307ecd13a7625ca7182fb63a44e75152760296bf
[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
33 /*
34 * The fourth extended filesystem constants/structures
35 */
36
37 /*
38 * Define EXT4FS_DEBUG to produce debug messages
39 */
40 #undef EXT4FS_DEBUG
41
42 /*
43 * Debug code
44 */
45 #ifdef EXT4FS_DEBUG
46 #define ext4_debug(f, a...) \
47 do { \
48 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
49 __FILE__, __LINE__, __func__); \
50 printk(KERN_DEBUG f, ## a); \
51 } while (0)
52 #else
53 #define ext4_debug(f, a...) do {} while (0)
54 #endif
55
56 /* data type for block offset of block group */
57 typedef int ext4_grpblk_t;
58
59 /* data type for filesystem-wide blocks number */
60 typedef unsigned long long ext4_fsblk_t;
61
62 /* data type for file logical block number */
63 typedef __u32 ext4_lblk_t;
64
65 /* data type for block group number */
66 typedef unsigned int ext4_group_t;
67
68 /*
69 * Flags used in mballoc's allocation_context flags field.
70 *
71 * Also used to show what's going on for debugging purposes when the
72 * flag field is exported via the traceport interface
73 */
74
75 /* prefer goal again. length */
76 #define EXT4_MB_HINT_MERGE 0x0001
77 /* blocks already reserved */
78 #define EXT4_MB_HINT_RESERVED 0x0002
79 /* metadata is being allocated */
80 #define EXT4_MB_HINT_METADATA 0x0004
81 /* first blocks in the file */
82 #define EXT4_MB_HINT_FIRST 0x0008
83 /* search for the best chunk */
84 #define EXT4_MB_HINT_BEST 0x0010
85 /* data is being allocated */
86 #define EXT4_MB_HINT_DATA 0x0020
87 /* don't preallocate (for tails) */
88 #define EXT4_MB_HINT_NOPREALLOC 0x0040
89 /* allocate for locality group */
90 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
91 /* allocate goal blocks or none */
92 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
93 /* goal is meaningful */
94 #define EXT4_MB_HINT_TRY_GOAL 0x0200
95 /* blocks already pre-reserved by delayed allocation */
96 #define EXT4_MB_DELALLOC_RESERVED 0x0400
97 /* We are doing stream allocation */
98 #define EXT4_MB_STREAM_ALLOC 0x0800
99
100
101 struct ext4_allocation_request {
102 /* target inode for block we're allocating */
103 struct inode *inode;
104 /* how many blocks we want to allocate */
105 unsigned int len;
106 /* logical block in target inode */
107 ext4_lblk_t logical;
108 /* the closest logical allocated block to the left */
109 ext4_lblk_t lleft;
110 /* the closest logical allocated block to the right */
111 ext4_lblk_t lright;
112 /* phys. target (a hint) */
113 ext4_fsblk_t goal;
114 /* phys. block for the closest logical allocated block to the left */
115 ext4_fsblk_t pleft;
116 /* phys. block for the closest logical allocated block to the right */
117 ext4_fsblk_t pright;
118 /* flags. see above EXT4_MB_HINT_* */
119 unsigned int flags;
120 };
121
122 /*
123 * For delayed allocation tracking
124 */
125 struct mpage_da_data {
126 struct inode *inode;
127 sector_t b_blocknr; /* start block number of extent */
128 size_t b_size; /* size of extent */
129 unsigned long b_state; /* state of the extent */
130 unsigned long first_page, next_page; /* extent of pages */
131 struct writeback_control *wbc;
132 int io_done;
133 int pages_written;
134 int retval;
135 };
136 #define DIO_AIO_UNWRITTEN 0x1
137 typedef struct ext4_io_end {
138 struct list_head list; /* per-file finished AIO list */
139 struct inode *inode; /* file being written to */
140 unsigned int flag; /* unwritten or not */
141 int error; /* I/O error code */
142 loff_t offset; /* offset in the file */
143 ssize_t size; /* size of the extent */
144 struct work_struct work; /* data work queue */
145 } ext4_io_end_t;
146
147 /*
148 * Special inodes numbers
149 */
150 #define EXT4_BAD_INO 1 /* Bad blocks inode */
151 #define EXT4_ROOT_INO 2 /* Root inode */
152 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
153 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
154 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
155 #define EXT4_JOURNAL_INO 8 /* Journal inode */
156
157 /* First non-reserved inode for old ext4 filesystems */
158 #define EXT4_GOOD_OLD_FIRST_INO 11
159
160 /*
161 * Maximal count of links to a file
162 */
163 #define EXT4_LINK_MAX 65000
164
165 /*
166 * Macro-instructions used to manage several block sizes
167 */
168 #define EXT4_MIN_BLOCK_SIZE 1024
169 #define EXT4_MAX_BLOCK_SIZE 65536
170 #define EXT4_MIN_BLOCK_LOG_SIZE 10
171 #ifdef __KERNEL__
172 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
173 #else
174 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
175 #endif
176 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
177 #ifdef __KERNEL__
178 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
179 #else
180 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
181 #endif
182 #ifdef __KERNEL__
183 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
184 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
185 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
186 #else
187 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
188 EXT4_GOOD_OLD_INODE_SIZE : \
189 (s)->s_inode_size)
190 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
191 EXT4_GOOD_OLD_FIRST_INO : \
192 (s)->s_first_ino)
193 #endif
194 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
195
196 /*
197 * Structure of a blocks group descriptor
198 */
199 struct ext4_group_desc
200 {
201 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
202 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
203 __le32 bg_inode_table_lo; /* Inodes table block */
204 __le16 bg_free_blocks_count_lo;/* Free blocks count */
205 __le16 bg_free_inodes_count_lo;/* Free inodes count */
206 __le16 bg_used_dirs_count_lo; /* Directories count */
207 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
208 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
209 __le16 bg_itable_unused_lo; /* Unused inodes count */
210 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
211 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
212 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
213 __le32 bg_inode_table_hi; /* Inodes table block MSB */
214 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
215 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
216 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
217 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
218 __u32 bg_reserved2[3];
219 };
220
221 /*
222 * Structure of a flex block group info
223 */
224
225 struct flex_groups {
226 atomic_t free_inodes;
227 atomic_t free_blocks;
228 atomic_t used_dirs;
229 };
230
231 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
232 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
233 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
234
235 /*
236 * Macro-instructions used to manage group descriptors
237 */
238 #define EXT4_MIN_DESC_SIZE 32
239 #define EXT4_MIN_DESC_SIZE_64BIT 64
240 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
241 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
242 #ifdef __KERNEL__
243 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
244 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
245 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
246 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
247 #else
248 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
249 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
250 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
251 #endif
252
253 /*
254 * Constants relative to the data blocks
255 */
256 #define EXT4_NDIR_BLOCKS 12
257 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
258 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
259 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
260 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
261
262 /*
263 * Inode flags
264 */
265 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
266 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
267 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
268 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
269 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
270 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
271 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
272 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
273 /* Reserved for compression usage... */
274 #define EXT4_DIRTY_FL 0x00000100
275 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
276 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
277 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
278 /* End compression flags --- maybe not all used */
279 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
280 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
281 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
282 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
283 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
284 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
285 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
286 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
287 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
288
289 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
290 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
291
292 /* Flags that should be inherited by new inodes from their parent. */
293 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
294 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
295 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
296 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
297 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
298
299 /* Flags that are appropriate for regular files (all but dir-specific ones). */
300 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
301
302 /* Flags that are appropriate for non-directories/regular files. */
303 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
304
305 /* Mask out flags that are inappropriate for the given type of inode. */
306 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
307 {
308 if (S_ISDIR(mode))
309 return flags;
310 else if (S_ISREG(mode))
311 return flags & EXT4_REG_FLMASK;
312 else
313 return flags & EXT4_OTHER_FLMASK;
314 }
315
316 /*
317 * Inode dynamic state flags
318 */
319 #define EXT4_STATE_JDATA 0x00000001 /* journaled data exists */
320 #define EXT4_STATE_NEW 0x00000002 /* inode is newly created */
321 #define EXT4_STATE_XATTR 0x00000004 /* has in-inode xattrs */
322 #define EXT4_STATE_NO_EXPAND 0x00000008 /* No space for expansion */
323 #define EXT4_STATE_DA_ALLOC_CLOSE 0x00000010 /* Alloc DA blks on close */
324 #define EXT4_STATE_EXT_MIGRATE 0x00000020 /* Inode is migrating */
325 #define EXT4_STATE_DIO_UNWRITTEN 0x00000040 /* need convert on dio done*/
326
327 /* Used to pass group descriptor data when online resize is done */
328 struct ext4_new_group_input {
329 __u32 group; /* Group number for this data */
330 __u64 block_bitmap; /* Absolute block number of block bitmap */
331 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
332 __u64 inode_table; /* Absolute block number of inode table start */
333 __u32 blocks_count; /* Total number of blocks in this group */
334 __u16 reserved_blocks; /* Number of reserved blocks in this group */
335 __u16 unused;
336 };
337
338 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
339 struct ext4_new_group_data {
340 __u32 group;
341 __u64 block_bitmap;
342 __u64 inode_bitmap;
343 __u64 inode_table;
344 __u32 blocks_count;
345 __u16 reserved_blocks;
346 __u16 unused;
347 __u32 free_blocks_count;
348 };
349
350 /*
351 * Flags used by ext4_get_blocks()
352 */
353 /* Allocate any needed blocks and/or convert an unitialized
354 extent to be an initialized ext4 */
355 #define EXT4_GET_BLOCKS_CREATE 0x0001
356 /* Request the creation of an unitialized extent */
357 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
358 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
359 EXT4_GET_BLOCKS_CREATE)
360 /* Caller is from the delayed allocation writeout path,
361 so set the magic i_delalloc_reserve_flag after taking the
362 inode allocation semaphore for */
363 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
364 /* caller is from the direct IO path, request to creation of an
365 unitialized extents if not allocated, split the uninitialized
366 extent if blocks has been preallocated already*/
367 #define EXT4_GET_BLOCKS_DIO 0x0008
368 #define EXT4_GET_BLOCKS_CONVERT 0x0010
369 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
370 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
371 /* Convert extent to initialized after direct IO complete */
372 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
373 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
374
375 /*
376 * Flags used by ext4_free_blocks
377 */
378 #define EXT4_FREE_BLOCKS_METADATA 0x0001
379 #define EXT4_FREE_BLOCKS_FORGET 0x0002
380 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
381
382 /*
383 * ioctl commands
384 */
385 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
386 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
387 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
388 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
389 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
390 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
391 #ifdef CONFIG_JBD2_DEBUG
392 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
393 #endif
394 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
395 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
396 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
397 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
398 #define EXT4_IOC_MIGRATE _IO('f', 9)
399 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
400 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
401 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
402 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
403
404 /*
405 * ioctl commands in 32 bit emulation
406 */
407 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
408 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
409 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
410 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
411 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
412 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
413 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
414 #ifdef CONFIG_JBD2_DEBUG
415 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
416 #endif
417 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
418 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
419
420
421 /*
422 * Mount options
423 */
424 struct ext4_mount_options {
425 unsigned long s_mount_opt;
426 uid_t s_resuid;
427 gid_t s_resgid;
428 unsigned long s_commit_interval;
429 u32 s_min_batch_time, s_max_batch_time;
430 #ifdef CONFIG_QUOTA
431 int s_jquota_fmt;
432 char *s_qf_names[MAXQUOTAS];
433 #endif
434 };
435
436 /* Max physical block we can addres w/o extents */
437 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
438
439 /*
440 * Structure of an inode on the disk
441 */
442 struct ext4_inode {
443 __le16 i_mode; /* File mode */
444 __le16 i_uid; /* Low 16 bits of Owner Uid */
445 __le32 i_size_lo; /* Size in bytes */
446 __le32 i_atime; /* Access time */
447 __le32 i_ctime; /* Inode Change time */
448 __le32 i_mtime; /* Modification time */
449 __le32 i_dtime; /* Deletion Time */
450 __le16 i_gid; /* Low 16 bits of Group Id */
451 __le16 i_links_count; /* Links count */
452 __le32 i_blocks_lo; /* Blocks count */
453 __le32 i_flags; /* File flags */
454 union {
455 struct {
456 __le32 l_i_version;
457 } linux1;
458 struct {
459 __u32 h_i_translator;
460 } hurd1;
461 struct {
462 __u32 m_i_reserved1;
463 } masix1;
464 } osd1; /* OS dependent 1 */
465 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
466 __le32 i_generation; /* File version (for NFS) */
467 __le32 i_file_acl_lo; /* File ACL */
468 __le32 i_size_high;
469 __le32 i_obso_faddr; /* Obsoleted fragment address */
470 union {
471 struct {
472 __le16 l_i_blocks_high; /* were l_i_reserved1 */
473 __le16 l_i_file_acl_high;
474 __le16 l_i_uid_high; /* these 2 fields */
475 __le16 l_i_gid_high; /* were reserved2[0] */
476 __u32 l_i_reserved2;
477 } linux2;
478 struct {
479 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
480 __u16 h_i_mode_high;
481 __u16 h_i_uid_high;
482 __u16 h_i_gid_high;
483 __u32 h_i_author;
484 } hurd2;
485 struct {
486 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
487 __le16 m_i_file_acl_high;
488 __u32 m_i_reserved2[2];
489 } masix2;
490 } osd2; /* OS dependent 2 */
491 __le16 i_extra_isize;
492 __le16 i_pad1;
493 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
494 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
495 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
496 __le32 i_crtime; /* File Creation time */
497 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
498 __le32 i_version_hi; /* high 32 bits for 64-bit version */
499 };
500
501 struct move_extent {
502 __u32 reserved; /* should be zero */
503 __u32 donor_fd; /* donor file descriptor */
504 __u64 orig_start; /* logical start offset in block for orig */
505 __u64 donor_start; /* logical start offset in block for donor */
506 __u64 len; /* block length to be moved */
507 __u64 moved_len; /* moved block length */
508 };
509
510 #define EXT4_EPOCH_BITS 2
511 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
512 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
513
514 /*
515 * Extended fields will fit into an inode if the filesystem was formatted
516 * with large inodes (-I 256 or larger) and there are not currently any EAs
517 * consuming all of the available space. For new inodes we always reserve
518 * enough space for the kernel's known extended fields, but for inodes
519 * created with an old kernel this might not have been the case. None of
520 * the extended inode fields is critical for correct filesystem operation.
521 * This macro checks if a certain field fits in the inode. Note that
522 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
523 */
524 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
525 ((offsetof(typeof(*ext4_inode), field) + \
526 sizeof((ext4_inode)->field)) \
527 <= (EXT4_GOOD_OLD_INODE_SIZE + \
528 (einode)->i_extra_isize)) \
529
530 static inline __le32 ext4_encode_extra_time(struct timespec *time)
531 {
532 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
533 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
534 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
535 }
536
537 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
538 {
539 if (sizeof(time->tv_sec) > 4)
540 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
541 << 32;
542 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
543 }
544
545 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
546 do { \
547 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
548 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
549 (raw_inode)->xtime ## _extra = \
550 ext4_encode_extra_time(&(inode)->xtime); \
551 } while (0)
552
553 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
554 do { \
555 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
556 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
557 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
558 (raw_inode)->xtime ## _extra = \
559 ext4_encode_extra_time(&(einode)->xtime); \
560 } while (0)
561
562 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
563 do { \
564 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
565 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
566 ext4_decode_extra_time(&(inode)->xtime, \
567 raw_inode->xtime ## _extra); \
568 } while (0)
569
570 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
571 do { \
572 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
573 (einode)->xtime.tv_sec = \
574 (signed)le32_to_cpu((raw_inode)->xtime); \
575 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
576 ext4_decode_extra_time(&(einode)->xtime, \
577 raw_inode->xtime ## _extra); \
578 } while (0)
579
580 #define i_disk_version osd1.linux1.l_i_version
581
582 #if defined(__KERNEL__) || defined(__linux__)
583 #define i_reserved1 osd1.linux1.l_i_reserved1
584 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
585 #define i_blocks_high osd2.linux2.l_i_blocks_high
586 #define i_uid_low i_uid
587 #define i_gid_low i_gid
588 #define i_uid_high osd2.linux2.l_i_uid_high
589 #define i_gid_high osd2.linux2.l_i_gid_high
590 #define i_reserved2 osd2.linux2.l_i_reserved2
591
592 #elif defined(__GNU__)
593
594 #define i_translator osd1.hurd1.h_i_translator
595 #define i_uid_high osd2.hurd2.h_i_uid_high
596 #define i_gid_high osd2.hurd2.h_i_gid_high
597 #define i_author osd2.hurd2.h_i_author
598
599 #elif defined(__masix__)
600
601 #define i_reserved1 osd1.masix1.m_i_reserved1
602 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
603 #define i_reserved2 osd2.masix2.m_i_reserved2
604
605 #endif /* defined(__KERNEL__) || defined(__linux__) */
606
607 /*
608 * storage for cached extent
609 */
610 struct ext4_ext_cache {
611 ext4_fsblk_t ec_start;
612 ext4_lblk_t ec_block;
613 __u32 ec_len; /* must be 32bit to return holes */
614 __u32 ec_type;
615 };
616
617 /*
618 * fourth extended file system inode data in memory
619 */
620 struct ext4_inode_info {
621 __le32 i_data[15]; /* unconverted */
622 __u32 i_flags;
623 ext4_fsblk_t i_file_acl;
624 __u32 i_dtime;
625
626 /*
627 * i_block_group is the number of the block group which contains
628 * this file's inode. Constant across the lifetime of the inode,
629 * it is ued for making block allocation decisions - we try to
630 * place a file's data blocks near its inode block, and new inodes
631 * near to their parent directory's inode.
632 */
633 ext4_group_t i_block_group;
634 __u32 i_state; /* Dynamic state flags for ext4 */
635
636 ext4_lblk_t i_dir_start_lookup;
637 #ifdef CONFIG_EXT4_FS_XATTR
638 /*
639 * Extended attributes can be read independently of the main file
640 * data. Taking i_mutex even when reading would cause contention
641 * between readers of EAs and writers of regular file data, so
642 * instead we synchronize on xattr_sem when reading or changing
643 * EAs.
644 */
645 struct rw_semaphore xattr_sem;
646 #endif
647
648 struct list_head i_orphan; /* unlinked but open inodes */
649
650 /*
651 * i_disksize keeps track of what the inode size is ON DISK, not
652 * in memory. During truncate, i_size is set to the new size by
653 * the VFS prior to calling ext4_truncate(), but the filesystem won't
654 * set i_disksize to 0 until the truncate is actually under way.
655 *
656 * The intent is that i_disksize always represents the blocks which
657 * are used by this file. This allows recovery to restart truncate
658 * on orphans if we crash during truncate. We actually write i_disksize
659 * into the on-disk inode when writing inodes out, instead of i_size.
660 *
661 * The only time when i_disksize and i_size may be different is when
662 * a truncate is in progress. The only things which change i_disksize
663 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
664 */
665 loff_t i_disksize;
666
667 /*
668 * i_data_sem is for serialising ext4_truncate() against
669 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
670 * data tree are chopped off during truncate. We can't do that in
671 * ext4 because whenever we perform intermediate commits during
672 * truncate, the inode and all the metadata blocks *must* be in a
673 * consistent state which allows truncation of the orphans to restart
674 * during recovery. Hence we must fix the get_block-vs-truncate race
675 * by other means, so we have i_data_sem.
676 */
677 struct rw_semaphore i_data_sem;
678 struct inode vfs_inode;
679 struct jbd2_inode jinode;
680
681 struct ext4_ext_cache i_cached_extent;
682 /*
683 * File creation time. Its function is same as that of
684 * struct timespec i_{a,c,m}time in the generic inode.
685 */
686 struct timespec i_crtime;
687
688 /* mballoc */
689 struct list_head i_prealloc_list;
690 spinlock_t i_prealloc_lock;
691
692 /* ialloc */
693 ext4_group_t i_last_alloc_group;
694
695 /* allocation reservation info for delalloc */
696 unsigned int i_reserved_data_blocks;
697 unsigned int i_reserved_meta_blocks;
698 unsigned int i_allocated_meta_blocks;
699 unsigned short i_delalloc_reserved_flag;
700 sector_t i_da_metadata_calc_last_lblock;
701 int i_da_metadata_calc_len;
702
703 /* on-disk additional length */
704 __u16 i_extra_isize;
705
706 spinlock_t i_block_reservation_lock;
707 #ifdef CONFIG_QUOTA
708 /* quota space reservation, managed internally by quota code */
709 qsize_t i_reserved_quota;
710 #endif
711
712 /* completed async DIOs that might need unwritten extents handling */
713 struct list_head i_aio_dio_complete_list;
714 /* current io_end structure for async DIO write*/
715 ext4_io_end_t *cur_aio_dio;
716
717 /*
718 * Transactions that contain inode's metadata needed to complete
719 * fsync and fdatasync, respectively.
720 */
721 tid_t i_sync_tid;
722 tid_t i_datasync_tid;
723 };
724
725 /*
726 * File system states
727 */
728 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
729 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
730 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
731
732 /*
733 * Misc. filesystem flags
734 */
735 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
736 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
737 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
738
739 /*
740 * Mount flags
741 */
742 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
743 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
744 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
745 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
746 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
747 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
748 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
749 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
750 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
751 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
752 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
753 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
754 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
755 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
756 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
757 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
758 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
759 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
760 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
761 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
762 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
763 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
764 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
765 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
766 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
767 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
768 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
769 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
770 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
771
772 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
773 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt
774 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
775 EXT4_MOUNT_##opt)
776
777 #define ext4_set_bit ext2_set_bit
778 #define ext4_set_bit_atomic ext2_set_bit_atomic
779 #define ext4_clear_bit ext2_clear_bit
780 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
781 #define ext4_test_bit ext2_test_bit
782 #define ext4_find_first_zero_bit ext2_find_first_zero_bit
783 #define ext4_find_next_zero_bit ext2_find_next_zero_bit
784 #define ext4_find_next_bit ext2_find_next_bit
785
786 /*
787 * Maximal mount counts between two filesystem checks
788 */
789 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
790 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
791
792 /*
793 * Behaviour when detecting errors
794 */
795 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
796 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
797 #define EXT4_ERRORS_PANIC 3 /* Panic */
798 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
799
800 /*
801 * Structure of the super block
802 */
803 struct ext4_super_block {
804 /*00*/ __le32 s_inodes_count; /* Inodes count */
805 __le32 s_blocks_count_lo; /* Blocks count */
806 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
807 __le32 s_free_blocks_count_lo; /* Free blocks count */
808 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
809 __le32 s_first_data_block; /* First Data Block */
810 __le32 s_log_block_size; /* Block size */
811 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
812 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
813 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
814 __le32 s_inodes_per_group; /* # Inodes per group */
815 __le32 s_mtime; /* Mount time */
816 /*30*/ __le32 s_wtime; /* Write time */
817 __le16 s_mnt_count; /* Mount count */
818 __le16 s_max_mnt_count; /* Maximal mount count */
819 __le16 s_magic; /* Magic signature */
820 __le16 s_state; /* File system state */
821 __le16 s_errors; /* Behaviour when detecting errors */
822 __le16 s_minor_rev_level; /* minor revision level */
823 /*40*/ __le32 s_lastcheck; /* time of last check */
824 __le32 s_checkinterval; /* max. time between checks */
825 __le32 s_creator_os; /* OS */
826 __le32 s_rev_level; /* Revision level */
827 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
828 __le16 s_def_resgid; /* Default gid for reserved blocks */
829 /*
830 * These fields are for EXT4_DYNAMIC_REV superblocks only.
831 *
832 * Note: the difference between the compatible feature set and
833 * the incompatible feature set is that if there is a bit set
834 * in the incompatible feature set that the kernel doesn't
835 * know about, it should refuse to mount the filesystem.
836 *
837 * e2fsck's requirements are more strict; if it doesn't know
838 * about a feature in either the compatible or incompatible
839 * feature set, it must abort and not try to meddle with
840 * things it doesn't understand...
841 */
842 __le32 s_first_ino; /* First non-reserved inode */
843 __le16 s_inode_size; /* size of inode structure */
844 __le16 s_block_group_nr; /* block group # of this superblock */
845 __le32 s_feature_compat; /* compatible feature set */
846 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
847 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
848 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
849 /*78*/ char s_volume_name[16]; /* volume name */
850 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
851 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
852 /*
853 * Performance hints. Directory preallocation should only
854 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
855 */
856 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
857 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
858 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
859 /*
860 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
861 */
862 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
863 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
864 __le32 s_journal_dev; /* device number of journal file */
865 __le32 s_last_orphan; /* start of list of inodes to delete */
866 __le32 s_hash_seed[4]; /* HTREE hash seed */
867 __u8 s_def_hash_version; /* Default hash version to use */
868 __u8 s_reserved_char_pad;
869 __le16 s_desc_size; /* size of group descriptor */
870 /*100*/ __le32 s_default_mount_opts;
871 __le32 s_first_meta_bg; /* First metablock block group */
872 __le32 s_mkfs_time; /* When the filesystem was created */
873 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
874 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
875 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
876 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
877 __le32 s_free_blocks_count_hi; /* Free blocks count */
878 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
879 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
880 __le32 s_flags; /* Miscellaneous flags */
881 __le16 s_raid_stride; /* RAID stride */
882 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */
883 __le64 s_mmp_block; /* Block for multi-mount protection */
884 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
885 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
886 __u8 s_reserved_char_pad2;
887 __le16 s_reserved_pad;
888 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
889 __u32 s_reserved[160]; /* Padding to the end of the block */
890 };
891
892 #ifdef __KERNEL__
893
894 /*
895 * run-time mount flags
896 */
897 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
898 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
899
900 /*
901 * fourth extended-fs super-block data in memory
902 */
903 struct ext4_sb_info {
904 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
905 unsigned long s_inodes_per_block;/* Number of inodes per block */
906 unsigned long s_blocks_per_group;/* Number of blocks in a group */
907 unsigned long s_inodes_per_group;/* Number of inodes in a group */
908 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
909 unsigned long s_gdb_count; /* Number of group descriptor blocks */
910 unsigned long s_desc_per_block; /* Number of group descriptors per block */
911 ext4_group_t s_groups_count; /* Number of groups in the fs */
912 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
913 unsigned long s_overhead_last; /* Last calculated overhead */
914 unsigned long s_blocks_last; /* Last seen block count */
915 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
916 struct buffer_head * s_sbh; /* Buffer containing the super block */
917 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
918 struct buffer_head **s_group_desc;
919 unsigned int s_mount_opt;
920 unsigned int s_mount_flags;
921 ext4_fsblk_t s_sb_block;
922 uid_t s_resuid;
923 gid_t s_resgid;
924 unsigned short s_mount_state;
925 unsigned short s_pad;
926 int s_addr_per_block_bits;
927 int s_desc_per_block_bits;
928 int s_inode_size;
929 int s_first_ino;
930 unsigned int s_inode_readahead_blks;
931 unsigned int s_inode_goal;
932 spinlock_t s_next_gen_lock;
933 u32 s_next_generation;
934 u32 s_hash_seed[4];
935 int s_def_hash_version;
936 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
937 struct percpu_counter s_freeblocks_counter;
938 struct percpu_counter s_freeinodes_counter;
939 struct percpu_counter s_dirs_counter;
940 struct percpu_counter s_dirtyblocks_counter;
941 struct blockgroup_lock *s_blockgroup_lock;
942 struct proc_dir_entry *s_proc;
943 struct kobject s_kobj;
944 struct completion s_kobj_unregister;
945
946 /* Journaling */
947 struct inode *s_journal_inode;
948 struct journal_s *s_journal;
949 struct list_head s_orphan;
950 struct mutex s_orphan_lock;
951 struct mutex s_resize_lock;
952 unsigned long s_commit_interval;
953 u32 s_max_batch_time;
954 u32 s_min_batch_time;
955 struct block_device *journal_bdev;
956 #ifdef CONFIG_JBD2_DEBUG
957 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
958 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
959 #endif
960 #ifdef CONFIG_QUOTA
961 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
962 int s_jquota_fmt; /* Format of quota to use */
963 #endif
964 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
965 struct rb_root system_blks;
966
967 #ifdef EXTENTS_STATS
968 /* ext4 extents stats */
969 unsigned long s_ext_min;
970 unsigned long s_ext_max;
971 unsigned long s_depth_max;
972 spinlock_t s_ext_stats_lock;
973 unsigned long s_ext_blocks;
974 unsigned long s_ext_extents;
975 #endif
976
977 /* for buddy allocator */
978 struct ext4_group_info ***s_group_info;
979 struct inode *s_buddy_cache;
980 long s_blocks_reserved;
981 spinlock_t s_reserve_lock;
982 spinlock_t s_md_lock;
983 tid_t s_last_transaction;
984 unsigned short *s_mb_offsets;
985 unsigned int *s_mb_maxs;
986
987 /* tunables */
988 unsigned long s_stripe;
989 unsigned int s_mb_stream_request;
990 unsigned int s_mb_max_to_scan;
991 unsigned int s_mb_min_to_scan;
992 unsigned int s_mb_stats;
993 unsigned int s_mb_order2_reqs;
994 unsigned int s_mb_group_prealloc;
995 unsigned int s_max_writeback_mb_bump;
996 /* where last allocation was done - for stream allocation */
997 unsigned long s_mb_last_group;
998 unsigned long s_mb_last_start;
999
1000 /* stats for buddy allocator */
1001 spinlock_t s_mb_pa_lock;
1002 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1003 atomic_t s_bal_success; /* we found long enough chunks */
1004 atomic_t s_bal_allocated; /* in blocks */
1005 atomic_t s_bal_ex_scanned; /* total extents scanned */
1006 atomic_t s_bal_goals; /* goal hits */
1007 atomic_t s_bal_breaks; /* too long searches */
1008 atomic_t s_bal_2orders; /* 2^order hits */
1009 spinlock_t s_bal_lock;
1010 unsigned long s_mb_buddies_generated;
1011 unsigned long long s_mb_generation_time;
1012 atomic_t s_mb_lost_chunks;
1013 atomic_t s_mb_preallocated;
1014 atomic_t s_mb_discarded;
1015 atomic_t s_lock_busy;
1016
1017 /* locality groups */
1018 struct ext4_locality_group *s_locality_groups;
1019
1020 /* for write statistics */
1021 unsigned long s_sectors_written_start;
1022 u64 s_kbytes_written;
1023
1024 unsigned int s_log_groups_per_flex;
1025 struct flex_groups *s_flex_groups;
1026
1027 /* workqueue for dio unwritten */
1028 struct workqueue_struct *dio_unwritten_wq;
1029 };
1030
1031 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1032 {
1033 return sb->s_fs_info;
1034 }
1035 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1036 {
1037 return container_of(inode, struct ext4_inode_info, vfs_inode);
1038 }
1039
1040 static inline struct timespec ext4_current_time(struct inode *inode)
1041 {
1042 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1043 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1044 }
1045
1046 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1047 {
1048 return ino == EXT4_ROOT_INO ||
1049 ino == EXT4_JOURNAL_INO ||
1050 ino == EXT4_RESIZE_INO ||
1051 (ino >= EXT4_FIRST_INO(sb) &&
1052 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1053 }
1054 #else
1055 /* Assume that user mode programs are passing in an ext4fs superblock, not
1056 * a kernel struct super_block. This will allow us to call the feature-test
1057 * macros from user land. */
1058 #define EXT4_SB(sb) (sb)
1059 #endif
1060
1061 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1062
1063 /*
1064 * Codes for operating systems
1065 */
1066 #define EXT4_OS_LINUX 0
1067 #define EXT4_OS_HURD 1
1068 #define EXT4_OS_MASIX 2
1069 #define EXT4_OS_FREEBSD 3
1070 #define EXT4_OS_LITES 4
1071
1072 /*
1073 * Revision levels
1074 */
1075 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1076 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1077
1078 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1079 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1080
1081 #define EXT4_GOOD_OLD_INODE_SIZE 128
1082
1083 /*
1084 * Feature set definitions
1085 */
1086
1087 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1088 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1089 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1090 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1091 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1092 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1093 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1094 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1095 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1096 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1097 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1098 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1099 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1100 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1101 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1102 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1103 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1104 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1105
1106 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1107 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1108 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1109 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1110 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1111 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1112
1113 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1114 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1115 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1116 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1117 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1118 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1119 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1120
1121 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1122 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1123 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1124 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1125 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1126 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1127 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1128 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1129 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1130
1131 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1132 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1133 EXT4_FEATURE_INCOMPAT_RECOVER| \
1134 EXT4_FEATURE_INCOMPAT_META_BG| \
1135 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1136 EXT4_FEATURE_INCOMPAT_64BIT| \
1137 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1138 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1139 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1140 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1141 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1142 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1143 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1144 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1145
1146 /*
1147 * Default values for user and/or group using reserved blocks
1148 */
1149 #define EXT4_DEF_RESUID 0
1150 #define EXT4_DEF_RESGID 0
1151
1152 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1153
1154 /*
1155 * Default mount options
1156 */
1157 #define EXT4_DEFM_DEBUG 0x0001
1158 #define EXT4_DEFM_BSDGROUPS 0x0002
1159 #define EXT4_DEFM_XATTR_USER 0x0004
1160 #define EXT4_DEFM_ACL 0x0008
1161 #define EXT4_DEFM_UID16 0x0010
1162 #define EXT4_DEFM_JMODE 0x0060
1163 #define EXT4_DEFM_JMODE_DATA 0x0020
1164 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1165 #define EXT4_DEFM_JMODE_WBACK 0x0060
1166
1167 /*
1168 * Default journal batch times
1169 */
1170 #define EXT4_DEF_MIN_BATCH_TIME 0
1171 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1172
1173 /*
1174 * Minimum number of groups in a flexgroup before we separate out
1175 * directories into the first block group of a flexgroup
1176 */
1177 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1178
1179 /*
1180 * Structure of a directory entry
1181 */
1182 #define EXT4_NAME_LEN 255
1183
1184 struct ext4_dir_entry {
1185 __le32 inode; /* Inode number */
1186 __le16 rec_len; /* Directory entry length */
1187 __le16 name_len; /* Name length */
1188 char name[EXT4_NAME_LEN]; /* File name */
1189 };
1190
1191 /*
1192 * The new version of the directory entry. Since EXT4 structures are
1193 * stored in intel byte order, and the name_len field could never be
1194 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1195 * file_type field.
1196 */
1197 struct ext4_dir_entry_2 {
1198 __le32 inode; /* Inode number */
1199 __le16 rec_len; /* Directory entry length */
1200 __u8 name_len; /* Name length */
1201 __u8 file_type;
1202 char name[EXT4_NAME_LEN]; /* File name */
1203 };
1204
1205 /*
1206 * Ext4 directory file types. Only the low 3 bits are used. The
1207 * other bits are reserved for now.
1208 */
1209 #define EXT4_FT_UNKNOWN 0
1210 #define EXT4_FT_REG_FILE 1
1211 #define EXT4_FT_DIR 2
1212 #define EXT4_FT_CHRDEV 3
1213 #define EXT4_FT_BLKDEV 4
1214 #define EXT4_FT_FIFO 5
1215 #define EXT4_FT_SOCK 6
1216 #define EXT4_FT_SYMLINK 7
1217
1218 #define EXT4_FT_MAX 8
1219
1220 /*
1221 * EXT4_DIR_PAD defines the directory entries boundaries
1222 *
1223 * NOTE: It must be a multiple of 4
1224 */
1225 #define EXT4_DIR_PAD 4
1226 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1227 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1228 ~EXT4_DIR_ROUND)
1229 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1230
1231 /*
1232 * Hash Tree Directory indexing
1233 * (c) Daniel Phillips, 2001
1234 */
1235
1236 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1237 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1238 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1239 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1240 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1241
1242 /* Legal values for the dx_root hash_version field: */
1243
1244 #define DX_HASH_LEGACY 0
1245 #define DX_HASH_HALF_MD4 1
1246 #define DX_HASH_TEA 2
1247 #define DX_HASH_LEGACY_UNSIGNED 3
1248 #define DX_HASH_HALF_MD4_UNSIGNED 4
1249 #define DX_HASH_TEA_UNSIGNED 5
1250
1251 #ifdef __KERNEL__
1252
1253 /* hash info structure used by the directory hash */
1254 struct dx_hash_info
1255 {
1256 u32 hash;
1257 u32 minor_hash;
1258 int hash_version;
1259 u32 *seed;
1260 };
1261
1262 #define EXT4_HTREE_EOF 0x7fffffff
1263
1264 /*
1265 * Control parameters used by ext4_htree_next_block
1266 */
1267 #define HASH_NB_ALWAYS 1
1268
1269
1270 /*
1271 * Describe an inode's exact location on disk and in memory
1272 */
1273 struct ext4_iloc
1274 {
1275 struct buffer_head *bh;
1276 unsigned long offset;
1277 ext4_group_t block_group;
1278 };
1279
1280 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1281 {
1282 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1283 }
1284
1285 /*
1286 * This structure is stuffed into the struct file's private_data field
1287 * for directories. It is where we put information so that we can do
1288 * readdir operations in hash tree order.
1289 */
1290 struct dir_private_info {
1291 struct rb_root root;
1292 struct rb_node *curr_node;
1293 struct fname *extra_fname;
1294 loff_t last_pos;
1295 __u32 curr_hash;
1296 __u32 curr_minor_hash;
1297 __u32 next_hash;
1298 };
1299
1300 /* calculate the first block number of the group */
1301 static inline ext4_fsblk_t
1302 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1303 {
1304 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1305 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1306 }
1307
1308 /*
1309 * Special error return code only used by dx_probe() and its callers.
1310 */
1311 #define ERR_BAD_DX_DIR -75000
1312
1313 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1314 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1315
1316 extern struct proc_dir_entry *ext4_proc_root;
1317
1318 /*
1319 * Function prototypes
1320 */
1321
1322 /*
1323 * Ok, these declarations are also in <linux/kernel.h> but none of the
1324 * ext4 source programs needs to include it so they are duplicated here.
1325 */
1326 # define NORET_TYPE /**/
1327 # define ATTRIB_NORET __attribute__((noreturn))
1328 # define NORET_AND noreturn,
1329
1330 /* bitmap.c */
1331 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1332
1333 /* balloc.c */
1334 extern unsigned int ext4_block_group(struct super_block *sb,
1335 ext4_fsblk_t blocknr);
1336 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1337 ext4_fsblk_t blocknr);
1338 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1339 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1340 ext4_group_t group);
1341 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1342 ext4_fsblk_t goal, unsigned long *count, int *errp);
1343 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1344 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1345 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1346 ext4_fsblk_t block, unsigned long count);
1347 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1348 extern void ext4_check_blocks_bitmap(struct super_block *);
1349 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1350 ext4_group_t block_group,
1351 struct buffer_head ** bh);
1352 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1353 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1354 ext4_group_t block_group);
1355 extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1356 struct buffer_head *bh,
1357 ext4_group_t group,
1358 struct ext4_group_desc *desc);
1359 #define ext4_free_blocks_after_init(sb, group, desc) \
1360 ext4_init_block_bitmap(sb, NULL, group, desc)
1361
1362 /* dir.c */
1363 extern int ext4_check_dir_entry(const char *, struct inode *,
1364 struct ext4_dir_entry_2 *,
1365 struct buffer_head *, unsigned int);
1366 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1367 __u32 minor_hash,
1368 struct ext4_dir_entry_2 *dirent);
1369 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1370
1371 /* fsync.c */
1372 extern int ext4_sync_file(struct file *, struct dentry *, int);
1373
1374 /* hash.c */
1375 extern int ext4fs_dirhash(const char *name, int len, struct
1376 dx_hash_info *hinfo);
1377
1378 /* ialloc.c */
1379 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1380 const struct qstr *qstr, __u32 goal);
1381 extern void ext4_free_inode(handle_t *, struct inode *);
1382 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1383 extern unsigned long ext4_count_free_inodes(struct super_block *);
1384 extern unsigned long ext4_count_dirs(struct super_block *);
1385 extern void ext4_check_inodes_bitmap(struct super_block *);
1386 extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1387 struct buffer_head *bh,
1388 ext4_group_t group,
1389 struct ext4_group_desc *desc);
1390 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1391
1392 /* mballoc.c */
1393 extern long ext4_mb_stats;
1394 extern long ext4_mb_max_to_scan;
1395 extern int ext4_mb_init(struct super_block *, int);
1396 extern int ext4_mb_release(struct super_block *);
1397 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1398 struct ext4_allocation_request *, int *);
1399 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1400 extern void ext4_discard_preallocations(struct inode *);
1401 extern int __init init_ext4_mballoc(void);
1402 extern void exit_ext4_mballoc(void);
1403 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1404 struct buffer_head *bh, ext4_fsblk_t block,
1405 unsigned long count, int flags);
1406 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1407 ext4_group_t i, struct ext4_group_desc *desc);
1408 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1409 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1410 ext4_group_t, int);
1411 /* inode.c */
1412 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1413 ext4_lblk_t, int, int *);
1414 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1415 ext4_lblk_t, int, int *);
1416 int ext4_get_block(struct inode *inode, sector_t iblock,
1417 struct buffer_head *bh_result, int create);
1418
1419 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1420 extern int ext4_write_inode(struct inode *, int);
1421 extern int ext4_setattr(struct dentry *, struct iattr *);
1422 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1423 struct kstat *stat);
1424 extern void ext4_delete_inode(struct inode *);
1425 extern int ext4_sync_inode(handle_t *, struct inode *);
1426 extern void ext4_dirty_inode(struct inode *);
1427 extern int ext4_change_inode_journal_flag(struct inode *, int);
1428 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1429 extern int ext4_can_truncate(struct inode *inode);
1430 extern void ext4_truncate(struct inode *);
1431 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1432 extern void ext4_set_inode_flags(struct inode *);
1433 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1434 extern int ext4_alloc_da_blocks(struct inode *inode);
1435 extern void ext4_set_aops(struct inode *inode);
1436 extern int ext4_writepage_trans_blocks(struct inode *);
1437 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1438 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1439 extern int ext4_block_truncate_page(handle_t *handle,
1440 struct address_space *mapping, loff_t from);
1441 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1442 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1443 extern int flush_aio_dio_completed_IO(struct inode *inode);
1444 extern void ext4_da_update_reserve_space(struct inode *inode,
1445 int used, int quota_claim);
1446 /* ioctl.c */
1447 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1448 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1449
1450 /* migrate.c */
1451 extern int ext4_ext_migrate(struct inode *);
1452
1453 /* namei.c */
1454 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1455 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
1456 extern int ext4_orphan_add(handle_t *, struct inode *);
1457 extern int ext4_orphan_del(handle_t *, struct inode *);
1458 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1459 __u32 start_minor_hash, __u32 *next_hash);
1460
1461 /* resize.c */
1462 extern int ext4_group_add(struct super_block *sb,
1463 struct ext4_new_group_data *input);
1464 extern int ext4_group_extend(struct super_block *sb,
1465 struct ext4_super_block *es,
1466 ext4_fsblk_t n_blocks_count);
1467
1468 /* super.c */
1469 extern void ext4_error(struct super_block *, const char *, const char *, ...)
1470 __attribute__ ((format (printf, 3, 4)));
1471 extern void __ext4_std_error(struct super_block *, const char *, int);
1472 extern void ext4_abort(struct super_block *, const char *, const char *, ...)
1473 __attribute__ ((format (printf, 3, 4)));
1474 extern void ext4_warning(struct super_block *, const char *, const char *, ...)
1475 __attribute__ ((format (printf, 3, 4)));
1476 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1477 __attribute__ ((format (printf, 3, 4)));
1478 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1479 const char *, const char *, ...)
1480 __attribute__ ((format (printf, 4, 5)));
1481 extern void ext4_update_dynamic_rev(struct super_block *sb);
1482 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1483 __u32 compat);
1484 extern int ext4_update_rocompat_feature(handle_t *handle,
1485 struct super_block *sb, __u32 rocompat);
1486 extern int ext4_update_incompat_feature(handle_t *handle,
1487 struct super_block *sb, __u32 incompat);
1488 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1489 struct ext4_group_desc *bg);
1490 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1491 struct ext4_group_desc *bg);
1492 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1493 struct ext4_group_desc *bg);
1494 extern __u32 ext4_free_blks_count(struct super_block *sb,
1495 struct ext4_group_desc *bg);
1496 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1497 struct ext4_group_desc *bg);
1498 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1499 struct ext4_group_desc *bg);
1500 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1501 struct ext4_group_desc *bg);
1502 extern void ext4_block_bitmap_set(struct super_block *sb,
1503 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1504 extern void ext4_inode_bitmap_set(struct super_block *sb,
1505 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1506 extern void ext4_inode_table_set(struct super_block *sb,
1507 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1508 extern void ext4_free_blks_set(struct super_block *sb,
1509 struct ext4_group_desc *bg, __u32 count);
1510 extern void ext4_free_inodes_set(struct super_block *sb,
1511 struct ext4_group_desc *bg, __u32 count);
1512 extern void ext4_used_dirs_set(struct super_block *sb,
1513 struct ext4_group_desc *bg, __u32 count);
1514 extern void ext4_itable_unused_set(struct super_block *sb,
1515 struct ext4_group_desc *bg, __u32 count);
1516 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1517 struct ext4_group_desc *gdp);
1518 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1519 struct ext4_group_desc *gdp);
1520
1521 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1522 {
1523 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1524 le32_to_cpu(es->s_blocks_count_lo);
1525 }
1526
1527 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1528 {
1529 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1530 le32_to_cpu(es->s_r_blocks_count_lo);
1531 }
1532
1533 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1534 {
1535 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1536 le32_to_cpu(es->s_free_blocks_count_lo);
1537 }
1538
1539 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1540 ext4_fsblk_t blk)
1541 {
1542 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1543 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1544 }
1545
1546 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1547 ext4_fsblk_t blk)
1548 {
1549 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1550 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1551 }
1552
1553 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1554 ext4_fsblk_t blk)
1555 {
1556 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1557 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1558 }
1559
1560 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1561 {
1562 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1563 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1564 le32_to_cpu(raw_inode->i_size_lo);
1565 else
1566 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1567 }
1568
1569 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1570 {
1571 raw_inode->i_size_lo = cpu_to_le32(i_size);
1572 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1573 }
1574
1575 static inline
1576 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1577 ext4_group_t group)
1578 {
1579 struct ext4_group_info ***grp_info;
1580 long indexv, indexh;
1581 grp_info = EXT4_SB(sb)->s_group_info;
1582 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1583 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1584 return grp_info[indexv][indexh];
1585 }
1586
1587 /*
1588 * Reading s_groups_count requires using smp_rmb() afterwards. See
1589 * the locking protocol documented in the comments of ext4_group_add()
1590 * in resize.c
1591 */
1592 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1593 {
1594 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
1595
1596 smp_rmb();
1597 return ngroups;
1598 }
1599
1600 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1601 ext4_group_t block_group)
1602 {
1603 return block_group >> sbi->s_log_groups_per_flex;
1604 }
1605
1606 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1607 {
1608 return 1 << sbi->s_log_groups_per_flex;
1609 }
1610
1611 #define ext4_std_error(sb, errno) \
1612 do { \
1613 if ((errno)) \
1614 __ext4_std_error((sb), __func__, (errno)); \
1615 } while (0)
1616
1617 #ifdef CONFIG_SMP
1618 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1619 * counters. So we need to make sure we have free blocks more
1620 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
1621 */
1622 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1623 #else
1624 #define EXT4_FREEBLOCKS_WATERMARK 0
1625 #endif
1626
1627 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1628 {
1629 /*
1630 * XXX: replace with spinlock if seen contended -bzzz
1631 */
1632 down_write(&EXT4_I(inode)->i_data_sem);
1633 if (newsize > EXT4_I(inode)->i_disksize)
1634 EXT4_I(inode)->i_disksize = newsize;
1635 up_write(&EXT4_I(inode)->i_data_sem);
1636 return ;
1637 }
1638
1639 struct ext4_group_info {
1640 unsigned long bb_state;
1641 struct rb_root bb_free_root;
1642 ext4_grpblk_t bb_first_free; /* first free block */
1643 ext4_grpblk_t bb_free; /* total free blocks */
1644 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
1645 struct list_head bb_prealloc_list;
1646 #ifdef DOUBLE_CHECK
1647 void *bb_bitmap;
1648 #endif
1649 struct rw_semaphore alloc_sem;
1650 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
1651 * regions, index is order.
1652 * bb_counters[3] = 5 means
1653 * 5 free 8-block regions. */
1654 };
1655
1656 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
1657
1658 #define EXT4_MB_GRP_NEED_INIT(grp) \
1659 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1660
1661 #define EXT4_MAX_CONTENTION 8
1662 #define EXT4_CONTENTION_THRESHOLD 2
1663
1664 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1665 ext4_group_t group)
1666 {
1667 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1668 }
1669
1670 /*
1671 * Returns true if the filesystem is busy enough that attempts to
1672 * access the block group locks has run into contention.
1673 */
1674 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1675 {
1676 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1677 }
1678
1679 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1680 {
1681 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1682 if (spin_trylock(lock))
1683 /*
1684 * We're able to grab the lock right away, so drop the
1685 * lock contention counter.
1686 */
1687 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1688 else {
1689 /*
1690 * The lock is busy, so bump the contention counter,
1691 * and then wait on the spin lock.
1692 */
1693 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1694 EXT4_MAX_CONTENTION);
1695 spin_lock(lock);
1696 }
1697 }
1698
1699 static inline void ext4_unlock_group(struct super_block *sb,
1700 ext4_group_t group)
1701 {
1702 spin_unlock(ext4_group_lock_ptr(sb, group));
1703 }
1704
1705 /*
1706 * Inodes and files operations
1707 */
1708
1709 /* dir.c */
1710 extern const struct file_operations ext4_dir_operations;
1711
1712 /* file.c */
1713 extern const struct inode_operations ext4_file_inode_operations;
1714 extern const struct file_operations ext4_file_operations;
1715
1716 /* namei.c */
1717 extern const struct inode_operations ext4_dir_inode_operations;
1718 extern const struct inode_operations ext4_special_inode_operations;
1719 extern struct dentry *ext4_get_parent(struct dentry *child);
1720
1721 /* symlink.c */
1722 extern const struct inode_operations ext4_symlink_inode_operations;
1723 extern const struct inode_operations ext4_fast_symlink_inode_operations;
1724
1725 /* block_validity */
1726 extern void ext4_release_system_zone(struct super_block *sb);
1727 extern int ext4_setup_system_zone(struct super_block *sb);
1728 extern int __init init_ext4_system_zone(void);
1729 extern void exit_ext4_system_zone(void);
1730 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1731 ext4_fsblk_t start_blk,
1732 unsigned int count);
1733
1734 /* extents.c */
1735 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1736 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1737 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1738 int chunk);
1739 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
1740 ext4_lblk_t iblock, unsigned int max_blocks,
1741 struct buffer_head *bh_result, int flags);
1742 extern void ext4_ext_truncate(struct inode *);
1743 extern void ext4_ext_init(struct super_block *);
1744 extern void ext4_ext_release(struct super_block *);
1745 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1746 loff_t len);
1747 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1748 ssize_t len);
1749 extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1750 sector_t block, unsigned int max_blocks,
1751 struct buffer_head *bh, int flags);
1752 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1753 __u64 start, __u64 len);
1754 /* move_extent.c */
1755 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1756 __u64 start_orig, __u64 start_donor,
1757 __u64 len, __u64 *moved_len);
1758
1759
1760 /*
1761 * Add new method to test wether block and inode bitmaps are properly
1762 * initialized. With uninit_bg reading the block from disk is not enough
1763 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1764 */
1765 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1766
1767 static inline int bitmap_uptodate(struct buffer_head *bh)
1768 {
1769 return (buffer_uptodate(bh) &&
1770 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1771 }
1772 static inline void set_bitmap_uptodate(struct buffer_head *bh)
1773 {
1774 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1775 }
1776
1777 #endif /* __KERNEL__ */
1778
1779 #endif /* _EXT4_H */
This page took 0.069684 seconds and 5 git commands to generate.