ext4: bigalloc changes to block bitmap initialization functions
[deliverable/linux.git] / fs / ext4 / ext4.h
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/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 #define EXT4_IO_END_QUEUED 0x0004
179
180 struct ext4_io_page {
181 struct page *p_page;
182 atomic_t p_count;
183 };
184
185 #define MAX_IO_PAGES 128
186
187 typedef struct ext4_io_end {
188 struct list_head list; /* per-file finished IO list */
189 struct inode *inode; /* file being written to */
190 unsigned int flag; /* unwritten or not */
191 struct page *page; /* page struct for buffer write */
192 loff_t offset; /* offset in the file */
193 ssize_t size; /* size of the extent */
194 struct work_struct work; /* data work queue */
195 struct kiocb *iocb; /* iocb struct for AIO */
196 int result; /* error value for AIO */
197 int num_io_pages;
198 struct ext4_io_page *pages[MAX_IO_PAGES];
199 } ext4_io_end_t;
200
201 struct ext4_io_submit {
202 int io_op;
203 struct bio *io_bio;
204 ext4_io_end_t *io_end;
205 struct ext4_io_page *io_page;
206 sector_t io_next_block;
207 };
208
209 /*
210 * Special inodes numbers
211 */
212 #define EXT4_BAD_INO 1 /* Bad blocks inode */
213 #define EXT4_ROOT_INO 2 /* Root inode */
214 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
218 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO 8 /* Journal inode */
220
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
223
224 /*
225 * Maximal count of links to a file
226 */
227 #define EXT4_LINK_MAX 65000
228
229 /*
230 * Macro-instructions used to manage several block sizes
231 */
232 #define EXT4_MIN_BLOCK_SIZE 1024
233 #define EXT4_MAX_BLOCK_SIZE 65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE 10
235 #define EXT4_MAX_BLOCK_LOG_SIZE 16
236 #ifdef __KERNEL__
237 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
238 #else
239 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
240 #endif
241 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
243 EXT4_SB(s)->s_cluster_bits)
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
247 #else
248 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
249 #endif
250 #ifdef __KERNEL__
251 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
254 #else
255 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 EXT4_GOOD_OLD_INODE_SIZE : \
257 (s)->s_inode_size)
258 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259 EXT4_GOOD_OLD_FIRST_INO : \
260 (s)->s_first_ino)
261 #endif
262 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
263
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270 (sbi)->s_cluster_bits)
271
272 /*
273 * Structure of a blocks group descriptor
274 */
275 struct ext4_group_desc
276 {
277 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
278 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
279 __le32 bg_inode_table_lo; /* Inodes table block */
280 __le16 bg_free_blocks_count_lo;/* Free blocks count */
281 __le16 bg_free_inodes_count_lo;/* Free inodes count */
282 __le16 bg_used_dirs_count_lo; /* Directories count */
283 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
284 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
285 __le16 bg_itable_unused_lo; /* Unused inodes count */
286 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
287 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
288 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
289 __le32 bg_inode_table_hi; /* Inodes table block MSB */
290 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
291 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
292 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
293 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
294 __u32 bg_reserved2[3];
295 };
296
297 /*
298 * Structure of a flex block group info
299 */
300
301 struct flex_groups {
302 atomic_t free_inodes;
303 atomic_t free_blocks;
304 atomic_t used_dirs;
305 };
306
307 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
308 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
309 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
310
311 /*
312 * Macro-instructions used to manage group descriptors
313 */
314 #define EXT4_MIN_DESC_SIZE 32
315 #define EXT4_MIN_DESC_SIZE_64BIT 64
316 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
317 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
318 #ifdef __KERNEL__
319 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
320 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
321 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
322 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
323 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
324 #else
325 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
326 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
327 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
328 #endif
329
330 /*
331 * Constants relative to the data blocks
332 */
333 #define EXT4_NDIR_BLOCKS 12
334 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
335 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
336 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
337 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
338
339 /*
340 * Inode flags
341 */
342 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
343 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
344 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
345 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
346 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
347 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
348 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
349 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
350 /* Reserved for compression usage... */
351 #define EXT4_DIRTY_FL 0x00000100
352 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
353 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
354 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
355 /* End compression flags --- maybe not all used */
356 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
357 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
358 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
359 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
360 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
361 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
362 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
363 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
364 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
365 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
366 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
367
368 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
369 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
370
371 /* Flags that should be inherited by new inodes from their parent. */
372 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
373 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
374 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
375 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
376
377 /* Flags that are appropriate for regular files (all but dir-specific ones). */
378 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
379
380 /* Flags that are appropriate for non-directories/regular files. */
381 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
382
383 /* Mask out flags that are inappropriate for the given type of inode. */
384 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
385 {
386 if (S_ISDIR(mode))
387 return flags;
388 else if (S_ISREG(mode))
389 return flags & EXT4_REG_FLMASK;
390 else
391 return flags & EXT4_OTHER_FLMASK;
392 }
393
394 /*
395 * Inode flags used for atomic set/get
396 */
397 enum {
398 EXT4_INODE_SECRM = 0, /* Secure deletion */
399 EXT4_INODE_UNRM = 1, /* Undelete */
400 EXT4_INODE_COMPR = 2, /* Compress file */
401 EXT4_INODE_SYNC = 3, /* Synchronous updates */
402 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
403 EXT4_INODE_APPEND = 5, /* writes to file may only append */
404 EXT4_INODE_NODUMP = 6, /* do not dump file */
405 EXT4_INODE_NOATIME = 7, /* do not update atime */
406 /* Reserved for compression usage... */
407 EXT4_INODE_DIRTY = 8,
408 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
409 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
410 EXT4_INODE_ECOMPR = 11, /* Compression error */
411 /* End compression flags --- maybe not all used */
412 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
413 EXT4_INODE_IMAGIC = 13, /* AFS directory */
414 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
415 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
416 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
417 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
418 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
419 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
420 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
421 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
422 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
423 };
424
425 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
426 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
427 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
428 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
429
430 /*
431 * Since it's pretty easy to mix up bit numbers and hex values, and we
432 * can't do a compile-time test for ENUM values, we use a run-time
433 * test to make sure that EXT4_XXX_FL is consistent with respect to
434 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
435 * out so it won't cost any extra space in the compiled kernel image.
436 * But it's important that these values are the same, since we are
437 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
438 * must be consistent with the values of FS_XXX_FL defined in
439 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
440 * ext4 filesystems, and of course the values defined in e2fsprogs.
441 *
442 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
443 */
444 static inline void ext4_check_flag_values(void)
445 {
446 CHECK_FLAG_VALUE(SECRM);
447 CHECK_FLAG_VALUE(UNRM);
448 CHECK_FLAG_VALUE(COMPR);
449 CHECK_FLAG_VALUE(SYNC);
450 CHECK_FLAG_VALUE(IMMUTABLE);
451 CHECK_FLAG_VALUE(APPEND);
452 CHECK_FLAG_VALUE(NODUMP);
453 CHECK_FLAG_VALUE(NOATIME);
454 CHECK_FLAG_VALUE(DIRTY);
455 CHECK_FLAG_VALUE(COMPRBLK);
456 CHECK_FLAG_VALUE(NOCOMPR);
457 CHECK_FLAG_VALUE(ECOMPR);
458 CHECK_FLAG_VALUE(INDEX);
459 CHECK_FLAG_VALUE(IMAGIC);
460 CHECK_FLAG_VALUE(JOURNAL_DATA);
461 CHECK_FLAG_VALUE(NOTAIL);
462 CHECK_FLAG_VALUE(DIRSYNC);
463 CHECK_FLAG_VALUE(TOPDIR);
464 CHECK_FLAG_VALUE(HUGE_FILE);
465 CHECK_FLAG_VALUE(EXTENTS);
466 CHECK_FLAG_VALUE(EA_INODE);
467 CHECK_FLAG_VALUE(EOFBLOCKS);
468 CHECK_FLAG_VALUE(RESERVED);
469 }
470
471 /* Used to pass group descriptor data when online resize is done */
472 struct ext4_new_group_input {
473 __u32 group; /* Group number for this data */
474 __u64 block_bitmap; /* Absolute block number of block bitmap */
475 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
476 __u64 inode_table; /* Absolute block number of inode table start */
477 __u32 blocks_count; /* Total number of blocks in this group */
478 __u16 reserved_blocks; /* Number of reserved blocks in this group */
479 __u16 unused;
480 };
481
482 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
483 struct compat_ext4_new_group_input {
484 u32 group;
485 compat_u64 block_bitmap;
486 compat_u64 inode_bitmap;
487 compat_u64 inode_table;
488 u32 blocks_count;
489 u16 reserved_blocks;
490 u16 unused;
491 };
492 #endif
493
494 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
495 struct ext4_new_group_data {
496 __u32 group;
497 __u64 block_bitmap;
498 __u64 inode_bitmap;
499 __u64 inode_table;
500 __u32 blocks_count;
501 __u16 reserved_blocks;
502 __u16 unused;
503 __u32 free_blocks_count;
504 };
505
506 /*
507 * Flags used by ext4_map_blocks()
508 */
509 /* Allocate any needed blocks and/or convert an unitialized
510 extent to be an initialized ext4 */
511 #define EXT4_GET_BLOCKS_CREATE 0x0001
512 /* Request the creation of an unitialized extent */
513 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
514 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
515 EXT4_GET_BLOCKS_CREATE)
516 /* Caller is from the delayed allocation writeout path,
517 so set the magic i_delalloc_reserve_flag after taking the
518 inode allocation semaphore for */
519 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
520 /* caller is from the direct IO path, request to creation of an
521 unitialized extents if not allocated, split the uninitialized
522 extent if blocks has been preallocated already*/
523 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
524 #define EXT4_GET_BLOCKS_CONVERT 0x0010
525 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
526 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
527 /* Convert extent to initialized after IO complete */
528 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
529 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
530 /* Punch out blocks of an extent */
531 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
532 /* Don't normalize allocation size (used for fallocate) */
533 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
534
535 /*
536 * Flags used by ext4_free_blocks
537 */
538 #define EXT4_FREE_BLOCKS_METADATA 0x0001
539 #define EXT4_FREE_BLOCKS_FORGET 0x0002
540 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
541 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
542
543 /*
544 * Flags used by ext4_discard_partial_page_buffers
545 */
546 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
547
548 /*
549 * ioctl commands
550 */
551 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
552 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
553 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
554 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
555 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
556 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
557 #ifdef CONFIG_JBD2_DEBUG
558 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
559 #endif
560 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
561 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
562 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
563 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
564 #define EXT4_IOC_MIGRATE _IO('f', 9)
565 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
566 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
567 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
568 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
569
570 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
571 /*
572 * ioctl commands in 32 bit emulation
573 */
574 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
575 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
576 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
577 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
578 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
579 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
580 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
581 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
582 #ifdef CONFIG_JBD2_DEBUG
583 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
584 #endif
585 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
586 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
587 #endif
588
589 /* Max physical block we can address w/o extents */
590 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
591
592 /*
593 * Structure of an inode on the disk
594 */
595 struct ext4_inode {
596 __le16 i_mode; /* File mode */
597 __le16 i_uid; /* Low 16 bits of Owner Uid */
598 __le32 i_size_lo; /* Size in bytes */
599 __le32 i_atime; /* Access time */
600 __le32 i_ctime; /* Inode Change time */
601 __le32 i_mtime; /* Modification time */
602 __le32 i_dtime; /* Deletion Time */
603 __le16 i_gid; /* Low 16 bits of Group Id */
604 __le16 i_links_count; /* Links count */
605 __le32 i_blocks_lo; /* Blocks count */
606 __le32 i_flags; /* File flags */
607 union {
608 struct {
609 __le32 l_i_version;
610 } linux1;
611 struct {
612 __u32 h_i_translator;
613 } hurd1;
614 struct {
615 __u32 m_i_reserved1;
616 } masix1;
617 } osd1; /* OS dependent 1 */
618 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
619 __le32 i_generation; /* File version (for NFS) */
620 __le32 i_file_acl_lo; /* File ACL */
621 __le32 i_size_high;
622 __le32 i_obso_faddr; /* Obsoleted fragment address */
623 union {
624 struct {
625 __le16 l_i_blocks_high; /* were l_i_reserved1 */
626 __le16 l_i_file_acl_high;
627 __le16 l_i_uid_high; /* these 2 fields */
628 __le16 l_i_gid_high; /* were reserved2[0] */
629 __u32 l_i_reserved2;
630 } linux2;
631 struct {
632 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
633 __u16 h_i_mode_high;
634 __u16 h_i_uid_high;
635 __u16 h_i_gid_high;
636 __u32 h_i_author;
637 } hurd2;
638 struct {
639 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
640 __le16 m_i_file_acl_high;
641 __u32 m_i_reserved2[2];
642 } masix2;
643 } osd2; /* OS dependent 2 */
644 __le16 i_extra_isize;
645 __le16 i_pad1;
646 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
647 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
648 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
649 __le32 i_crtime; /* File Creation time */
650 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
651 __le32 i_version_hi; /* high 32 bits for 64-bit version */
652 };
653
654 struct move_extent {
655 __u32 reserved; /* should be zero */
656 __u32 donor_fd; /* donor file descriptor */
657 __u64 orig_start; /* logical start offset in block for orig */
658 __u64 donor_start; /* logical start offset in block for donor */
659 __u64 len; /* block length to be moved */
660 __u64 moved_len; /* moved block length */
661 };
662
663 #define EXT4_EPOCH_BITS 2
664 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
665 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
666
667 /*
668 * Extended fields will fit into an inode if the filesystem was formatted
669 * with large inodes (-I 256 or larger) and there are not currently any EAs
670 * consuming all of the available space. For new inodes we always reserve
671 * enough space for the kernel's known extended fields, but for inodes
672 * created with an old kernel this might not have been the case. None of
673 * the extended inode fields is critical for correct filesystem operation.
674 * This macro checks if a certain field fits in the inode. Note that
675 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
676 */
677 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
678 ((offsetof(typeof(*ext4_inode), field) + \
679 sizeof((ext4_inode)->field)) \
680 <= (EXT4_GOOD_OLD_INODE_SIZE + \
681 (einode)->i_extra_isize)) \
682
683 static inline __le32 ext4_encode_extra_time(struct timespec *time)
684 {
685 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
686 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
687 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
688 }
689
690 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
691 {
692 if (sizeof(time->tv_sec) > 4)
693 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
694 << 32;
695 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
696 }
697
698 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
699 do { \
700 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
701 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
702 (raw_inode)->xtime ## _extra = \
703 ext4_encode_extra_time(&(inode)->xtime); \
704 } while (0)
705
706 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
707 do { \
708 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
709 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
710 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
711 (raw_inode)->xtime ## _extra = \
712 ext4_encode_extra_time(&(einode)->xtime); \
713 } while (0)
714
715 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
716 do { \
717 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
718 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
719 ext4_decode_extra_time(&(inode)->xtime, \
720 raw_inode->xtime ## _extra); \
721 else \
722 (inode)->xtime.tv_nsec = 0; \
723 } while (0)
724
725 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
726 do { \
727 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
728 (einode)->xtime.tv_sec = \
729 (signed)le32_to_cpu((raw_inode)->xtime); \
730 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
731 ext4_decode_extra_time(&(einode)->xtime, \
732 raw_inode->xtime ## _extra); \
733 else \
734 (einode)->xtime.tv_nsec = 0; \
735 } while (0)
736
737 #define i_disk_version osd1.linux1.l_i_version
738
739 #if defined(__KERNEL__) || defined(__linux__)
740 #define i_reserved1 osd1.linux1.l_i_reserved1
741 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
742 #define i_blocks_high osd2.linux2.l_i_blocks_high
743 #define i_uid_low i_uid
744 #define i_gid_low i_gid
745 #define i_uid_high osd2.linux2.l_i_uid_high
746 #define i_gid_high osd2.linux2.l_i_gid_high
747 #define i_reserved2 osd2.linux2.l_i_reserved2
748
749 #elif defined(__GNU__)
750
751 #define i_translator osd1.hurd1.h_i_translator
752 #define i_uid_high osd2.hurd2.h_i_uid_high
753 #define i_gid_high osd2.hurd2.h_i_gid_high
754 #define i_author osd2.hurd2.h_i_author
755
756 #elif defined(__masix__)
757
758 #define i_reserved1 osd1.masix1.m_i_reserved1
759 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
760 #define i_reserved2 osd2.masix2.m_i_reserved2
761
762 #endif /* defined(__KERNEL__) || defined(__linux__) */
763
764 /*
765 * storage for cached extent
766 * If ec_len == 0, then the cache is invalid.
767 * If ec_start == 0, then the cache represents a gap (null mapping)
768 */
769 struct ext4_ext_cache {
770 ext4_fsblk_t ec_start;
771 ext4_lblk_t ec_block;
772 __u32 ec_len; /* must be 32bit to return holes */
773 };
774
775 /*
776 * fourth extended file system inode data in memory
777 */
778 struct ext4_inode_info {
779 __le32 i_data[15]; /* unconverted */
780 __u32 i_dtime;
781 ext4_fsblk_t i_file_acl;
782
783 /*
784 * i_block_group is the number of the block group which contains
785 * this file's inode. Constant across the lifetime of the inode,
786 * it is ued for making block allocation decisions - we try to
787 * place a file's data blocks near its inode block, and new inodes
788 * near to their parent directory's inode.
789 */
790 ext4_group_t i_block_group;
791 ext4_lblk_t i_dir_start_lookup;
792 #if (BITS_PER_LONG < 64)
793 unsigned long i_state_flags; /* Dynamic state flags */
794 #endif
795 unsigned long i_flags;
796
797 #ifdef CONFIG_EXT4_FS_XATTR
798 /*
799 * Extended attributes can be read independently of the main file
800 * data. Taking i_mutex even when reading would cause contention
801 * between readers of EAs and writers of regular file data, so
802 * instead we synchronize on xattr_sem when reading or changing
803 * EAs.
804 */
805 struct rw_semaphore xattr_sem;
806 #endif
807
808 struct list_head i_orphan; /* unlinked but open inodes */
809
810 /*
811 * i_disksize keeps track of what the inode size is ON DISK, not
812 * in memory. During truncate, i_size is set to the new size by
813 * the VFS prior to calling ext4_truncate(), but the filesystem won't
814 * set i_disksize to 0 until the truncate is actually under way.
815 *
816 * The intent is that i_disksize always represents the blocks which
817 * are used by this file. This allows recovery to restart truncate
818 * on orphans if we crash during truncate. We actually write i_disksize
819 * into the on-disk inode when writing inodes out, instead of i_size.
820 *
821 * The only time when i_disksize and i_size may be different is when
822 * a truncate is in progress. The only things which change i_disksize
823 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
824 */
825 loff_t i_disksize;
826
827 /*
828 * i_data_sem is for serialising ext4_truncate() against
829 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
830 * data tree are chopped off during truncate. We can't do that in
831 * ext4 because whenever we perform intermediate commits during
832 * truncate, the inode and all the metadata blocks *must* be in a
833 * consistent state which allows truncation of the orphans to restart
834 * during recovery. Hence we must fix the get_block-vs-truncate race
835 * by other means, so we have i_data_sem.
836 */
837 struct rw_semaphore i_data_sem;
838 struct inode vfs_inode;
839 struct jbd2_inode *jinode;
840
841 struct ext4_ext_cache i_cached_extent;
842 /*
843 * File creation time. Its function is same as that of
844 * struct timespec i_{a,c,m}time in the generic inode.
845 */
846 struct timespec i_crtime;
847
848 /* mballoc */
849 struct list_head i_prealloc_list;
850 spinlock_t i_prealloc_lock;
851
852 /* ialloc */
853 ext4_group_t i_last_alloc_group;
854
855 /* allocation reservation info for delalloc */
856 unsigned int i_reserved_data_blocks;
857 unsigned int i_reserved_meta_blocks;
858 unsigned int i_allocated_meta_blocks;
859 ext4_lblk_t i_da_metadata_calc_last_lblock;
860 int i_da_metadata_calc_len;
861
862 /* on-disk additional length */
863 __u16 i_extra_isize;
864
865 #ifdef CONFIG_QUOTA
866 /* quota space reservation, managed internally by quota code */
867 qsize_t i_reserved_quota;
868 #endif
869
870 /* completed IOs that might need unwritten extents handling */
871 struct list_head i_completed_io_list;
872 spinlock_t i_completed_io_lock;
873 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
874 /* current io_end structure for async DIO write*/
875 ext4_io_end_t *cur_aio_dio;
876 atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
877
878 spinlock_t i_block_reservation_lock;
879
880 /*
881 * Transactions that contain inode's metadata needed to complete
882 * fsync and fdatasync, respectively.
883 */
884 tid_t i_sync_tid;
885 tid_t i_datasync_tid;
886 };
887
888 /*
889 * File system states
890 */
891 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
892 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
893 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
894
895 /*
896 * Misc. filesystem flags
897 */
898 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
899 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
900 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
901
902 /*
903 * Mount flags
904 */
905 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
906 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
907 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
908 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
909 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
910 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
911 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
912 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
913 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
914 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
915 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
916 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
917 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
918 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
919 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
920 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
921 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
922 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
923 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
924 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
925 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
926 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
927 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
928 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
929 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
930 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
931 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
932 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
933 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
934 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
935 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
936
937 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
938 specified delalloc */
939
940 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
941 ~EXT4_MOUNT_##opt
942 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
943 EXT4_MOUNT_##opt
944 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
945 EXT4_MOUNT_##opt)
946
947 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
948 ~EXT4_MOUNT2_##opt
949 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
950 EXT4_MOUNT2_##opt
951 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
952 EXT4_MOUNT2_##opt)
953
954 #define ext4_set_bit __test_and_set_bit_le
955 #define ext4_set_bit_atomic ext2_set_bit_atomic
956 #define ext4_clear_bit __test_and_clear_bit_le
957 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
958 #define ext4_test_bit test_bit_le
959 #define ext4_find_first_zero_bit find_first_zero_bit_le
960 #define ext4_find_next_zero_bit find_next_zero_bit_le
961 #define ext4_find_next_bit find_next_bit_le
962
963 extern void ext4_set_bits(void *bm, int cur, int len);
964
965 /*
966 * Maximal mount counts between two filesystem checks
967 */
968 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
969 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
970
971 /*
972 * Behaviour when detecting errors
973 */
974 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
975 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
976 #define EXT4_ERRORS_PANIC 3 /* Panic */
977 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
978
979 /*
980 * Structure of the super block
981 */
982 struct ext4_super_block {
983 /*00*/ __le32 s_inodes_count; /* Inodes count */
984 __le32 s_blocks_count_lo; /* Blocks count */
985 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
986 __le32 s_free_blocks_count_lo; /* Free blocks count */
987 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
988 __le32 s_first_data_block; /* First Data Block */
989 __le32 s_log_block_size; /* Block size */
990 __le32 s_log_cluster_size; /* Allocation cluster size */
991 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
992 __le32 s_clusters_per_group; /* # Clusters per group */
993 __le32 s_inodes_per_group; /* # Inodes per group */
994 __le32 s_mtime; /* Mount time */
995 /*30*/ __le32 s_wtime; /* Write time */
996 __le16 s_mnt_count; /* Mount count */
997 __le16 s_max_mnt_count; /* Maximal mount count */
998 __le16 s_magic; /* Magic signature */
999 __le16 s_state; /* File system state */
1000 __le16 s_errors; /* Behaviour when detecting errors */
1001 __le16 s_minor_rev_level; /* minor revision level */
1002 /*40*/ __le32 s_lastcheck; /* time of last check */
1003 __le32 s_checkinterval; /* max. time between checks */
1004 __le32 s_creator_os; /* OS */
1005 __le32 s_rev_level; /* Revision level */
1006 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1007 __le16 s_def_resgid; /* Default gid for reserved blocks */
1008 /*
1009 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1010 *
1011 * Note: the difference between the compatible feature set and
1012 * the incompatible feature set is that if there is a bit set
1013 * in the incompatible feature set that the kernel doesn't
1014 * know about, it should refuse to mount the filesystem.
1015 *
1016 * e2fsck's requirements are more strict; if it doesn't know
1017 * about a feature in either the compatible or incompatible
1018 * feature set, it must abort and not try to meddle with
1019 * things it doesn't understand...
1020 */
1021 __le32 s_first_ino; /* First non-reserved inode */
1022 __le16 s_inode_size; /* size of inode structure */
1023 __le16 s_block_group_nr; /* block group # of this superblock */
1024 __le32 s_feature_compat; /* compatible feature set */
1025 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1026 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1027 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1028 /*78*/ char s_volume_name[16]; /* volume name */
1029 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1030 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1031 /*
1032 * Performance hints. Directory preallocation should only
1033 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1034 */
1035 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1036 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1037 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1038 /*
1039 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1040 */
1041 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1042 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1043 __le32 s_journal_dev; /* device number of journal file */
1044 __le32 s_last_orphan; /* start of list of inodes to delete */
1045 __le32 s_hash_seed[4]; /* HTREE hash seed */
1046 __u8 s_def_hash_version; /* Default hash version to use */
1047 __u8 s_jnl_backup_type;
1048 __le16 s_desc_size; /* size of group descriptor */
1049 /*100*/ __le32 s_default_mount_opts;
1050 __le32 s_first_meta_bg; /* First metablock block group */
1051 __le32 s_mkfs_time; /* When the filesystem was created */
1052 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1053 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1054 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1055 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1056 __le32 s_free_blocks_count_hi; /* Free blocks count */
1057 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1058 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1059 __le32 s_flags; /* Miscellaneous flags */
1060 __le16 s_raid_stride; /* RAID stride */
1061 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1062 __le64 s_mmp_block; /* Block for multi-mount protection */
1063 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1064 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1065 __u8 s_reserved_char_pad;
1066 __le16 s_reserved_pad;
1067 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1068 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1069 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1070 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1071 snapshot's future use */
1072 __le32 s_snapshot_list; /* inode number of the head of the
1073 on-disk snapshot list */
1074 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1075 __le32 s_error_count; /* number of fs errors */
1076 __le32 s_first_error_time; /* first time an error happened */
1077 __le32 s_first_error_ino; /* inode involved in first error */
1078 __le64 s_first_error_block; /* block involved of first error */
1079 __u8 s_first_error_func[32]; /* function where the error happened */
1080 __le32 s_first_error_line; /* line number where error happened */
1081 __le32 s_last_error_time; /* most recent time of an error */
1082 __le32 s_last_error_ino; /* inode involved in last error */
1083 __le32 s_last_error_line; /* line number where error happened */
1084 __le64 s_last_error_block; /* block involved of last error */
1085 __u8 s_last_error_func[32]; /* function where the error happened */
1086 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1087 __u8 s_mount_opts[64];
1088 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1089 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1090 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1091 __le32 s_reserved[109]; /* Padding to the end of the block */
1092 };
1093
1094 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1095
1096 #ifdef __KERNEL__
1097
1098 /*
1099 * run-time mount flags
1100 */
1101 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1102 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1103
1104 /*
1105 * fourth extended-fs super-block data in memory
1106 */
1107 struct ext4_sb_info {
1108 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1109 unsigned long s_inodes_per_block;/* Number of inodes per block */
1110 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1111 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1112 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1113 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1114 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1115 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1116 ext4_group_t s_groups_count; /* Number of groups in the fs */
1117 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1118 unsigned long s_overhead_last; /* Last calculated overhead */
1119 unsigned long s_blocks_last; /* Last seen block count */
1120 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1121 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1122 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1123 struct buffer_head * s_sbh; /* Buffer containing the super block */
1124 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1125 struct buffer_head **s_group_desc;
1126 unsigned int s_mount_opt;
1127 unsigned int s_mount_opt2;
1128 unsigned int s_mount_flags;
1129 ext4_fsblk_t s_sb_block;
1130 uid_t s_resuid;
1131 gid_t s_resgid;
1132 unsigned short s_mount_state;
1133 unsigned short s_pad;
1134 int s_addr_per_block_bits;
1135 int s_desc_per_block_bits;
1136 int s_inode_size;
1137 int s_first_ino;
1138 unsigned int s_inode_readahead_blks;
1139 unsigned int s_inode_goal;
1140 spinlock_t s_next_gen_lock;
1141 u32 s_next_generation;
1142 u32 s_hash_seed[4];
1143 int s_def_hash_version;
1144 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1145 struct percpu_counter s_freeblocks_counter;
1146 struct percpu_counter s_freeinodes_counter;
1147 struct percpu_counter s_dirs_counter;
1148 struct percpu_counter s_dirtyblocks_counter;
1149 struct blockgroup_lock *s_blockgroup_lock;
1150 struct proc_dir_entry *s_proc;
1151 struct kobject s_kobj;
1152 struct completion s_kobj_unregister;
1153
1154 /* Journaling */
1155 struct journal_s *s_journal;
1156 struct list_head s_orphan;
1157 struct mutex s_orphan_lock;
1158 unsigned long s_resize_flags; /* Flags indicating if there
1159 is a resizer */
1160 unsigned long s_commit_interval;
1161 u32 s_max_batch_time;
1162 u32 s_min_batch_time;
1163 struct block_device *journal_bdev;
1164 #ifdef CONFIG_JBD2_DEBUG
1165 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
1166 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
1167 #endif
1168 #ifdef CONFIG_QUOTA
1169 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1170 int s_jquota_fmt; /* Format of quota to use */
1171 #endif
1172 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1173 struct rb_root system_blks;
1174
1175 #ifdef EXTENTS_STATS
1176 /* ext4 extents stats */
1177 unsigned long s_ext_min;
1178 unsigned long s_ext_max;
1179 unsigned long s_depth_max;
1180 spinlock_t s_ext_stats_lock;
1181 unsigned long s_ext_blocks;
1182 unsigned long s_ext_extents;
1183 #endif
1184 /* ext4 extent cache stats */
1185 unsigned long extent_cache_hits;
1186 unsigned long extent_cache_misses;
1187
1188 /* for buddy allocator */
1189 struct ext4_group_info ***s_group_info;
1190 struct inode *s_buddy_cache;
1191 spinlock_t s_md_lock;
1192 unsigned short *s_mb_offsets;
1193 unsigned int *s_mb_maxs;
1194
1195 /* tunables */
1196 unsigned long s_stripe;
1197 unsigned int s_mb_stream_request;
1198 unsigned int s_mb_max_to_scan;
1199 unsigned int s_mb_min_to_scan;
1200 unsigned int s_mb_stats;
1201 unsigned int s_mb_order2_reqs;
1202 unsigned int s_mb_group_prealloc;
1203 unsigned int s_max_writeback_mb_bump;
1204 /* where last allocation was done - for stream allocation */
1205 unsigned long s_mb_last_group;
1206 unsigned long s_mb_last_start;
1207
1208 /* stats for buddy allocator */
1209 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1210 atomic_t s_bal_success; /* we found long enough chunks */
1211 atomic_t s_bal_allocated; /* in blocks */
1212 atomic_t s_bal_ex_scanned; /* total extents scanned */
1213 atomic_t s_bal_goals; /* goal hits */
1214 atomic_t s_bal_breaks; /* too long searches */
1215 atomic_t s_bal_2orders; /* 2^order hits */
1216 spinlock_t s_bal_lock;
1217 unsigned long s_mb_buddies_generated;
1218 unsigned long long s_mb_generation_time;
1219 atomic_t s_mb_lost_chunks;
1220 atomic_t s_mb_preallocated;
1221 atomic_t s_mb_discarded;
1222 atomic_t s_lock_busy;
1223
1224 /* locality groups */
1225 struct ext4_locality_group __percpu *s_locality_groups;
1226
1227 /* for write statistics */
1228 unsigned long s_sectors_written_start;
1229 u64 s_kbytes_written;
1230
1231 unsigned int s_log_groups_per_flex;
1232 struct flex_groups *s_flex_groups;
1233
1234 /* workqueue for dio unwritten */
1235 struct workqueue_struct *dio_unwritten_wq;
1236
1237 /* timer for periodic error stats printing */
1238 struct timer_list s_err_report;
1239
1240 /* Lazy inode table initialization info */
1241 struct ext4_li_request *s_li_request;
1242 /* Wait multiplier for lazy initialization thread */
1243 unsigned int s_li_wait_mult;
1244
1245 /* Kernel thread for multiple mount protection */
1246 struct task_struct *s_mmp_tsk;
1247
1248 /* record the last minlen when FITRIM is called. */
1249 atomic_t s_last_trim_minblks;
1250 };
1251
1252 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1253 {
1254 return sb->s_fs_info;
1255 }
1256 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1257 {
1258 return container_of(inode, struct ext4_inode_info, vfs_inode);
1259 }
1260
1261 static inline struct timespec ext4_current_time(struct inode *inode)
1262 {
1263 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1264 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1265 }
1266
1267 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1268 {
1269 return ino == EXT4_ROOT_INO ||
1270 ino == EXT4_JOURNAL_INO ||
1271 ino == EXT4_RESIZE_INO ||
1272 (ino >= EXT4_FIRST_INO(sb) &&
1273 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1274 }
1275
1276 /*
1277 * Inode dynamic state flags
1278 */
1279 enum {
1280 EXT4_STATE_JDATA, /* journaled data exists */
1281 EXT4_STATE_NEW, /* inode is newly created */
1282 EXT4_STATE_XATTR, /* has in-inode xattrs */
1283 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1284 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1285 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1286 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1287 EXT4_STATE_NEWENTRY, /* File just added to dir */
1288 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1289 };
1290
1291 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1292 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1293 { \
1294 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1295 } \
1296 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1297 { \
1298 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1299 } \
1300 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1301 { \
1302 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1303 }
1304
1305 EXT4_INODE_BIT_FNS(flag, flags, 0)
1306 #if (BITS_PER_LONG < 64)
1307 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1308
1309 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1310 {
1311 (ei)->i_state_flags = 0;
1312 }
1313 #else
1314 EXT4_INODE_BIT_FNS(state, flags, 32)
1315
1316 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1317 {
1318 /* We depend on the fact that callers will set i_flags */
1319 }
1320 #endif
1321 #else
1322 /* Assume that user mode programs are passing in an ext4fs superblock, not
1323 * a kernel struct super_block. This will allow us to call the feature-test
1324 * macros from user land. */
1325 #define EXT4_SB(sb) (sb)
1326 #endif
1327
1328 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1329
1330 /*
1331 * Codes for operating systems
1332 */
1333 #define EXT4_OS_LINUX 0
1334 #define EXT4_OS_HURD 1
1335 #define EXT4_OS_MASIX 2
1336 #define EXT4_OS_FREEBSD 3
1337 #define EXT4_OS_LITES 4
1338
1339 /*
1340 * Revision levels
1341 */
1342 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1343 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1344
1345 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1346 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1347
1348 #define EXT4_GOOD_OLD_INODE_SIZE 128
1349
1350 /*
1351 * Feature set definitions
1352 */
1353
1354 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1355 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1356 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1357 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1358 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1359 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1360 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1361 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1362 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1363 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1364 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1365 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1366 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1367 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1368 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1369 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1370 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1371 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1372
1373 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1374 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1375 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1376 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1377 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1378 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1379
1380 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1381 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1382 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1383 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1384 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1385 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1386 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1387 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1388 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1389
1390 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1391 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1392 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1393 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1394 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1395 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1396 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1397 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1398 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1399 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1400 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1401
1402 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1403 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1404 EXT4_FEATURE_INCOMPAT_META_BG)
1405 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1406 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1407 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1408
1409 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1410 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1411 EXT4_FEATURE_INCOMPAT_RECOVER| \
1412 EXT4_FEATURE_INCOMPAT_META_BG)
1413 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1414 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1415 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1416
1417 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1418 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1419 EXT4_FEATURE_INCOMPAT_RECOVER| \
1420 EXT4_FEATURE_INCOMPAT_META_BG| \
1421 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1422 EXT4_FEATURE_INCOMPAT_64BIT| \
1423 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1424 EXT4_FEATURE_INCOMPAT_MMP)
1425 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1426 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1427 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1428 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1429 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1430 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1431 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1432
1433 /*
1434 * Default values for user and/or group using reserved blocks
1435 */
1436 #define EXT4_DEF_RESUID 0
1437 #define EXT4_DEF_RESGID 0
1438
1439 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1440
1441 /*
1442 * Default mount options
1443 */
1444 #define EXT4_DEFM_DEBUG 0x0001
1445 #define EXT4_DEFM_BSDGROUPS 0x0002
1446 #define EXT4_DEFM_XATTR_USER 0x0004
1447 #define EXT4_DEFM_ACL 0x0008
1448 #define EXT4_DEFM_UID16 0x0010
1449 #define EXT4_DEFM_JMODE 0x0060
1450 #define EXT4_DEFM_JMODE_DATA 0x0020
1451 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1452 #define EXT4_DEFM_JMODE_WBACK 0x0060
1453 #define EXT4_DEFM_NOBARRIER 0x0100
1454 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1455 #define EXT4_DEFM_DISCARD 0x0400
1456 #define EXT4_DEFM_NODELALLOC 0x0800
1457
1458 /*
1459 * Default journal batch times
1460 */
1461 #define EXT4_DEF_MIN_BATCH_TIME 0
1462 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1463
1464 /*
1465 * Minimum number of groups in a flexgroup before we separate out
1466 * directories into the first block group of a flexgroup
1467 */
1468 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1469
1470 /*
1471 * Structure of a directory entry
1472 */
1473 #define EXT4_NAME_LEN 255
1474
1475 struct ext4_dir_entry {
1476 __le32 inode; /* Inode number */
1477 __le16 rec_len; /* Directory entry length */
1478 __le16 name_len; /* Name length */
1479 char name[EXT4_NAME_LEN]; /* File name */
1480 };
1481
1482 /*
1483 * The new version of the directory entry. Since EXT4 structures are
1484 * stored in intel byte order, and the name_len field could never be
1485 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1486 * file_type field.
1487 */
1488 struct ext4_dir_entry_2 {
1489 __le32 inode; /* Inode number */
1490 __le16 rec_len; /* Directory entry length */
1491 __u8 name_len; /* Name length */
1492 __u8 file_type;
1493 char name[EXT4_NAME_LEN]; /* File name */
1494 };
1495
1496 /*
1497 * Ext4 directory file types. Only the low 3 bits are used. The
1498 * other bits are reserved for now.
1499 */
1500 #define EXT4_FT_UNKNOWN 0
1501 #define EXT4_FT_REG_FILE 1
1502 #define EXT4_FT_DIR 2
1503 #define EXT4_FT_CHRDEV 3
1504 #define EXT4_FT_BLKDEV 4
1505 #define EXT4_FT_FIFO 5
1506 #define EXT4_FT_SOCK 6
1507 #define EXT4_FT_SYMLINK 7
1508
1509 #define EXT4_FT_MAX 8
1510
1511 /*
1512 * EXT4_DIR_PAD defines the directory entries boundaries
1513 *
1514 * NOTE: It must be a multiple of 4
1515 */
1516 #define EXT4_DIR_PAD 4
1517 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1518 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1519 ~EXT4_DIR_ROUND)
1520 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1521
1522 /*
1523 * If we ever get support for fs block sizes > page_size, we'll need
1524 * to remove the #if statements in the next two functions...
1525 */
1526 static inline unsigned int
1527 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1528 {
1529 unsigned len = le16_to_cpu(dlen);
1530
1531 #if (PAGE_CACHE_SIZE >= 65536)
1532 if (len == EXT4_MAX_REC_LEN || len == 0)
1533 return blocksize;
1534 return (len & 65532) | ((len & 3) << 16);
1535 #else
1536 return len;
1537 #endif
1538 }
1539
1540 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1541 {
1542 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1543 BUG();
1544 #if (PAGE_CACHE_SIZE >= 65536)
1545 if (len < 65536)
1546 return cpu_to_le16(len);
1547 if (len == blocksize) {
1548 if (blocksize == 65536)
1549 return cpu_to_le16(EXT4_MAX_REC_LEN);
1550 else
1551 return cpu_to_le16(0);
1552 }
1553 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1554 #else
1555 return cpu_to_le16(len);
1556 #endif
1557 }
1558
1559 /*
1560 * Hash Tree Directory indexing
1561 * (c) Daniel Phillips, 2001
1562 */
1563
1564 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1565 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1566 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1567 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1568 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1569
1570 /* Legal values for the dx_root hash_version field: */
1571
1572 #define DX_HASH_LEGACY 0
1573 #define DX_HASH_HALF_MD4 1
1574 #define DX_HASH_TEA 2
1575 #define DX_HASH_LEGACY_UNSIGNED 3
1576 #define DX_HASH_HALF_MD4_UNSIGNED 4
1577 #define DX_HASH_TEA_UNSIGNED 5
1578
1579 #ifdef __KERNEL__
1580
1581 /* hash info structure used by the directory hash */
1582 struct dx_hash_info
1583 {
1584 u32 hash;
1585 u32 minor_hash;
1586 int hash_version;
1587 u32 *seed;
1588 };
1589
1590 #define EXT4_HTREE_EOF 0x7fffffff
1591
1592 /*
1593 * Control parameters used by ext4_htree_next_block
1594 */
1595 #define HASH_NB_ALWAYS 1
1596
1597
1598 /*
1599 * Describe an inode's exact location on disk and in memory
1600 */
1601 struct ext4_iloc
1602 {
1603 struct buffer_head *bh;
1604 unsigned long offset;
1605 ext4_group_t block_group;
1606 };
1607
1608 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1609 {
1610 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1611 }
1612
1613 /*
1614 * This structure is stuffed into the struct file's private_data field
1615 * for directories. It is where we put information so that we can do
1616 * readdir operations in hash tree order.
1617 */
1618 struct dir_private_info {
1619 struct rb_root root;
1620 struct rb_node *curr_node;
1621 struct fname *extra_fname;
1622 loff_t last_pos;
1623 __u32 curr_hash;
1624 __u32 curr_minor_hash;
1625 __u32 next_hash;
1626 };
1627
1628 /* calculate the first block number of the group */
1629 static inline ext4_fsblk_t
1630 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1631 {
1632 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1633 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1634 }
1635
1636 /*
1637 * Special error return code only used by dx_probe() and its callers.
1638 */
1639 #define ERR_BAD_DX_DIR -75000
1640
1641 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1642 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1643
1644 /*
1645 * Timeout and state flag for lazy initialization inode thread.
1646 */
1647 #define EXT4_DEF_LI_WAIT_MULT 10
1648 #define EXT4_DEF_LI_MAX_START_DELAY 5
1649 #define EXT4_LAZYINIT_QUIT 0x0001
1650 #define EXT4_LAZYINIT_RUNNING 0x0002
1651
1652 /*
1653 * Lazy inode table initialization info
1654 */
1655 struct ext4_lazy_init {
1656 unsigned long li_state;
1657 struct list_head li_request_list;
1658 struct mutex li_list_mtx;
1659 };
1660
1661 struct ext4_li_request {
1662 struct super_block *lr_super;
1663 struct ext4_sb_info *lr_sbi;
1664 ext4_group_t lr_next_group;
1665 struct list_head lr_request;
1666 unsigned long lr_next_sched;
1667 unsigned long lr_timeout;
1668 };
1669
1670 struct ext4_features {
1671 struct kobject f_kobj;
1672 struct completion f_kobj_unregister;
1673 };
1674
1675 /*
1676 * This structure will be used for multiple mount protection. It will be
1677 * written into the block number saved in the s_mmp_block field in the
1678 * superblock. Programs that check MMP should assume that if
1679 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1680 * to use the filesystem, regardless of how old the timestamp is.
1681 */
1682 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1683 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1684 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1685 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1686
1687 struct mmp_struct {
1688 __le32 mmp_magic; /* Magic number for MMP */
1689 __le32 mmp_seq; /* Sequence no. updated periodically */
1690
1691 /*
1692 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1693 * purposes and do not affect the correctness of the algorithm
1694 */
1695 __le64 mmp_time; /* Time last updated */
1696 char mmp_nodename[64]; /* Node which last updated MMP block */
1697 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1698
1699 /*
1700 * mmp_check_interval is used to verify if the MMP block has been
1701 * updated on the block device. The value is updated based on the
1702 * maximum time to write the MMP block during an update cycle.
1703 */
1704 __le16 mmp_check_interval;
1705
1706 __le16 mmp_pad1;
1707 __le32 mmp_pad2[227];
1708 };
1709
1710 /* arguments passed to the mmp thread */
1711 struct mmpd_data {
1712 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1713 struct super_block *sb; /* super block of the fs */
1714 };
1715
1716 /*
1717 * Check interval multiplier
1718 * The MMP block is written every update interval and initially checked every
1719 * update interval x the multiplier (the value is then adapted based on the
1720 * write latency). The reason is that writes can be delayed under load and we
1721 * don't want readers to incorrectly assume that the filesystem is no longer
1722 * in use.
1723 */
1724 #define EXT4_MMP_CHECK_MULT 2UL
1725
1726 /*
1727 * Minimum interval for MMP checking in seconds.
1728 */
1729 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1730
1731 /*
1732 * Maximum interval for MMP checking in seconds.
1733 */
1734 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1735
1736 /*
1737 * Function prototypes
1738 */
1739
1740 /*
1741 * Ok, these declarations are also in <linux/kernel.h> but none of the
1742 * ext4 source programs needs to include it so they are duplicated here.
1743 */
1744 # define NORET_TYPE /**/
1745 # define ATTRIB_NORET __attribute__((noreturn))
1746 # define NORET_AND noreturn,
1747
1748 /* bitmap.c */
1749 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1750
1751 /* balloc.c */
1752 extern unsigned int ext4_block_group(struct super_block *sb,
1753 ext4_fsblk_t blocknr);
1754 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1755 ext4_fsblk_t blocknr);
1756 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1757 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1758 ext4_group_t group);
1759 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1760 ext4_fsblk_t goal,
1761 unsigned int flags,
1762 unsigned long *count,
1763 int *errp);
1764 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi,
1765 s64 nblocks, unsigned int flags);
1766 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1767 extern void ext4_check_blocks_bitmap(struct super_block *);
1768 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1769 ext4_group_t block_group,
1770 struct buffer_head ** bh);
1771 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1772 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1773 ext4_group_t block_group);
1774 extern void ext4_init_block_bitmap(struct super_block *sb,
1775 struct buffer_head *bh,
1776 ext4_group_t group,
1777 struct ext4_group_desc *desc);
1778 extern unsigned ext4_free_blocks_after_init(struct super_block *sb,
1779 ext4_group_t block_group,
1780 struct ext4_group_desc *gdp);
1781 extern unsigned ext4_num_base_meta_clusters(struct super_block *sb,
1782 ext4_group_t block_group);
1783 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1784 ext4_group_t block_group,
1785 struct ext4_group_desc *gdp);
1786 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1787
1788 /* dir.c */
1789 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1790 struct file *,
1791 struct ext4_dir_entry_2 *,
1792 struct buffer_head *, unsigned int);
1793 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1794 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1795 (de), (bh), (offset)))
1796 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1797 __u32 minor_hash,
1798 struct ext4_dir_entry_2 *dirent);
1799 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1800
1801 /* fsync.c */
1802 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1803 extern int ext4_flush_completed_IO(struct inode *);
1804
1805 /* hash.c */
1806 extern int ext4fs_dirhash(const char *name, int len, struct
1807 dx_hash_info *hinfo);
1808
1809 /* ialloc.c */
1810 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1811 const struct qstr *qstr, __u32 goal);
1812 extern void ext4_free_inode(handle_t *, struct inode *);
1813 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1814 extern unsigned long ext4_count_free_inodes(struct super_block *);
1815 extern unsigned long ext4_count_dirs(struct super_block *);
1816 extern void ext4_check_inodes_bitmap(struct super_block *);
1817 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1818 extern int ext4_init_inode_table(struct super_block *sb,
1819 ext4_group_t group, int barrier);
1820
1821 /* mballoc.c */
1822 extern long ext4_mb_stats;
1823 extern long ext4_mb_max_to_scan;
1824 extern int ext4_mb_init(struct super_block *, int);
1825 extern int ext4_mb_release(struct super_block *);
1826 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1827 struct ext4_allocation_request *, int *);
1828 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1829 extern void ext4_discard_preallocations(struct inode *);
1830 extern int __init ext4_init_mballoc(void);
1831 extern void ext4_exit_mballoc(void);
1832 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1833 struct buffer_head *bh, ext4_fsblk_t block,
1834 unsigned long count, int flags);
1835 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1836 ext4_group_t i, struct ext4_group_desc *desc);
1837 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1838 ext4_fsblk_t block, unsigned long count);
1839 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1840
1841 /* inode.c */
1842 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1843 ext4_lblk_t, int, int *);
1844 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1845 ext4_lblk_t, int, int *);
1846 int ext4_get_block(struct inode *inode, sector_t iblock,
1847 struct buffer_head *bh_result, int create);
1848
1849 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1850 extern int ext4_write_inode(struct inode *, struct writeback_control *);
1851 extern int ext4_setattr(struct dentry *, struct iattr *);
1852 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1853 struct kstat *stat);
1854 extern void ext4_evict_inode(struct inode *);
1855 extern void ext4_clear_inode(struct inode *);
1856 extern int ext4_sync_inode(handle_t *, struct inode *);
1857 extern void ext4_dirty_inode(struct inode *, int);
1858 extern int ext4_change_inode_journal_flag(struct inode *, int);
1859 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1860 extern int ext4_can_truncate(struct inode *inode);
1861 extern void ext4_truncate(struct inode *);
1862 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1863 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1864 extern void ext4_set_inode_flags(struct inode *);
1865 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1866 extern int ext4_alloc_da_blocks(struct inode *inode);
1867 extern void ext4_set_aops(struct inode *inode);
1868 extern int ext4_writepage_trans_blocks(struct inode *);
1869 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1870 extern int ext4_block_truncate_page(handle_t *handle,
1871 struct address_space *mapping, loff_t from);
1872 extern int ext4_block_zero_page_range(handle_t *handle,
1873 struct address_space *mapping, loff_t from, loff_t length);
1874 extern int ext4_discard_partial_page_buffers(handle_t *handle,
1875 struct address_space *mapping, loff_t from,
1876 loff_t length, int flags);
1877 extern int ext4_discard_partial_page_buffers_no_lock(handle_t *handle,
1878 struct inode *inode, struct page *page, loff_t from,
1879 loff_t length, int flags);
1880 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1881 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1882 extern void ext4_da_update_reserve_space(struct inode *inode,
1883 int used, int quota_claim);
1884
1885 /* indirect.c */
1886 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
1887 struct ext4_map_blocks *map, int flags);
1888 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
1889 const struct iovec *iov, loff_t offset,
1890 unsigned long nr_segs);
1891 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
1892 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
1893 extern void ext4_ind_truncate(struct inode *inode);
1894
1895 /* ioctl.c */
1896 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1897 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1898
1899 /* migrate.c */
1900 extern int ext4_ext_migrate(struct inode *);
1901
1902 /* namei.c */
1903 extern int ext4_orphan_add(handle_t *, struct inode *);
1904 extern int ext4_orphan_del(handle_t *, struct inode *);
1905 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1906 __u32 start_minor_hash, __u32 *next_hash);
1907
1908 /* resize.c */
1909 extern int ext4_group_add(struct super_block *sb,
1910 struct ext4_new_group_data *input);
1911 extern int ext4_group_extend(struct super_block *sb,
1912 struct ext4_super_block *es,
1913 ext4_fsblk_t n_blocks_count);
1914
1915 /* super.c */
1916 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
1917 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
1918 extern void ext4_kvfree(void *ptr);
1919 extern void __ext4_error(struct super_block *, const char *, unsigned int,
1920 const char *, ...)
1921 __attribute__ ((format (printf, 4, 5)));
1922 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1923 __LINE__, ## message)
1924 extern void ext4_error_inode(struct inode *, const char *, unsigned int,
1925 ext4_fsblk_t, const char *, ...)
1926 __attribute__ ((format (printf, 5, 6)));
1927 extern void ext4_error_file(struct file *, const char *, unsigned int,
1928 ext4_fsblk_t, const char *, ...)
1929 __attribute__ ((format (printf, 5, 6)));
1930 extern void __ext4_std_error(struct super_block *, const char *,
1931 unsigned int, int);
1932 extern void __ext4_abort(struct super_block *, const char *, unsigned int,
1933 const char *, ...)
1934 __attribute__ ((format (printf, 4, 5)));
1935 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1936 __LINE__, ## message)
1937 extern void __ext4_warning(struct super_block *, const char *, unsigned int,
1938 const char *, ...)
1939 __attribute__ ((format (printf, 4, 5)));
1940 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1941 __LINE__, ## message)
1942 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1943 __attribute__ ((format (printf, 3, 4)));
1944 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
1945 const char *, unsigned int, const char *);
1946 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
1947 __LINE__, msg)
1948 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1949 struct super_block *, ext4_group_t, \
1950 unsigned long, ext4_fsblk_t, \
1951 const char *, ...)
1952 __attribute__ ((format (printf, 7, 8)));
1953 #define ext4_grp_locked_error(sb, grp, message...) \
1954 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1955 extern void ext4_update_dynamic_rev(struct super_block *sb);
1956 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1957 __u32 compat);
1958 extern int ext4_update_rocompat_feature(handle_t *handle,
1959 struct super_block *sb, __u32 rocompat);
1960 extern int ext4_update_incompat_feature(handle_t *handle,
1961 struct super_block *sb, __u32 incompat);
1962 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1963 struct ext4_group_desc *bg);
1964 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1965 struct ext4_group_desc *bg);
1966 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1967 struct ext4_group_desc *bg);
1968 extern __u32 ext4_free_blks_count(struct super_block *sb,
1969 struct ext4_group_desc *bg);
1970 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1971 struct ext4_group_desc *bg);
1972 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1973 struct ext4_group_desc *bg);
1974 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1975 struct ext4_group_desc *bg);
1976 extern void ext4_block_bitmap_set(struct super_block *sb,
1977 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1978 extern void ext4_inode_bitmap_set(struct super_block *sb,
1979 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1980 extern void ext4_inode_table_set(struct super_block *sb,
1981 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1982 extern void ext4_free_blks_set(struct super_block *sb,
1983 struct ext4_group_desc *bg, __u32 count);
1984 extern void ext4_free_inodes_set(struct super_block *sb,
1985 struct ext4_group_desc *bg, __u32 count);
1986 extern void ext4_used_dirs_set(struct super_block *sb,
1987 struct ext4_group_desc *bg, __u32 count);
1988 extern void ext4_itable_unused_set(struct super_block *sb,
1989 struct ext4_group_desc *bg, __u32 count);
1990 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1991 struct ext4_group_desc *gdp);
1992 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1993 struct ext4_group_desc *gdp);
1994
1995 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1996 {
1997 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1998 le32_to_cpu(es->s_blocks_count_lo);
1999 }
2000
2001 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2002 {
2003 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2004 le32_to_cpu(es->s_r_blocks_count_lo);
2005 }
2006
2007 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2008 {
2009 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2010 le32_to_cpu(es->s_free_blocks_count_lo);
2011 }
2012
2013 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2014 ext4_fsblk_t blk)
2015 {
2016 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2017 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2018 }
2019
2020 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2021 ext4_fsblk_t blk)
2022 {
2023 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2024 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2025 }
2026
2027 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2028 ext4_fsblk_t blk)
2029 {
2030 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2031 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2032 }
2033
2034 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2035 {
2036 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2037 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2038 le32_to_cpu(raw_inode->i_size_lo);
2039 else
2040 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2041 }
2042
2043 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2044 {
2045 raw_inode->i_size_lo = cpu_to_le32(i_size);
2046 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2047 }
2048
2049 static inline
2050 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2051 ext4_group_t group)
2052 {
2053 struct ext4_group_info ***grp_info;
2054 long indexv, indexh;
2055 grp_info = EXT4_SB(sb)->s_group_info;
2056 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2057 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2058 return grp_info[indexv][indexh];
2059 }
2060
2061 /*
2062 * Reading s_groups_count requires using smp_rmb() afterwards. See
2063 * the locking protocol documented in the comments of ext4_group_add()
2064 * in resize.c
2065 */
2066 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2067 {
2068 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2069
2070 smp_rmb();
2071 return ngroups;
2072 }
2073
2074 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2075 ext4_group_t block_group)
2076 {
2077 return block_group >> sbi->s_log_groups_per_flex;
2078 }
2079
2080 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2081 {
2082 return 1 << sbi->s_log_groups_per_flex;
2083 }
2084
2085 #define ext4_std_error(sb, errno) \
2086 do { \
2087 if ((errno)) \
2088 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2089 } while (0)
2090
2091 #ifdef CONFIG_SMP
2092 /* Each CPU can accumulate percpu_counter_batch blocks in their local
2093 * counters. So we need to make sure we have free blocks more
2094 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2095 */
2096 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2097 #else
2098 #define EXT4_FREEBLOCKS_WATERMARK 0
2099 #endif
2100
2101 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2102 {
2103 /*
2104 * XXX: replace with spinlock if seen contended -bzzz
2105 */
2106 down_write(&EXT4_I(inode)->i_data_sem);
2107 if (newsize > EXT4_I(inode)->i_disksize)
2108 EXT4_I(inode)->i_disksize = newsize;
2109 up_write(&EXT4_I(inode)->i_data_sem);
2110 return ;
2111 }
2112
2113 struct ext4_group_info {
2114 unsigned long bb_state;
2115 struct rb_root bb_free_root;
2116 ext4_grpblk_t bb_first_free; /* first free block */
2117 ext4_grpblk_t bb_free; /* total free blocks */
2118 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2119 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2120 struct list_head bb_prealloc_list;
2121 #ifdef DOUBLE_CHECK
2122 void *bb_bitmap;
2123 #endif
2124 struct rw_semaphore alloc_sem;
2125 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2126 * regions, index is order.
2127 * bb_counters[3] = 5 means
2128 * 5 free 8-block regions. */
2129 };
2130
2131 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2132 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2133
2134 #define EXT4_MB_GRP_NEED_INIT(grp) \
2135 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2136
2137 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2138 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2139 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2140 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2141 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2142 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2143
2144 #define EXT4_MAX_CONTENTION 8
2145 #define EXT4_CONTENTION_THRESHOLD 2
2146
2147 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2148 ext4_group_t group)
2149 {
2150 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2151 }
2152
2153 /*
2154 * Returns true if the filesystem is busy enough that attempts to
2155 * access the block group locks has run into contention.
2156 */
2157 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2158 {
2159 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2160 }
2161
2162 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2163 {
2164 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2165 if (spin_trylock(lock))
2166 /*
2167 * We're able to grab the lock right away, so drop the
2168 * lock contention counter.
2169 */
2170 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2171 else {
2172 /*
2173 * The lock is busy, so bump the contention counter,
2174 * and then wait on the spin lock.
2175 */
2176 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2177 EXT4_MAX_CONTENTION);
2178 spin_lock(lock);
2179 }
2180 }
2181
2182 static inline void ext4_unlock_group(struct super_block *sb,
2183 ext4_group_t group)
2184 {
2185 spin_unlock(ext4_group_lock_ptr(sb, group));
2186 }
2187
2188 static inline void ext4_mark_super_dirty(struct super_block *sb)
2189 {
2190 if (EXT4_SB(sb)->s_journal == NULL)
2191 sb->s_dirt =1;
2192 }
2193
2194 /*
2195 * Block validity checking
2196 */
2197 #define ext4_check_indirect_blockref(inode, bh) \
2198 ext4_check_blockref(__func__, __LINE__, inode, \
2199 (__le32 *)(bh)->b_data, \
2200 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2201
2202 #define ext4_ind_check_inode(inode) \
2203 ext4_check_blockref(__func__, __LINE__, inode, \
2204 EXT4_I(inode)->i_data, \
2205 EXT4_NDIR_BLOCKS)
2206
2207 /*
2208 * Inodes and files operations
2209 */
2210
2211 /* dir.c */
2212 extern const struct file_operations ext4_dir_operations;
2213
2214 /* file.c */
2215 extern const struct inode_operations ext4_file_inode_operations;
2216 extern const struct file_operations ext4_file_operations;
2217 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2218
2219 /* namei.c */
2220 extern const struct inode_operations ext4_dir_inode_operations;
2221 extern const struct inode_operations ext4_special_inode_operations;
2222 extern struct dentry *ext4_get_parent(struct dentry *child);
2223
2224 /* symlink.c */
2225 extern const struct inode_operations ext4_symlink_inode_operations;
2226 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2227
2228 /* block_validity */
2229 extern void ext4_release_system_zone(struct super_block *sb);
2230 extern int ext4_setup_system_zone(struct super_block *sb);
2231 extern int __init ext4_init_system_zone(void);
2232 extern void ext4_exit_system_zone(void);
2233 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2234 ext4_fsblk_t start_blk,
2235 unsigned int count);
2236 extern int ext4_check_blockref(const char *, unsigned int,
2237 struct inode *, __le32 *, unsigned int);
2238
2239 /* extents.c */
2240 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2241 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2242 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2243 int chunk);
2244 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2245 struct ext4_map_blocks *map, int flags);
2246 extern void ext4_ext_truncate(struct inode *);
2247 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2248 loff_t length);
2249 extern void ext4_ext_init(struct super_block *);
2250 extern void ext4_ext_release(struct super_block *);
2251 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2252 loff_t len);
2253 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2254 ssize_t len);
2255 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2256 struct ext4_map_blocks *map, int flags);
2257 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2258 __u64 start, __u64 len);
2259 /* move_extent.c */
2260 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2261 __u64 start_orig, __u64 start_donor,
2262 __u64 len, __u64 *moved_len);
2263
2264 /* page-io.c */
2265 extern int __init ext4_init_pageio(void);
2266 extern void ext4_exit_pageio(void);
2267 extern void ext4_ioend_wait(struct inode *);
2268 extern void ext4_free_io_end(ext4_io_end_t *io);
2269 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2270 extern int ext4_end_io_nolock(ext4_io_end_t *io);
2271 extern void ext4_io_submit(struct ext4_io_submit *io);
2272 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2273 struct page *page,
2274 int len,
2275 struct writeback_control *wbc);
2276
2277 /* mmp.c */
2278 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2279
2280 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2281 enum ext4_state_bits {
2282 BH_Uninit /* blocks are allocated but uninitialized on disk */
2283 = BH_JBDPrivateStart,
2284 };
2285
2286 BUFFER_FNS(Uninit, uninit)
2287 TAS_BUFFER_FNS(Uninit, uninit)
2288
2289 /*
2290 * Add new method to test wether block and inode bitmaps are properly
2291 * initialized. With uninit_bg reading the block from disk is not enough
2292 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2293 */
2294 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2295
2296 static inline int bitmap_uptodate(struct buffer_head *bh)
2297 {
2298 return (buffer_uptodate(bh) &&
2299 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2300 }
2301 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2302 {
2303 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2304 }
2305
2306 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2307
2308 /* For ioend & aio unwritten conversion wait queues */
2309 #define EXT4_WQ_HASH_SZ 37
2310 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2311 EXT4_WQ_HASH_SZ])
2312 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2313 EXT4_WQ_HASH_SZ])
2314 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2315 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2316
2317 #define EXT4_RESIZING 0
2318 extern int ext4_resize_begin(struct super_block *sb);
2319 extern void ext4_resize_end(struct super_block *sb);
2320
2321 #endif /* __KERNEL__ */
2322
2323 #endif /* _EXT4_H */
This page took 0.08406 seconds and 5 git commands to generate.