for ENUM values, we use a run-time
421 * test to make sure that EXT4_XXX_FL is consistent with respect to
422 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
423 * out so it won't cost any extra space in the compiled kernel image.
424 * But it's important that these values are the same, since we are
425 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
426 * must be consistent with the values of FS_XXX_FL defined in
427 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
428 * ext4 filesystems, and of course the values defined in e2fsprogs.
429 *
430 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
431 */
432 static inline void ext4_check_flag_values(void)
433 {
434 CHECK_FLAG_VALUE(SECRM);
435 CHECK_FLAG_VALUE(UNRM);
436 CHECK_FLAG_VALUE(COMPR);
437 CHECK_FLAG_VALUE(SYNC);
438 CHECK_FLAG_VALUE(IMMUTABLE);
439 CHECK_FLAG_VALUE(APPEND);
440 CHECK_FLAG_VALUE(NODUMP);
441 CHECK_FLAG_VALUE(NOATIME);
442 CHECK_FLAG_VALUE(DIRTY);
443 CHECK_FLAG_VALUE(COMPRBLK);
444 CHECK_FLAG_VALUE(NOCOMPR);
445 CHECK_FLAG_VALUE(ECOMPR);
446 CHECK_FLAG_VALUE(INDEX);
447 CHECK_FLAG_VALUE(IMAGIC);
448 CHECK_FLAG_VALUE(JOURNAL_DATA);
449 CHECK_FLAG_VALUE(NOTAIL);
450 CHECK_FLAG_VALUE(DIRSYNC);
451 CHECK_FLAG_VALUE(TOPDIR);
452 CHECK_FLAG_VALUE(HUGE_FILE);
453 CHECK_FLAG_VALUE(EXTENTS);
454 CHECK_FLAG_VALUE(EA_INODE);
455 CHECK_FLAG_VALUE(EOFBLOCKS);
456 CHECK_FLAG_VALUE(RESERVED);
457 }
458
459 /* Used to pass group descriptor data when online resize is done */
460 struct ext4_new_group_input {
461 __u32 group; /* Group number for this data */
462 __u64 block_bitmap; /* Absolute block number of block bitmap */
463 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
464 __u64 inode_table; /* Absolute block number of inode table start */
465 __u32 blocks_count; /* Total number of blocks in this group */
466 __u16 reserved_blocks; /* Number of reserved blocks in this group */
467 __u16 unused;
468 };
469
470 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
471 struct compat_ext4_new_group_input {
472 u32 group;
473 compat_u64 block_bitmap;
474 compat_u64 inode_bitmap;
475 compat_u64 inode_table;
476 u32 blocks_count;
477 u16 reserved_blocks;
478 u16 unused;
479 };
480 #endif
481
482 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
483 struct ext4_new_group_data {
484 __u32 group;
485 __u64 block_bitmap;
486 __u64 inode_bitmap;
487 __u64 inode_table;
488 __u32 blocks_count;
489 __u16 reserved_blocks;
490 __u16 unused;
491 __u32 free_blocks_count;
492 };
493
494 /*
495 * Flags used by ext4_map_blocks()
496 */
497 /* Allocate any needed blocks and/or convert an unitialized
498 extent to be an initialized ext4 */
499 #define EXT4_GET_BLOCKS_CREATE 0x0001
500 /* Request the creation of an unitialized extent */
501 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
502 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
503 EXT4_GET_BLOCKS_CREATE)
504 /* Caller is from the delayed allocation writeout path,
505 so set the magic i_delalloc_reserve_flag after taking the
506 inode allocation semaphore for */
507 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
508 /* caller is from the direct IO path, request to creation of an
509 unitialized extents if not allocated, split the uninitialized
510 extent if blocks has been preallocated already*/
511 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
512 #define EXT4_GET_BLOCKS_CONVERT 0x0010
513 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
514 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
515 /* Convert extent to initialized after IO complete */
516 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
517 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
518 /* Punch out blocks of an extent */
519 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
520 /* Don't normalize allocation size (used for fallocate) */
521 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
522
523 /*
524 * Flags used by ext4_free_blocks
525 */
526 #define EXT4_FREE_BLOCKS_METADATA 0x0001
527 #define EXT4_FREE_BLOCKS_FORGET 0x0002
528 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
529 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
530
531 /*
532 * ioctl commands
533 */
534 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
535 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
536 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
537 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
538 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
539 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
540 #ifdef CONFIG_JBD2_DEBUG
541 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
542 #endif
543 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
544 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
545 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
546 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
547 #define EXT4_IOC_MIGRATE _IO('f', 9)
548 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
549 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
550 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
551 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
552
553 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
554 /*
555 * ioctl commands in 32 bit emulation
556 */
557 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
558 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
559 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
560 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
561 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
562 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
563 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
564 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
565 #ifdef CONFIG_JBD2_DEBUG
566 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
567 #endif
568 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
569 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
570 #endif
571
572 /* Max physical block we can address w/o extents */
573 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
574
575 /*
576 * Structure of an inode on the disk
577 */
578 struct ext4_inode {
579 __le16 i_mode; /* File mode */
580 __le16 i_uid; /* Low 16 bits of Owner Uid */
581 __le32 i_size_lo; /* Size in bytes */
582 __le32 i_atime; /* Access time */
583 __le32 i_ctime; /* Inode Change time */
584 __le32 i_mtime; /* Modification time */
585 __le32 i_dtime; /* Deletion Time */
586 __le16 i_gid; /* Low 16 bits of Group Id */
587 __le16 i_links_count; /* Links count */
588 __le32 i_blocks_lo; /* Blocks count */
589 __le32 i_flags; /* File flags */
590 union {
591 struct {
592 __le32 l_i_version;
593 } linux1;
594 struct {
595 __u32 h_i_translator;
596 } hurd1;
597 struct {
598 __u32 m_i_reserved1;
599 } masix1;
600 } osd1; /* OS dependent 1 */
601 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
602 __le32 i_generation; /* File version (for NFS) */
603 __le32 i_file_acl_lo; /* File ACL */
604 __le32 i_size_high;
605 __le32 i_obso_faddr; /* Obsoleted fragment address */
606 union {
607 struct {
608 __le16 l_i_blocks_high; /* were l_i_reserved1 */
609 __le16 l_i_file_acl_high;
610 __le16 l_i_uid_high; /* these 2 fields */
611 __le16 l_i_gid_high; /* were reserved2[0] */
612 __u32 l_i_reserved2;
613 } linux2;
614 struct {
615 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
616 __u16 h_i_mode_high;
617 __u16 h_i_uid_high;
618 __u16 h_i_gid_high;
619 __u32 h_i_author;
620 } hurd2;
621 struct {
622 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
623 __le16 m_i_file_acl_high;
624 __u32 m_i_reserved2[2];
625 } masix2;
626 } osd2; /* OS dependent 2 */
627 __le16 i_extra_isize;
628 __le16 i_pad1;
629 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
630 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
631 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
632 __le32 i_crtime; /* File Creation time */
633 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
634 __le32 i_version_hi; /* high 32 bits for 64-bit version */
635 };
636
637 struct move_extent {
638 __u32 reserved; /* should be zero */
639 __u32 donor_fd; /* donor file descriptor */
640 __u64 orig_start; /* logical start offset in block for orig */
641 __u64 donor_start; /* logical start offset in block for donor */
642 __u64 len; /* block length to be moved */
643 __u64 moved_len; /* moved block length */
644 };
645
646 #define EXT4_EPOCH_BITS 2
647 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
648 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
649
650 /*
651 * Extended fields will fit into an inode if the filesystem was formatted
652 * with large inodes (-I 256 or larger) and there are not currently any EAs
653 * consuming all of the available space. For new inodes we always reserve
654 * enough space for the kernel's known extended fields, but for inodes
655 * created with an old kernel this might not have been the case. None of
656 * the extended inode fields is critical for correct filesystem operation.
657 * This macro checks if a certain field fits in the inode. Note that
658 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
659 */
660 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
661 ((offsetof(typeof(*ext4_inode), field) + \
662 sizeof((ext4_inode)->field)) \
663 <= (EXT4_GOOD_OLD_INODE_SIZE + \
664 (einode)->i_extra_isize)) \
665
666 static inline __le32 ext4_encode_extra_time(struct timespec *time)
667 {
668 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
669 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
670 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
671 }
672
673 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
674 {
675 if (sizeof(time->tv_sec) > 4)
676 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
677 << 32;
678 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
679 }
680
681 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
682 do { \
683 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
684 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
685 (raw_inode)->xtime ## _extra = \
686 ext4_encode_extra_time(&(inode)->xtime); \
687 } while (0)
688
689 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
690 do { \
691 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
692 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
693 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
694 (raw_inode)->xtime ## _extra = \
695 ext4_encode_extra_time(&(einode)->xtime); \
696 } while (0)
697
698 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
699 do { \
700 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
701 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
702 ext4_decode_extra_time(&(inode)->xtime, \
703 raw_inode->xtime ## _extra); \
704 else \
705 (inode)->xtime.tv_nsec = 0; \
706 } while (0)
707
708 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
709 do { \
710 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
711 (einode)->xtime.tv_sec = \
712 (signed)le32_to_cpu((raw_inode)->xtime); \
713 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
714 ext4_decode_extra_time(&(einode)->xtime, \
715 raw_inode->xtime ## _extra); \
716 else \
717 (einode)->xtime.tv_nsec = 0; \
718 } while (0)
719
720 #define i_disk_version osd1.linux1.l_i_version
721
722 #if defined(__KERNEL__) || defined(__linux__)
723 #define i_reserved1 osd1.linux1.l_i_reserved1
724 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
725 #define i_blocks_high osd2.linux2.l_i_blocks_high
726 #define i_uid_low i_uid
727 #define i_gid_low i_gid
728 #define i_uid_high osd2.linux2.l_i_uid_high
729 #define i_gid_high osd2.linux2.l_i_gid_high
730 #define i_reserved2 osd2.linux2.l_i_reserved2
731
732 #elif defined(__GNU__)
733
734 #define i_translator osd1.hurd1.h_i_translator
735 #define i_uid_high osd2.hurd2.h_i_uid_high
736 #define i_gid_high osd2.hurd2.h_i_gid_high
737 #define i_author osd2.hurd2.h_i_author
738
739 #elif defined(__masix__)
740
741 #define i_reserved1 osd1.masix1.m_i_reserved1
742 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
743 #define i_reserved2 osd2.masix2.m_i_reserved2
744
745 #endif /* defined(__KERNEL__) || defined(__linux__) */
746
747 /*
748 * storage for cached extent
749 * If ec_len == 0, then the cache is invalid.
750 * If ec_start == 0, then the cache represents a gap (null mapping)
751 */
752 struct ext4_ext_cache {
753 ext4_fsblk_t ec_start;
754 ext4_lblk_t ec_block;
755 __u32 ec_len; /* must be 32bit to return holes */
756 };
757
758 /*
759 * fourth extended file system inode data in memory
760 */
761 struct ext4_inode_info {
762 __le32 i_data[15]; /* unconverted */
763 __u32 i_dtime;
764 ext4_fsblk_t i_file_acl;
765
766 /*
767 * i_block_group is the number of the block group which contains
768 * this file's inode. Constant across the lifetime of the inode,
769 * it is ued for making block allocation decisions - we try to
770 * place a file's data blocks near its inode block, and new inodes
771 * near to their parent directory's inode.
772 */
773 ext4_group_t i_block_group;
774 ext4_lblk_t i_dir_start_lookup;
775 #if (BITS_PER_LONG < 64)
776 unsigned long i_state_flags; /* Dynamic state flags */
777 #endif
778 unsigned long i_flags;
779
780 #ifdef CONFIG_EXT4_FS_XATTR
781 /*
782 * Extended attributes can be read independently of the main file
783 * data. Taking i_mutex even when reading would cause contention
784 * between readers of EAs and writers of regular file data, so
785 * instead we synchronize on xattr_sem when reading or changing
786 * EAs.
787 */
788 struct rw_semaphore xattr_sem;
789 #endif
790
791 struct list_head i_orphan; /* unlinked but open inodes */
792
793 /*
794 * i_disksize keeps track of what the inode size is ON DISK, not
795 * in memory. During truncate, i_size is set to the new size by
796 * the VFS prior to calling ext4_truncate(), but the filesystem won't
797 * set i_disksize to 0 until the truncate is actually under way.
798 *
799 * The intent is that i_disksize always represents the blocks which
800 * are used by this file. This allows recovery to restart truncate
801 * on orphans if we crash during truncate. We actually write i_disksize
802 * into the on-disk inode when writing inodes out, instead of i_size.
803 *
804 * The only time when i_disksize and i_size may be different is when
805 * a truncate is in progress. The only things which change i_disksize
806 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
807 */
808 loff_t i_disksize;
809
810 /*
811 * i_data_sem is for serialising ext4_truncate() against
812 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
813 * data tree are chopped off during truncate. We can't do that in
814 * ext4 because whenever we perform intermediate commits during
815 * truncate, the inode and all the metadata blocks *must* be in a
816 * consistent state which allows truncation of the orphans to restart
817 * during recovery. Hence we must fix the get_block-vs-truncate race
818 * by other means, so we have i_data_sem.
819 */
820 struct rw_semaphore i_data_sem;
821 struct inode vfs_inode;
822 struct jbd2_inode *jinode;
823
824 struct ext4_ext_cache i_cached_extent;
825 /*
826 * File creation time. Its function is same as that of
827 * struct timespec i_{a,c,m}time in the generic inode.
828 */
829 struct timespec i_crtime;
830
831 /* mballoc */
832 struct list_head i_prealloc_list;
833 spinlock_t i_prealloc_lock;
834
835 /* ialloc */
836 ext4_group_t i_last_alloc_group;
837
838 /* allocation reservation info for delalloc */
839 unsigned int i_reserved_data_blocks;
840 unsigned int i_reserved_meta_blocks;
841 unsigned int i_allocated_meta_blocks;
842 ext4_lblk_t i_da_metadata_calc_last_lblock;
843 int i_da_metadata_calc_len;
844
845 /* on-disk additional length */
846 __u16 i_extra_isize;
847
848 #ifdef CONFIG_QUOTA
849 /* quota space reservation, managed internally by quota code */
850 qsize_t i_reserved_quota;
851 #endif
852
853 /* completed IOs that might need unwritten extents handling */
854 struct list_head i_completed_io_list;
855 spinlock_t i_completed_io_lock;
856 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
857 /* current io_end structure for async DIO write*/
858 ext4_io_end_t *cur_aio_dio;
859 atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
860
861 spinlock_t i_block_reservation_lock;
862
863 /*
864 * Transactions that contain inode's metadata needed to complete
865 * fsync and fdatasync, respectively.
866 */
867 tid_t i_sync_tid;
868 tid_t i_datasync_tid;
869 };
870
871 /*
872 * File system states
873 */
874 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
875 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
876 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
877
878 /*
879 * Misc. filesystem flags
880 */
881 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
882 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
883 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
884
885 /*
886 * Mount flags
887 */
888 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
889 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
890 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
891 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
892 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
893 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
894 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
895 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
896 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
897 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
898 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
899 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
900 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
901 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
902 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
903 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
904 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
905 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
906 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
907 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
908 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
909 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
910 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
911 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
912 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
913 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
914 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
915 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
916 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
917 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
918 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
919
920 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
921 ~EXT4_MOUNT_##opt
922 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
923 EXT4_MOUNT_##opt
924 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
925 EXT4_MOUNT_##opt)
926
927 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
928 ~EXT4_MOUNT2_##opt
929 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
930 EXT4_MOUNT2_##opt
931 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
932 EXT4_MOUNT2_##opt)
933
934 #define ext4_test_and_set_bit __test_and_set_bit_le
935 #define ext4_set_bit __set_bit_le
936 #define ext4_set_bit_atomic ext2_set_bit_atomic
937 #define ext4_test_and_clear_bit __test_and_clear_bit_le
938 #define ext4_clear_bit __clear_bit_le
939 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
940 #define ext4_test_bit test_bit_le
941 #define ext4_find_next_zero_bit find_next_zero_bit_le
942 #define ext4_find_next_bit find_next_bit_le
943
944 /*
945 * Maximal mount counts between two filesystem checks
946 */
947 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
948 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
949
950 /*
951 * Behaviour when detecting errors
952 */
953 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
954 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
955 #define EXT4_ERRORS_PANIC 3 /* Panic */
956 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
957
958 /*
959 * Structure of the super block
960 */
961 struct ext4_super_block {
962 /*00*/ __le32 s_inodes_count; /* Inodes count */
963 __le32 s_blocks_count_lo; /* Blocks count */
964 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
965 __le32 s_free_blocks_count_lo; /* Free blocks count */
966 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
967 __le32 s_first_data_block; /* First Data Block */
968 __le32 s_log_block_size; /* Block size */
969 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
970 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
971 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
972 __le32 s_inodes_per_group; /* # Inodes per group */
973 __le32 s_mtime; /* Mount time */
974 /*30*/ __le32 s_wtime; /* Write time */
975 __le16 s_mnt_count; /* Mount count */
976 __le16 s_max_mnt_count; /* Maximal mount count */
977 __le16 s_magic; /* Magic signature */
978 __le16 s_state; /* File system state */
979 __le16 s_errors; /* Behaviour when detecting errors */
980 __le16 s_minor_rev_level; /* minor revision level */
981 /*40*/ __le32 s_lastcheck; /* time of last check */
982 __le32 s_checkinterval; /* max. time between checks */
983 __le32 s_creator_os; /* OS */
984 __le32 s_rev_level; /* Revision level */
985 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
986 __le16 s_def_resgid; /* Default gid for reserved blocks */
987 /*
988 * These fields are for EXT4_DYNAMIC_REV superblocks only.
989 *
990 * Note: the difference between the compatible feature set and
991 * the incompatible feature set is that if there is a bit set
992 * in the incompatible feature set that the kernel doesn't
993 * know about, it should refuse to mount the filesystem.
994 *
995 * e2fsck's requirements are more strict; if it doesn't know
996 * about a feature in either the compatible or incompatible
997 * feature set, it must abort and not try to meddle with
998 * things it doesn't understand...
999 */
1000 __le32 s_first_ino; /* First non-reserved inode */
1001 __le16 s_inode_size; /* size of inode structure */
1002 __le16 s_block_group_nr; /* block group # of this superblock */
1003 __le32 s_feature_compat; /* compatible feature set */
1004 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1005 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1006 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1007 /*78*/ char s_volume_name[16]; /* volume name */
1008 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1009 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1010 /*
1011 * Performance hints. Directory preallocation should only
1012 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1013 */
1014 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1015 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1016 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1017 /*
1018 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1019 */
1020 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1021 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1022 __le32 s_journal_dev; /* device number of journal file */
1023 __le32 s_last_orphan; /* start of list of inodes to delete */
1024 __le32 s_hash_seed[4]; /* HTREE hash seed */
1025 __u8 s_def_hash_version; /* Default hash version to use */
1026 __u8 s_jnl_backup_type;
1027 __le16 s_desc_size; /* size of group descriptor */
1028 /*100*/ __le32 s_default_mount_opts;
1029 __le32 s_first_meta_bg; /* First metablock block group */
1030 __le32 s_mkfs_time; /* When the filesystem was created */
1031 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1032 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1033 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1034 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1035 __le32 s_free_blocks_count_hi; /* Free blocks count */
1036 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1037 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1038 __le32 s_flags; /* Miscellaneous flags */
1039 __le16 s_raid_stride; /* RAID stride */
1040 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1041 __le64 s_mmp_block; /* Block for multi-mount protection */
1042 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1043 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1044 __u8 s_reserved_char_pad;
1045 __le16 s_reserved_pad;
1046 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1047 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1048 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1049 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1050 snapshot's future use */
1051 __le32 s_snapshot_list; /* inode number of the head of the
1052 on-disk snapshot list */
1053 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1054 __le32 s_error_count; /* number of fs errors */
1055 __le32 s_first_error_time; /* first time an error happened */
1056 __le32 s_first_error_ino; /* inode involved in first error */
1057 __le64 s_first_error_block; /* block involved of first error */
1058 __u8 s_first_error_func[32]; /* function where the error happened */
1059 __le32 s_first_error_line; /* line number where error happened */
1060 __le32 s_last_error_time; /* most recent time of an error */
1061 __le32 s_last_error_ino; /* inode involved in last error */
1062 __le32 s_last_error_line; /* line number where error happened */
1063 __le64 s_last_error_block; /* block involved of last error */
1064 __u8 s_last_error_func[32]; /* function where the error happened */
1065 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1066 __u8 s_mount_opts[64];
1067 __le32 s_reserved[112]; /* Padding to the end of the block */
1068 };
1069
1070 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1071
1072 #ifdef __KERNEL__
1073
1074 /*
1075 * run-time mount flags
1076 */
1077 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1078 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1079
1080 /*
1081 * fourth extended-fs super-block data in memory
1082 */
1083 struct ext4_sb_info {
1084 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1085 unsigned long s_inodes_per_block;/* Number of inodes per block */
1086 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1087 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1088 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1089 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1090 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1091 ext4_group_t s_groups_count; /* Number of groups in the fs */
1092 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1093 unsigned long s_overhead_last; /* Last calculated overhead */
1094 unsigned long s_blocks_last; /* Last seen block count */
1095 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1096 struct buffer_head * s_sbh; /* Buffer containing the super block */
1097 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1098 struct buffer_head **s_group_desc;
1099 unsigned int s_mount_opt;
1100 unsigned int s_mount_opt2;
1101 unsigned int s_mount_flags;
1102 ext4_fsblk_t s_sb_block;
1103 uid_t s_resuid;
1104 gid_t s_resgid;
1105 unsigned short s_mount_state;
1106 unsigned short s_pad;
1107 int s_addr_per_block_bits;
1108 int s_desc_per_block_bits;
1109 int s_inode_size;
1110 int s_first_ino;
1111 unsigned int s_inode_readahead_blks;
1112 unsigned int s_inode_goal;
1113 spinlock_t s_next_gen_lock;
1114 u32 s_next_generation;
1115 u32 s_hash_seed[4];
1116 int s_def_hash_version;
1117 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1118 struct percpu_counter s_freeblocks_counter;
1119 struct percpu_counter s_freeinodes_counter;
1120 struct percpu_counter s_dirs_counter;
1121 struct percpu_counter s_dirtyblocks_counter;
1122 struct blockgroup_lock *s_blockgroup_lock;
1123 struct proc_dir_entry *s_proc;
1124 struct kobject s_kobj;
1125 struct completion s_kobj_unregister;
1126
1127 /* Journaling */
1128 struct journal_s *s_journal;
1129 struct list_head s_orphan;
1130 struct mutex s_orphan_lock;
1131 struct mutex s_resize_lock;
1132 unsigned long s_commit_interval;
1133 u32 s_max_batch_time;
1134 u32 s_min_batch_time;
1135 struct block_device *journal_bdev;
1136 #ifdef CONFIG_JBD2_DEBUG
1137 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
1138 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
1139 #endif
1140 #ifdef CONFIG_QUOTA
1141 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1142 int s_jquota_fmt; /* Format of quota to use */
1143 #endif
1144 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1145 struct rb_root system_blks;
1146
1147 #ifdef EXTENTS_STATS
1148 /* ext4 extents stats */
1149 unsigned long s_ext_min;
1150 unsigned long s_ext_max;
1151 unsigned long s_depth_max;
1152 spinlock_t s_ext_stats_lock;
1153 unsigned long s_ext_blocks;
1154 unsigned long s_ext_extents;
1155 #endif
1156 /* ext4 extent cache stats */
1157 unsigned long extent_cache_hits;
1158 unsigned long extent_cache_misses;
1159
1160 /* for buddy allocator */
1161 struct ext4_group_info ***s_group_info;
1162 struct inode *s_buddy_cache;
1163 spinlock_t s_md_lock;
1164 unsigned short *s_mb_offsets;
1165 unsigned int *s_mb_maxs;
1166
1167 /* tunables */
1168 unsigned long s_stripe;
1169 unsigned int s_mb_stream_request;
1170 unsigned int s_mb_max_to_scan;
1171 unsigned int s_mb_min_to_scan;
1172 unsigned int s_mb_stats;
1173 unsigned int s_mb_order2_reqs;
1174 unsigned int s_mb_group_prealloc;
1175 unsigned int s_max_writeback_mb_bump;
1176 /* where last allocation was done - for stream allocation */
1177 unsigned long s_mb_last_group;
1178 unsigned long s_mb_last_start;
1179
1180 /* stats for buddy allocator */
1181 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1182 atomic_t s_bal_success; /* we found long enough chunks */
1183 atomic_t s_bal_allocated; /* in blocks */
1184 atomic_t s_bal_ex_scanned; /* total extents scanned */
1185 atomic_t s_bal_goals; /* goal hits */
1186 atomic_t s_bal_breaks; /* too long searches */
1187 atomic_t s_bal_2orders; /* 2^order hits */
1188 spinlock_t s_bal_lock;
1189 unsigned long s_mb_buddies_generated;
1190 unsigned long long s_mb_generation_time;
1191 atomic_t s_mb_lost_chunks;
1192 atomic_t s_mb_preallocated;
1193 atomic_t s_mb_discarded;
1194 atomic_t s_lock_busy;
1195
1196 /* locality groups */
1197 struct ext4_locality_group __percpu *s_locality_groups;
1198
1199 /* for write statistics */
1200 unsigned long s_sectors_written_start;
1201 u64 s_kbytes_written;
1202
1203 unsigned int s_log_groups_per_flex;
1204 struct flex_groups *s_flex_groups;
1205
1206 /* workqueue for dio unwritten */
1207 struct workqueue_struct *dio_unwritten_wq;
1208
1209 /* timer for periodic error stats printing */
1210 struct timer_list s_err_report;
1211
1212 /* Lazy inode table initialization info */
1213 struct ext4_li_request *s_li_request;
1214 /* Wait multiplier for lazy initialization thread */
1215 unsigned int s_li_wait_mult;
1216
1217 /* Kernel thread for multiple mount protection */
1218 struct task_struct *s_mmp_tsk;
1219
1220 /* record the last minlen when FITRIM is called. */
1221 atomic_t s_last_trim_minblks;
1222 };
1223
1224 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1225 {
1226 return sb->s_fs_info;
1227 }
1228 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1229 {
1230 return container_of(inode, struct ext4_inode_info, vfs_inode);
1231 }
1232
1233 static inline struct timespec ext4_current_time(struct inode *inode)
1234 {
1235 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1236 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1237 }
1238
1239 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1240 {
1241 return ino == EXT4_ROOT_INO ||
1242 ino == EXT4_JOURNAL_INO ||
1243 ino == EXT4_RESIZE_INO ||
1244 (ino >= EXT4_FIRST_INO(sb) &&
1245 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1246 }
1247
1248 /*
1249 * Inode dynamic state flags
1250 */
1251 enum {
1252 EXT4_STATE_JDATA, /* journaled data exists */
1253 EXT4_STATE_NEW, /* inode is newly created */
1254 EXT4_STATE_XATTR, /* has in-inode xattrs */
1255 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1256 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1257 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1258 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1259 EXT4_STATE_NEWENTRY, /* File just added to dir */
1260 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1261 };
1262
1263 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1264 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1265 { \
1266 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1267 } \
1268 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1269 { \
1270 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1271 } \
1272 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1273 { \
1274 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1275 }
1276
1277 EXT4_INODE_BIT_FNS(flag, flags, 0)
1278 #if (BITS_PER_LONG < 64)
1279 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1280
1281 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1282 {
1283 (ei)->i_state_flags = 0;
1284 }
1285 #else
1286 EXT4_INODE_BIT_FNS(state, flags, 32)
1287
1288 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1289 {
1290 /* We depend on the fact that callers will set i_flags */
1291 }
1292 #endif
1293 #else
1294 /* Assume that user mode programs are passing in an ext4fs superblock, not
1295 * a kernel struct super_block. This will allow us to call the feature-test
1296 * macros from user land. */
1297 #define EXT4_SB(sb) (sb)
1298 #endif
1299
1300 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1301
1302 /*
1303 * Codes for operating systems
1304 */
1305 #define EXT4_OS_LINUX 0
1306 #define EXT4_OS_HURD 1
1307 #define EXT4_OS_MASIX 2
1308 #define EXT4_OS_FREEBSD 3
1309 #define EXT4_OS_LITES 4
1310
1311 /*
1312 * Revision levels
1313 */
1314 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1315 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1316
1317 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1318 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1319
1320 #define EXT4_GOOD_OLD_INODE_SIZE 128
1321
1322 /*
1323 * Feature set definitions
1324 */
1325
1326 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1327 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1328 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1329 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1330 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1331 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1332 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1333 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1334 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1335 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1336 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1337 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1338 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1339 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1340 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1341 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1342 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1343 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1344
1345 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1346 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1347 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1348 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1349 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1350 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1351
1352 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1353 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1354 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1355 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1356 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1357 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1358 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1359 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1360
1361 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1362 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1363 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1364 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1365 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1366 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1367 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1368 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1369 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1370 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1371 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1372
1373 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1374 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1375 EXT4_FEATURE_INCOMPAT_META_BG)
1376 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1377 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1378 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1379
1380 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1381 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1382 EXT4_FEATURE_INCOMPAT_RECOVER| \
1383 EXT4_FEATURE_INCOMPAT_META_BG)
1384 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1385 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1386 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1387
1388 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1389 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1390 EXT4_FEATURE_INCOMPAT_RECOVER| \
1391 EXT4_FEATURE_INCOMPAT_META_BG| \
1392 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1393 EXT4_FEATURE_INCOMPAT_64BIT| \
1394 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1395 EXT4_FEATURE_INCOMPAT_MMP)
1396 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1397 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1398 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1399 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1400 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1401 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1402 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1403
1404 /*
1405 * Default values for user and/or group using reserved blocks
1406 */
1407 #define EXT4_DEF_RESUID 0
1408 #define EXT4_DEF_RESGID 0
1409
1410 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1411
1412 /*
1413 * Default mount options
1414 */
1415 #define EXT4_DEFM_DEBUG 0x0001
1416 #define EXT4_DEFM_BSDGROUPS 0x0002
1417 #define EXT4_DEFM_XATTR_USER 0x0004
1418 #define EXT4_DEFM_ACL 0x0008
1419 #define EXT4_DEFM_UID16 0x0010
1420 #define EXT4_DEFM_JMODE 0x0060
1421 #define EXT4_DEFM_JMODE_DATA 0x0020
1422 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1423 #define EXT4_DEFM_JMODE_WBACK 0x0060
1424 #define EXT4_DEFM_NOBARRIER 0x0100
1425 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1426 #define EXT4_DEFM_DISCARD 0x0400
1427 #define EXT4_DEFM_NODELALLOC 0x0800
1428
1429 /*
1430 * Default journal batch times
1431 */
1432 #define EXT4_DEF_MIN_BATCH_TIME 0
1433 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1434
1435 /*
1436 * Minimum number of groups in a flexgroup before we separate out
1437 * directories into the first block group of a flexgroup
1438 */
1439 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1440
1441 /*
1442 * Structure of a directory entry
1443 */
1444 #define EXT4_NAME_LEN 255
1445
1446 struct ext4_dir_entry {
1447 __le32 inode; /* Inode number */
1448 __le16 rec_len; /* Directory entry length */
1449 __le16 name_len; /* Name length */
1450 char name[EXT4_NAME_LEN]; /* File name */
1451 };
1452
1453 /*
1454 * The new version of the directory entry. Since EXT4 structures are
1455 * stored in intel byte order, and the name_len field could never be
1456 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1457 * file_type field.
1458 */
1459 struct ext4_dir_entry_2 {
1460 __le32 inode; /* Inode number */
1461 __le16 rec_len; /* Directory entry length */
1462 __u8 name_len; /* Name length */
1463 __u8 file_type;
1464 char name[EXT4_NAME_LEN]; /* File name */
1465 };
1466
1467 /*
1468 * Ext4 directory file types. Only the low 3 bits are used. The
1469 * other bits are reserved for now.
1470 */
1471 #define EXT4_FT_UNKNOWN 0
1472 #define EXT4_FT_REG_FILE 1
1473 #define EXT4_FT_DIR 2
1474 #define EXT4_FT_CHRDEV 3
1475 #define EXT4_FT_BLKDEV 4
1476 #define EXT4_FT_FIFO 5
1477 #define EXT4_FT_SOCK 6
1478 #define EXT4_FT_SYMLINK 7
1479
1480 #define EXT4_FT_MAX 8
1481
1482 /*
1483 * EXT4_DIR_PAD defines the directory entries boundaries
1484 *
1485 * NOTE: It must be a multiple of 4
1486 */
1487 #define EXT4_DIR_PAD 4
1488 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1489 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1490 ~EXT4_DIR_ROUND)
1491 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1492
1493 /*
1494 * If we ever get support for fs block sizes > page_size, we'll need
1495 * to remove the #if statements in the next two functions...
1496 */
1497 static inline unsigned int
1498 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1499 {
1500 unsigned len = le16_to_cpu(dlen);
1501
1502 #if (PAGE_CACHE_SIZE >= 65536)
1503 if (len == EXT4_MAX_REC_LEN || len == 0)
1504 return blocksize;
1505 return (len & 65532) | ((len & 3) << 16);
1506 #else
1507 return len;
1508 #endif
1509 }
1510
1511 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1512 {
1513 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1514 BUG();
1515 #if (PAGE_CACHE_SIZE >= 65536)
1516 if (len < 65536)
1517 return cpu_to_le16(len);
1518 if (len == blocksize) {
1519 if (blocksize == 65536)
1520 return cpu_to_le16(EXT4_MAX_REC_LEN);
1521 else
1522 return cpu_to_le16(0);
1523 }
1524 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1525 #else
1526 return cpu_to_le16(len);
1527 #endif
1528 }
1529
1530 /*
1531 * Hash Tree Directory indexing
1532 * (c) Daniel Phillips, 2001
1533 */
1534
1535 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1536 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1537 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1538 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1539 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1540
1541 /* Legal values for the dx_root hash_version field: */
1542
1543 #define DX_HASH_LEGACY 0
1544 #define DX_HASH_HALF_MD4 1
1545 #define DX_HASH_TEA 2
1546 #define DX_HASH_LEGACY_UNSIGNED 3
1547 #define DX_HASH_HALF_MD4_UNSIGNED 4
1548 #define DX_HASH_TEA_UNSIGNED 5
1549
1550 #ifdef __KERNEL__
1551
1552 /* hash info structure used by the directory hash */
1553 struct dx_hash_info
1554 {
1555 u32 hash;
1556 u32 minor_hash;
1557 int hash_version;
1558 u32 *seed;
1559 };
1560
1561 #define EXT4_HTREE_EOF 0x7fffffff
1562
1563 /*
1564 * Control parameters used by ext4_htree_next_block
1565 */
1566 #define HASH_NB_ALWAYS 1
1567
1568
1569 /*
1570 * Describe an inode's exact location on disk and in memory
1571 */
1572 struct ext4_iloc
1573 {
1574 struct buffer_head *bh;
1575 unsigned long offset;
1576 ext4_group_t block_group;
1577 };
1578
1579 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1580 {
1581 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1582 }
1583
1584 /*
1585 * This structure is stuffed into the struct file's private_data field
1586 * for directories. It is where we put information so that we can do
1587 * readdir operations in hash tree order.
1588 */
1589 struct dir_private_info {
1590 struct rb_root root;
1591 struct rb_node *curr_node;
1592 struct fname *extra_fname;
1593 loff_t last_pos;
1594 __u32 curr_hash;
1595 __u32 curr_minor_hash;
1596 __u32 next_hash;
1597 };
1598
1599 /* calculate the first block number of the group */
1600 static inline ext4_fsblk_t
1601 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1602 {
1603 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1604 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1605 }
1606
1607 /*
1608 * Special error return code only used by dx_probe() and its callers.
1609 */
1610 #define ERR_BAD_DX_DIR -75000
1611
1612 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1613 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1614
1615 /*
1616 * Timeout and state flag for lazy initialization inode thread.
1617 */
1618 #define EXT4_DEF_LI_WAIT_MULT 10
1619 #define EXT4_DEF_LI_MAX_START_DELAY 5
1620 #define EXT4_LAZYINIT_QUIT 0x0001
1621 #define EXT4_LAZYINIT_RUNNING 0x0002
1622
1623 /*
1624 * Lazy inode table initialization info
1625 */
1626 struct ext4_lazy_init {
1627 unsigned long li_state;
1628 struct list_head li_request_list;
1629 struct mutex li_list_mtx;
1630 };
1631
1632 struct ext4_li_request {
1633 struct super_block *lr_super;
1634 struct ext4_sb_info *lr_sbi;
1635 ext4_group_t lr_next_group;
1636 struct list_head lr_request;
1637 unsigned long lr_next_sched;
1638 unsigned long lr_timeout;
1639 };
1640
1641 struct ext4_features {
1642 struct kobject f_kobj;
1643 struct completion f_kobj_unregister;
1644 };
1645
1646 /*
1647 * This structure will be used for multiple mount protection. It will be
1648 * written into the block number saved in the s_mmp_block field in the
1649 * superblock. Programs that check MMP should assume that if
1650 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1651 * to use the filesystem, regardless of how old the timestamp is.
1652 */
1653 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1654 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1655 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1656 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1657
1658 struct mmp_struct {
1659 __le32 mmp_magic; /* Magic number for MMP */
1660 __le32 mmp_seq; /* Sequence no. updated periodically */
1661
1662 /*
1663 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1664 * purposes and do not affect the correctness of the algorithm
1665 */
1666 __le64 mmp_time; /* Time last updated */
1667 char mmp_nodename[64]; /* Node which last updated MMP block */
1668 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1669
1670 /*
1671 * mmp_check_interval is used to verify if the MMP block has been
1672 * updated on the block device. The value is updated based on the
1673 * maximum time to write the MMP block during an update cycle.
1674 */
1675 __le16 mmp_check_interval;
1676
1677 __le16 mmp_pad1;
1678 __le32 mmp_pad2[227];
1679 };
1680
1681 /* arguments passed to the mmp thread */
1682 struct mmpd_data {
1683 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1684 struct super_block *sb; /* super block of the fs */
1685 };
1686
1687 /*
1688 * Check interval multiplier
1689 * The MMP block is written every update interval and initially checked every
1690 * update interval x the multiplier (the value is then adapted based on the
1691 * write latency). The reason is that writes can be delayed under load and we
1692 * don't want readers to incorrectly assume that the filesystem is no longer
1693 * in use.
1694 */
1695 #define EXT4_MMP_CHECK_MULT 2UL
1696
1697 /*
1698 * Minimum interval for MMP checking in seconds.
1699 */
1700 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1701
1702 /*
1703 * Maximum interval for MMP checking in seconds.
1704 */
1705 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1706
1707 /*
1708 * Function prototypes
1709 */
1710
1711 /*
1712 * Ok, these declarations are also in <linux/kernel.h> but none of the
1713 * ext4 source programs needs to include it so they are duplicated here.
1714 */
1715 # define NORET_TYPE /**/
1716 # define ATTRIB_NORET __attribute__((noreturn))
1717 # define NORET_AND noreturn,
1718
1719 /* bitmap.c */
1720 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1721
1722 /* balloc.c */
1723 extern unsigned int ext4_block_group(struct super_block *sb,
1724 ext4_fsblk_t blocknr);
1725 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1726 ext4_fsblk_t blocknr);
1727 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1728 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1729 ext4_group_t group);
1730 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1731 ext4_fsblk_t goal,
1732 unsigned int flags,
1733 unsigned long *count,
1734 int *errp);
1735 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi,
1736 s64 nblocks, unsigned int flags);
1737 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1738 extern void ext4_check_blocks_bitmap(struct super_block *);
1739 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1740 ext4_group_t block_group,
1741 struct buffer_head ** bh);
1742 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1743 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1744 ext4_group_t block_group);
1745 extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1746 struct buffer_head *bh,
1747 ext4_group_t group,
1748 struct ext4_group_desc *desc);
1749 #define ext4_free_blocks_after_init(sb, group, desc) \
1750 ext4_init_block_bitmap(sb, NULL, group, desc)
1751 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1752
1753 /* dir.c */
1754 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1755 struct file *,
1756 struct ext4_dir_entry_2 *,
1757 struct buffer_head *, unsigned int);
1758 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1759 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1760 (de), (bh), (offset)))
1761 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1762 __u32 minor_hash,
1763 struct ext4_dir_entry_2 *dirent);
1764 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1765
1766 /* fsync.c */
1767 extern int ext4_sync_file(struct file *, int);
1768 extern int ext4_flush_completed_IO(struct inode *);
1769
1770 /* hash.c */
1771 extern int ext4fs_dirhash(const char *name, int len, struct
1772 dx_hash_info *hinfo);
1773
1774 /* ialloc.c */
1775 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1776 const struct qstr *qstr, __u32 goal);
1777 extern void ext4_free_inode(handle_t *, struct inode *);
1778 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1779 extern unsigned long ext4_count_free_inodes(struct super_block *);
1780 extern unsigned long ext4_count_dirs(struct super_block *);
1781 extern void ext4_check_inodes_bitmap(struct super_block *);
1782 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1783 extern int ext4_init_inode_table(struct super_block *sb,
1784 ext4_group_t group, int barrier);
1785
1786 /* mballoc.c */
1787 extern long ext4_mb_stats;
1788 extern long ext4_mb_max_to_scan;
1789 extern int ext4_mb_init(struct super_block *, int);
1790 extern int ext4_mb_release(struct super_block *);
1791 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1792 struct ext4_allocation_request *, int *);
1793 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1794 extern void ext4_discard_preallocations(struct inode *);
1795 extern int __init ext4_init_mballoc(void);
1796 extern void ext4_exit_mballoc(void);
1797 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1798 struct buffer_head *bh, ext4_fsblk_t block,
1799 unsigned long count, int flags);
1800 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1801 ext4_group_t i, struct ext4_group_desc *desc);
1802 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1803 ext4_fsblk_t block, unsigned long count);
1804 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1805
1806 /* inode.c */
1807 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1808 ext4_lblk_t, int, int *);
1809 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1810 ext4_lblk_t, int, int *);
1811 int ext4_get_block(struct inode *inode, sector_t iblock,
1812 struct buffer_head *bh_result, int create);
1813
1814 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1815 extern int ext4_write_inode(struct inode *, struct writeback_control *);
1816 extern int ext4_setattr(struct dentry *, struct iattr *);
1817 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1818 struct kstat *stat);
1819 extern void ext4_evict_inode(struct inode *);
1820 extern void ext4_clear_inode(struct inode *);
1821 extern int ext4_sync_inode(handle_t *, struct inode *);
1822 extern void ext4_dirty_inode(struct inode *, int);
1823 extern int ext4_change_inode_journal_flag(struct inode *, int);
1824 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1825 extern int ext4_can_truncate(struct inode *inode);
1826 extern void ext4_truncate(struct inode *);
1827 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1828 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1829 extern void ext4_set_inode_flags(struct inode *);
1830 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1831 extern int ext4_alloc_da_blocks(struct inode *inode);
1832 extern void ext4_set_aops(struct inode *inode);
1833 extern int ext4_writepage_trans_blocks(struct inode *);
1834 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1835 extern int ext4_block_truncate_page(handle_t *handle,
1836 struct address_space *mapping, loff_t from);
1837 extern int ext4_block_zero_page_range(handle_t *handle,
1838 struct address_space *mapping, loff_t from, loff_t length);
1839 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1840 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1841 extern void ext4_da_update_reserve_space(struct inode *inode,
1842 int used, int quota_claim);
1843
1844 /* indirect.c */
1845 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
1846 struct ext4_map_blocks *map, int flags);
1847 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
1848 const struct iovec *iov, loff_t offset,
1849 unsigned long nr_segs);
1850 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
1851 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
1852 extern void ext4_ind_truncate(struct inode *inode);
1853
1854 /* ioctl.c */
1855 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1856 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1857
1858 /* migrate.c */
1859 extern int ext4_ext_migrate(struct inode *);
1860
1861 /* namei.c */
1862 extern int ext4_orphan_add(handle_t *, struct inode *);
1863 extern int ext4_orphan_del(handle_t *, struct inode *);
1864 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1865 __u32 start_minor_hash, __u32 *next_hash);
1866
1867 /* resize.c */
1868 extern int ext4_group_add(struct super_block *sb,
1869 struct ext4_new_group_data *input);
1870 extern int ext4_group_extend(struct super_block *sb,
1871 struct ext4_super_block *es,
1872 ext4_fsblk_t n_blocks_count);
1873
1874 /* super.c */
1875 extern void __ext4_error(struct super_block *, const char *, unsigned int,
1876 const char *, ...)
1877 __attribute__ ((format (printf, 4, 5)));
1878 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1879 __LINE__, ## message)
1880 extern void ext4_error_inode(struct inode *, const char *, unsigned int,
1881 ext4_fsblk_t, const char *, ...)
1882 __attribute__ ((format (printf, 5, 6)));
1883 extern void ext4_error_file(struct file *, const char *, unsigned int,
1884 ext4_fsblk_t, const char *, ...)
1885 __attribute__ ((format (printf, 5, 6)));
1886 extern void __ext4_std_error(struct super_block *, const char *,
1887 unsigned int, int);
1888 extern void __ext4_abort(struct super_block *, const char *, unsigned int,
1889 const char *, ...)
1890 __attribute__ ((format (printf, 4, 5)));
1891 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1892 __LINE__, ## message)
1893 extern void __ext4_warning(struct super_block *, const char *, unsigned int,
1894 const char *, ...)
1895 __attribute__ ((format (printf, 4, 5)));
1896 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1897 __LINE__, ## message)
1898 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1899 __attribute__ ((format (printf, 3, 4)));
1900 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
1901 const char *, unsigned int, const char *);
1902 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
1903 __LINE__, msg)
1904 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1905 struct super_block *, ext4_group_t, \
1906 unsigned long, ext4_fsblk_t, \
1907 const char *, ...)
1908 __attribute__ ((format (printf, 7, 8)));
1909 #define ext4_grp_locked_error(sb, grp, message...) \
1910 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1911 extern void ext4_update_dynamic_rev(struct super_block *sb);
1912 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1913 __u32 compat);
1914 extern int ext4_update_rocompat_feature(handle_t *handle,
1915 struct super_block *sb, __u32 rocompat);
1916 extern int ext4_update_incompat_feature(handle_t *handle,
1917 struct super_block *sb, __u32 incompat);
1918 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1919 struct ext4_group_desc *bg);
1920 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1921 struct ext4_group_desc *bg);
1922 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1923 struct ext4_group_desc *bg);
1924 extern __u32 ext4_free_blks_count(struct super_block *sb,
1925 struct ext4_group_desc *bg);
1926 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1927 struct ext4_group_desc *bg);
1928 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1929 struct ext4_group_desc *bg);
1930 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1931 struct ext4_group_desc *bg);
1932 extern void ext4_block_bitmap_set(struct super_block *sb,
1933 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1934 extern void ext4_inode_bitmap_set(struct super_block *sb,
1935 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1936 extern void ext4_inode_table_set(struct super_block *sb,
1937 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1938 extern void ext4_free_blks_set(struct super_block *sb,
1939 struct ext4_group_desc *bg, __u32 count);
1940 extern void ext4_free_inodes_set(struct super_block *sb,
1941 struct ext4_group_desc *bg, __u32 count);
1942 extern void ext4_used_dirs_set(struct super_block *sb,
1943 struct ext4_group_desc *bg, __u32 count);
1944 extern void ext4_itable_unused_set(struct super_block *sb,
1945 struct ext4_group_desc *bg, __u32 count);
1946 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1947 struct ext4_group_desc *gdp);
1948 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1949 struct ext4_group_desc *gdp);
1950
1951 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1952 {
1953 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1954 le32_to_cpu(es->s_blocks_count_lo);
1955 }
1956
1957 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1958 {
1959 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1960 le32_to_cpu(es->s_r_blocks_count_lo);
1961 }
1962
1963 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1964 {
1965 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1966 le32_to_cpu(es->s_free_blocks_count_lo);
1967 }
1968
1969 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1970 ext4_fsblk_t blk)
1971 {
1972 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1973 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1974 }
1975
1976 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1977 ext4_fsblk_t blk)
1978 {
1979 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1980 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1981 }
1982
1983 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1984 ext4_fsblk_t blk)
1985 {
1986 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1987 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1988 }
1989
1990 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1991 {
1992 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1993 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1994 le32_to_cpu(raw_inode->i_size_lo);
1995 else
1996 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1997 }
1998
1999 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2000 {
2001 raw_inode->i_size_lo = cpu_to_le32(i_size);
2002 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2003 }
2004
2005 static inline
2006 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2007 ext4_group_t group)
2008 {
2009 struct ext4_group_info ***grp_info;
2010 long indexv, indexh;
2011 grp_info = EXT4_SB(sb)->s_group_info;
2012 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2013 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2014 return grp_info[indexv][indexh];
2015 }
2016
2017 /*
2018 * Reading s_groups_count requires using smp_rmb() afterwards. See
2019 * the locking protocol documented in the comments of ext4_group_add()
2020 * in resize.c
2021 */
2022 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2023 {
2024 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2025
2026 smp_rmb();
2027 return ngroups;
2028 }
2029
2030 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2031 ext4_group_t block_group)
2032 {
2033 return block_group >> sbi->s_log_groups_per_flex;
2034 }
2035
2036 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2037 {
2038 return 1 << sbi->s_log_groups_per_flex;
2039 }
2040
2041 #define ext4_std_error(sb, errno) \
2042 do { \
2043 if ((errno)) \
2044 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2045 } while (0)
2046
2047 #ifdef CONFIG_SMP
2048 /* Each CPU can accumulate percpu_counter_batch blocks in their local
2049 * counters. So we need to make sure we have free blocks more
2050 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2051 */
2052 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2053 #else
2054 #define EXT4_FREEBLOCKS_WATERMARK 0
2055 #endif
2056
2057 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2058 {
2059 /*
2060 * XXX: replace with spinlock if seen contended -bzzz
2061 */
2062 down_write(&EXT4_I(inode)->i_data_sem);
2063 if (newsize > EXT4_I(inode)->i_disksize)
2064 EXT4_I(inode)->i_disksize = newsize;
2065 up_write(&EXT4_I(inode)->i_data_sem);
2066 return ;
2067 }
2068
2069 struct ext4_group_info {
2070 unsigned long bb_state;
2071 struct rb_root bb_free_root;
2072 ext4_grpblk_t bb_first_free; /* first free block */
2073 ext4_grpblk_t bb_free; /* total free blocks */
2074 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2075 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2076 struct list_head bb_prealloc_list;
2077 #ifdef DOUBLE_CHECK
2078 void *bb_bitmap;
2079 #endif
2080 struct rw_semaphore alloc_sem;
2081 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2082 * regions, index is order.
2083 * bb_counters[3] = 5 means
2084 * 5 free 8-block regions. */
2085 };
2086
2087 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2088 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2089
2090 #define EXT4_MB_GRP_NEED_INIT(grp) \
2091 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2092
2093 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2094 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2095 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2096 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2097 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2098 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2099
2100 #define EXT4_MAX_CONTENTION 8
2101 #define EXT4_CONTENTION_THRESHOLD 2
2102
2103 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2104 ext4_group_t group)
2105 {
2106 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2107 }
2108
2109 /*
2110 * Returns true if the filesystem is busy enough that attempts to
2111 * access the block group locks has run into contention.
2112 */
2113 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2114 {
2115 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2116 }
2117
2118 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2119 {
2120 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2121 if (spin_trylock(lock))
2122 /*
2123 * We're able to grab the lock right away, so drop the
2124 * lock contention counter.
2125 */
2126 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2127 else {
2128 /*
2129 * The lock is busy, so bump the contention counter,
2130 * and then wait on the spin lock.
2131 */
2132 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2133 EXT4_MAX_CONTENTION);
2134 spin_lock(lock);
2135 }
2136 }
2137
2138 static inline void ext4_unlock_group(struct super_block *sb,
2139 ext4_group_t group)
2140 {
2141 spin_unlock(ext4_group_lock_ptr(sb, group));
2142 }
2143
2144 static inline void ext4_mark_super_dirty(struct super_block *sb)
2145 {
2146 if (EXT4_SB(sb)->s_journal == NULL)
2147 sb->s_dirt =1;
2148 }
2149
2150 /*
2151 * Block validity checking
2152 */
2153 #define ext4_check_indirect_blockref(inode, bh) \
2154 ext4_check_blockref(__func__, __LINE__, inode, \
2155 (__le32 *)(bh)->b_data, \
2156 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2157
2158 #define ext4_ind_check_inode(inode) \
2159 ext4_check_blockref(__func__, __LINE__, inode, \
2160 EXT4_I(inode)->i_data, \
2161 EXT4_NDIR_BLOCKS)
2162
2163 /*
2164 * Inodes and files operations
2165 */
2166
2167 /* dir.c */
2168 extern const struct file_operations ext4_dir_operations;
2169
2170 /* file.c */
2171 extern const struct inode_operations ext4_file_inode_operations;
2172 extern const struct file_operations ext4_file_operations;
2173 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2174
2175 /* namei.c */
2176 extern const struct inode_operations ext4_dir_inode_operations;
2177 extern const struct inode_operations ext4_special_inode_operations;
2178 extern struct dentry *ext4_get_parent(struct dentry *child);
2179
2180 /* symlink.c */
2181 extern const struct inode_operations ext4_symlink_inode_operations;
2182 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2183
2184 /* block_validity */
2185 extern void ext4_release_system_zone(struct super_block *sb);
2186 extern int ext4_setup_system_zone(struct super_block *sb);
2187 extern int __init ext4_init_system_zone(void);
2188 extern void ext4_exit_system_zone(void);
2189 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2190 ext4_fsblk_t start_blk,
2191 unsigned int count);
2192 extern int ext4_check_blockref(const char *, unsigned int,
2193 struct inode *, __le32 *, unsigned int);
2194
2195 /* extents.c */
2196 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2197 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2198 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2199 int chunk);
2200 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2201 struct ext4_map_blocks *map, int flags);
2202 extern void ext4_ext_truncate(struct inode *);
2203 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2204 loff_t length);
2205 extern void ext4_ext_init(struct super_block *);
2206 extern void ext4_ext_release(struct super_block *);
2207 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2208 loff_t len);
2209 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2210 ssize_t len);
2211 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2212 struct ext4_map_blocks *map, int flags);
2213 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2214 __u64 start, __u64 len);
2215 /* move_extent.c */
2216 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2217 __u64 start_orig, __u64 start_donor,
2218 __u64 len, __u64 *moved_len);
2219
2220 /* page-io.c */
2221 extern int __init ext4_init_pageio(void);
2222 extern void ext4_exit_pageio(void);
2223 extern void ext4_ioend_wait(struct inode *);
2224 extern void ext4_free_io_end(ext4_io_end_t *io);
2225 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2226 extern int ext4_end_io_nolock(ext4_io_end_t *io);
2227 extern void ext4_io_submit(struct ext4_io_submit *io);
2228 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2229 struct page *page,
2230 int len,
2231 struct writeback_control *wbc);
2232
2233 /* mmp.c */
2234 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2235
2236 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2237 enum ext4_state_bits {
2238 BH_Uninit /* blocks are allocated but uninitialized on disk */
2239 = BH_JBDPrivateStart,
2240 };
2241
2242 BUFFER_FNS(Uninit, uninit)
2243 TAS_BUFFER_FNS(Uninit, uninit)
2244
2245 /*
2246 * Add new method to test wether block and inode bitmaps are properly
2247 * initialized. With uninit_bg reading the block from disk is not enough
2248 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2249 */
2250 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2251
2252 static inline int bitmap_uptodate(struct buffer_head *bh)
2253 {
2254 return (buffer_uptodate(bh) &&
2255 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2256 }
2257 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2258 {
2259 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2260 }
2261
2262 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2263
2264 /* For ioend & aio unwritten conversion wait queues */
2265 #define EXT4_WQ_HASH_SZ 37
2266 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2267 EXT4_WQ_HASH_SZ])
2268 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2269 EXT4_WQ_HASH_SZ])
2270 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2271 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2272
2273 #endif /* __KERNEL__ */
2274
2275 #endif /* _EXT4_H */
linux-next