Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
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#include <crypto/hash.h>
33#ifdef __KERNEL__
34#include <linux/compat.h>
35#endif
36
37/*
38 * The fourth extended filesystem constants/structures
39 */
40
41/*
42 * Define EXT4FS_DEBUG to produce debug messages
43 */
44#undef EXT4FS_DEBUG
45
46/*
47 * Debug code
48 */
49#ifdef EXT4FS_DEBUG
50#define ext4_debug(f, a...) \
51 do { \
52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
53 __FILE__, __LINE__, __func__); \
54 printk(KERN_DEBUG f, ## a); \
55 } while (0)
56#else
57#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
58#endif
59
60/*
61 * Turn on EXT_DEBUG to get lots of info about extents operations.
62 */
63#define EXT_DEBUG__
64#ifdef EXT_DEBUG
65#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
66#else
67#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
68#endif
69
70#define EXT4_ERROR_INODE(inode, fmt, a...) \
71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
72
73#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
75
76#define EXT4_ERROR_FILE(file, block, fmt, a...) \
77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
78
79/* data type for block offset of block group */
80typedef int ext4_grpblk_t;
81
82/* data type for filesystem-wide blocks number */
83typedef unsigned long long ext4_fsblk_t;
84
85/* data type for file logical block number */
86typedef __u32 ext4_lblk_t;
87
88/* data type for block group number */
89typedef unsigned int ext4_group_t;
90
91/*
92 * Flags used in mballoc's allocation_context flags field.
93 *
94 * Also used to show what's going on for debugging purposes when the
95 * flag field is exported via the traceport interface
96 */
97
98/* prefer goal again. length */
99#define EXT4_MB_HINT_MERGE 0x0001
100/* blocks already reserved */
101#define EXT4_MB_HINT_RESERVED 0x0002
102/* metadata is being allocated */
103#define EXT4_MB_HINT_METADATA 0x0004
104/* first blocks in the file */
105#define EXT4_MB_HINT_FIRST 0x0008
106/* search for the best chunk */
107#define EXT4_MB_HINT_BEST 0x0010
108/* data is being allocated */
109#define EXT4_MB_HINT_DATA 0x0020
110/* don't preallocate (for tails) */
111#define EXT4_MB_HINT_NOPREALLOC 0x0040
112/* allocate for locality group */
113#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
114/* allocate goal blocks or none */
115#define EXT4_MB_HINT_GOAL_ONLY 0x0100
116/* goal is meaningful */
117#define EXT4_MB_HINT_TRY_GOAL 0x0200
118/* blocks already pre-reserved by delayed allocation */
119#define EXT4_MB_DELALLOC_RESERVED 0x0400
120/* We are doing stream allocation */
121#define EXT4_MB_STREAM_ALLOC 0x0800
122/* Use reserved root blocks if needed */
123#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
124
125struct ext4_allocation_request {
126 /* target inode for block we're allocating */
127 struct inode *inode;
128 /* how many blocks we want to allocate */
129 unsigned int len;
130 /* logical block in target inode */
131 ext4_lblk_t logical;
132 /* the closest logical allocated block to the left */
133 ext4_lblk_t lleft;
134 /* the closest logical allocated block to the right */
135 ext4_lblk_t lright;
136 /* phys. target (a hint) */
137 ext4_fsblk_t goal;
138 /* phys. block for the closest logical allocated block to the left */
139 ext4_fsblk_t pleft;
140 /* phys. block for the closest logical allocated block to the right */
141 ext4_fsblk_t pright;
142 /* flags. see above EXT4_MB_HINT_* */
143 unsigned int flags;
144};
145
146/*
147 * Logical to physical block mapping, used by ext4_map_blocks()
148 *
149 * This structure is used to pass requests into ext4_map_blocks() as
150 * well as to store the information returned by ext4_map_blocks(). It
151 * takes less room on the stack than a struct buffer_head.
152 */
153#define EXT4_MAP_NEW (1 << BH_New)
154#define EXT4_MAP_MAPPED (1 << BH_Mapped)
155#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
156#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
157#define EXT4_MAP_UNINIT (1 << BH_Uninit)
158/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
159 * ext4_map_blocks wants to know whether or not the underlying cluster has
160 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
161 * the requested mapping was from previously mapped (or delayed allocated)
162 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
163 * should never appear on buffer_head's state flags.
164 */
165#define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
166#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
167 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
168 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
169
170struct ext4_map_blocks {
171 ext4_fsblk_t m_pblk;
172 ext4_lblk_t m_lblk;
173 unsigned int m_len;
174 unsigned int m_flags;
175};
176
177/*
178 * For delayed allocation tracking
179 */
180struct mpage_da_data {
181 struct inode *inode;
182 sector_t b_blocknr; /* start block number of extent */
183 size_t b_size; /* size of extent */
184 unsigned long b_state; /* state of the extent */
185 unsigned long first_page, next_page; /* extent of pages */
186 struct writeback_control *wbc;
187 int io_done;
188 int pages_written;
189 int retval;
190};
191
192/*
193 * Flags for ext4_io_end->flags
194 */
195#define EXT4_IO_END_UNWRITTEN 0x0001
196#define EXT4_IO_END_ERROR 0x0002
197#define EXT4_IO_END_DIRECT 0x0004
198
199struct ext4_io_page {
200 struct page *p_page;
201 atomic_t p_count;
202};
203
204#define MAX_IO_PAGES 128
205
206/*
207 * For converting uninitialized extents on a work queue.
208 *
209 * 'page' is only used from the writepage() path; 'pages' is only used for
210 * buffered writes; they are used to keep page references until conversion
211 * takes place. For AIO/DIO, neither field is filled in.
212 */
213typedef struct ext4_io_end {
214 struct list_head list; /* per-file finished IO list */
215 struct inode *inode; /* file being written to */
216 unsigned int flag; /* unwritten or not */
217 loff_t offset; /* offset in the file */
218 ssize_t size; /* size of the extent */
219 struct kiocb *iocb; /* iocb struct for AIO */
220 int result; /* error value for AIO */
221 int num_io_pages; /* for writepages() */
222 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
223} ext4_io_end_t;
224
225struct ext4_io_submit {
226 int io_op;
227 struct bio *io_bio;
228 ext4_io_end_t *io_end;
229 struct ext4_io_page *io_page;
230 sector_t io_next_block;
231};
232
233/*
234 * Special inodes numbers
235 */
236#define EXT4_BAD_INO 1 /* Bad blocks inode */
237#define EXT4_ROOT_INO 2 /* Root inode */
238#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
239#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
240#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
241#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
242#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
243#define EXT4_JOURNAL_INO 8 /* Journal inode */
244
245/* First non-reserved inode for old ext4 filesystems */
246#define EXT4_GOOD_OLD_FIRST_INO 11
247
248/*
249 * Maximal count of links to a file
250 */
251#define EXT4_LINK_MAX 65000
252
253/*
254 * Macro-instructions used to manage several block sizes
255 */
256#define EXT4_MIN_BLOCK_SIZE 1024
257#define EXT4_MAX_BLOCK_SIZE 65536
258#define EXT4_MIN_BLOCK_LOG_SIZE 10
259#define EXT4_MAX_BLOCK_LOG_SIZE 16
260#ifdef __KERNEL__
261# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
262#else
263# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
264#endif
265#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
266#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
267 EXT4_SB(s)->s_cluster_bits)
268#ifdef __KERNEL__
269# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
270# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
271#else
272# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
273#endif
274#ifdef __KERNEL__
275#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
276#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
277#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
278#else
279#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
280 EXT4_GOOD_OLD_INODE_SIZE : \
281 (s)->s_inode_size)
282#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
283 EXT4_GOOD_OLD_FIRST_INO : \
284 (s)->s_first_ino)
285#endif
286#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
287
288/* Translate a block number to a cluster number */
289#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
290/* Translate a cluster number to a block number */
291#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
292/* Translate # of blks to # of clusters */
293#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
294 (sbi)->s_cluster_bits)
295
296/*
297 * Structure of a blocks group descriptor
298 */
299struct ext4_group_desc
300{
301 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
302 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
303 __le32 bg_inode_table_lo; /* Inodes table block */
304 __le16 bg_free_blocks_count_lo;/* Free blocks count */
305 __le16 bg_free_inodes_count_lo;/* Free inodes count */
306 __le16 bg_used_dirs_count_lo; /* Directories count */
307 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
308 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
309 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
310 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
311 __le16 bg_itable_unused_lo; /* Unused inodes count */
312 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
313 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
314 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
315 __le32 bg_inode_table_hi; /* Inodes table block MSB */
316 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
317 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
318 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
319 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
320 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
321 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
322 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
323 __u32 bg_reserved;
324};
325
326#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
328 sizeof(__le16))
329#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
330 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
331 sizeof(__le16))
332
333/*
334 * Structure of a flex block group info
335 */
336
337struct flex_groups {
338 atomic_t free_inodes;
339 atomic_t free_clusters;
340 atomic_t used_dirs;
341};
342
343#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
344#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
345#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
346
347/*
348 * Macro-instructions used to manage group descriptors
349 */
350#define EXT4_MIN_DESC_SIZE 32
351#define EXT4_MIN_DESC_SIZE_64BIT 64
352#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
353#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
354#ifdef __KERNEL__
355# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
356# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
357# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
358# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
359# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
360#else
361# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
362# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
363# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
364#endif
365
366/*
367 * Constants relative to the data blocks
368 */
369#define EXT4_NDIR_BLOCKS 12
370#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
371#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
372#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
373#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
374
375/*
376 * Inode flags
377 */
378#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
379#define EXT4_UNRM_FL 0x00000002 /* Undelete */
380#define EXT4_COMPR_FL 0x00000004 /* Compress file */
381#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
382#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
383#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
384#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
385#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
386/* Reserved for compression usage... */
387#define EXT4_DIRTY_FL 0x00000100
388#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
389#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
390#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
391/* End compression flags --- maybe not all used */
392#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
393#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
394#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
395#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
396#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
397#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
398#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
399#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
400#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
401#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
402#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
403#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
404
405#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
406#define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
407
408/* Flags that should be inherited by new inodes from their parent. */
409#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
410 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
411 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
412 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
413
414/* Flags that are appropriate for regular files (all but dir-specific ones). */
415#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
416
417/* Flags that are appropriate for non-directories/regular files. */
418#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
419
420/* Mask out flags that are inappropriate for the given type of inode. */
421static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
422{
423 if (S_ISDIR(mode))
424 return flags;
425 else if (S_ISREG(mode))
426 return flags & EXT4_REG_FLMASK;
427 else
428 return flags & EXT4_OTHER_FLMASK;
429}
430
431/*
432 * Inode flags used for atomic set/get
433 */
434enum {
435 EXT4_INODE_SECRM = 0, /* Secure deletion */
436 EXT4_INODE_UNRM = 1, /* Undelete */
437 EXT4_INODE_COMPR = 2, /* Compress file */
438 EXT4_INODE_SYNC = 3, /* Synchronous updates */
439 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
440 EXT4_INODE_APPEND = 5, /* writes to file may only append */
441 EXT4_INODE_NODUMP = 6, /* do not dump file */
442 EXT4_INODE_NOATIME = 7, /* do not update atime */
443/* Reserved for compression usage... */
444 EXT4_INODE_DIRTY = 8,
445 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
446 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
447 EXT4_INODE_ECOMPR = 11, /* Compression error */
448/* End compression flags --- maybe not all used */
449 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
450 EXT4_INODE_IMAGIC = 13, /* AFS directory */
451 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
452 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
453 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
454 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
455 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
456 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
457 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
458 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
459 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
460 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
461};
462
463/*
464 * Since it's pretty easy to mix up bit numbers and hex values, we use a
465 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
466 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
467 * any extra space in the compiled kernel image, otherwise, the build will fail.
468 * It's important that these values are the same, since we are using
469 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
470 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
471 * values found in ext2, ext3 and ext4 filesystems, and of course the values
472 * defined in e2fsprogs.
473 *
474 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
475 */
476#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
477#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
478
479static inline void ext4_check_flag_values(void)
480{
481 CHECK_FLAG_VALUE(SECRM);
482 CHECK_FLAG_VALUE(UNRM);
483 CHECK_FLAG_VALUE(COMPR);
484 CHECK_FLAG_VALUE(SYNC);
485 CHECK_FLAG_VALUE(IMMUTABLE);
486 CHECK_FLAG_VALUE(APPEND);
487 CHECK_FLAG_VALUE(NODUMP);
488 CHECK_FLAG_VALUE(NOATIME);
489 CHECK_FLAG_VALUE(DIRTY);
490 CHECK_FLAG_VALUE(COMPRBLK);
491 CHECK_FLAG_VALUE(NOCOMPR);
492 CHECK_FLAG_VALUE(ECOMPR);
493 CHECK_FLAG_VALUE(INDEX);
494 CHECK_FLAG_VALUE(IMAGIC);
495 CHECK_FLAG_VALUE(JOURNAL_DATA);
496 CHECK_FLAG_VALUE(NOTAIL);
497 CHECK_FLAG_VALUE(DIRSYNC);
498 CHECK_FLAG_VALUE(TOPDIR);
499 CHECK_FLAG_VALUE(HUGE_FILE);
500 CHECK_FLAG_VALUE(EXTENTS);
501 CHECK_FLAG_VALUE(EA_INODE);
502 CHECK_FLAG_VALUE(EOFBLOCKS);
503 CHECK_FLAG_VALUE(INLINE_DATA);
504 CHECK_FLAG_VALUE(RESERVED);
505}
506
507/* Used to pass group descriptor data when online resize is done */
508struct ext4_new_group_input {
509 __u32 group; /* Group number for this data */
510 __u64 block_bitmap; /* Absolute block number of block bitmap */
511 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
512 __u64 inode_table; /* Absolute block number of inode table start */
513 __u32 blocks_count; /* Total number of blocks in this group */
514 __u16 reserved_blocks; /* Number of reserved blocks in this group */
515 __u16 unused;
516};
517
518#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
519struct compat_ext4_new_group_input {
520 u32 group;
521 compat_u64 block_bitmap;
522 compat_u64 inode_bitmap;
523 compat_u64 inode_table;
524 u32 blocks_count;
525 u16 reserved_blocks;
526 u16 unused;
527};
528#endif
529
530/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
531struct ext4_new_group_data {
532 __u32 group;
533 __u64 block_bitmap;
534 __u64 inode_bitmap;
535 __u64 inode_table;
536 __u32 blocks_count;
537 __u16 reserved_blocks;
538 __u16 unused;
539 __u32 free_blocks_count;
540};
541
542/* Indexes used to index group tables in ext4_new_group_data */
543enum {
544 BLOCK_BITMAP = 0, /* block bitmap */
545 INODE_BITMAP, /* inode bitmap */
546 INODE_TABLE, /* inode tables */
547 GROUP_TABLE_COUNT,
548};
549
550/*
551 * Flags used by ext4_map_blocks()
552 */
553 /* Allocate any needed blocks and/or convert an unitialized
554 extent to be an initialized ext4 */
555#define EXT4_GET_BLOCKS_CREATE 0x0001
556 /* Request the creation of an unitialized extent */
557#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
558#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
559 EXT4_GET_BLOCKS_CREATE)
560 /* Caller is from the delayed allocation writeout path,
561 so set the magic i_delalloc_reserve_flag after taking the
562 inode allocation semaphore for */
563#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
564 /* caller is from the direct IO path, request to creation of an
565 unitialized extents if not allocated, split the uninitialized
566 extent if blocks has been preallocated already*/
567#define EXT4_GET_BLOCKS_PRE_IO 0x0008
568#define EXT4_GET_BLOCKS_CONVERT 0x0010
569#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
570 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
571 /* Convert extent to initialized after IO complete */
572#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
573 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
574 /* Punch out blocks of an extent */
575#define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
576 /* Don't normalize allocation size (used for fallocate) */
577#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
578 /* Request will not result in inode size update (user for fallocate) */
579#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
580 /* Do not take i_data_sem locking in ext4_map_blocks */
581#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
582 /* Do not put hole in extent cache */
583#define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
584
585/*
586 * Flags used by ext4_free_blocks
587 */
588#define EXT4_FREE_BLOCKS_METADATA 0x0001
589#define EXT4_FREE_BLOCKS_FORGET 0x0002
590#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
591#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
592#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
593#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
594
595/*
596 * Flags used by ext4_discard_partial_page_buffers
597 */
598#define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
599
600/*
601 * ioctl commands
602 */
603#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
604#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
605#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
606#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
607#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
608#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
609#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
610#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
611#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
612#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
613#define EXT4_IOC_MIGRATE _IO('f', 9)
614 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
615 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
616#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
617#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
618#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
619
620#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
621/*
622 * ioctl commands in 32 bit emulation
623 */
624#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
625#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
626#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
627#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
628#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
629#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
630#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
631#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
632#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
633#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
634#endif
635
636/* Max physical block we can address w/o extents */
637#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
638
639/*
640 * Structure of an inode on the disk
641 */
642struct ext4_inode {
643 __le16 i_mode; /* File mode */
644 __le16 i_uid; /* Low 16 bits of Owner Uid */
645 __le32 i_size_lo; /* Size in bytes */
646 __le32 i_atime; /* Access time */
647 __le32 i_ctime; /* Inode Change time */
648 __le32 i_mtime; /* Modification time */
649 __le32 i_dtime; /* Deletion Time */
650 __le16 i_gid; /* Low 16 bits of Group Id */
651 __le16 i_links_count; /* Links count */
652 __le32 i_blocks_lo; /* Blocks count */
653 __le32 i_flags; /* File flags */
654 union {
655 struct {
656 __le32 l_i_version;
657 } linux1;
658 struct {
659 __u32 h_i_translator;
660 } hurd1;
661 struct {
662 __u32 m_i_reserved1;
663 } masix1;
664 } osd1; /* OS dependent 1 */
665 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
666 __le32 i_generation; /* File version (for NFS) */
667 __le32 i_file_acl_lo; /* File ACL */
668 __le32 i_size_high;
669 __le32 i_obso_faddr; /* Obsoleted fragment address */
670 union {
671 struct {
672 __le16 l_i_blocks_high; /* were l_i_reserved1 */
673 __le16 l_i_file_acl_high;
674 __le16 l_i_uid_high; /* these 2 fields */
675 __le16 l_i_gid_high; /* were reserved2[0] */
676 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
677 __le16 l_i_reserved;
678 } linux2;
679 struct {
680 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
681 __u16 h_i_mode_high;
682 __u16 h_i_uid_high;
683 __u16 h_i_gid_high;
684 __u32 h_i_author;
685 } hurd2;
686 struct {
687 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
688 __le16 m_i_file_acl_high;
689 __u32 m_i_reserved2[2];
690 } masix2;
691 } osd2; /* OS dependent 2 */
692 __le16 i_extra_isize;
693 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
694 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
695 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
696 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
697 __le32 i_crtime; /* File Creation time */
698 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
699 __le32 i_version_hi; /* high 32 bits for 64-bit version */
700};
701
702struct move_extent {
703 __u32 reserved; /* should be zero */
704 __u32 donor_fd; /* donor file descriptor */
705 __u64 orig_start; /* logical start offset in block for orig */
706 __u64 donor_start; /* logical start offset in block for donor */
707 __u64 len; /* block length to be moved */
708 __u64 moved_len; /* moved block length */
709};
710
711#define EXT4_EPOCH_BITS 2
712#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
713#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
714
715/*
716 * Extended fields will fit into an inode if the filesystem was formatted
717 * with large inodes (-I 256 or larger) and there are not currently any EAs
718 * consuming all of the available space. For new inodes we always reserve
719 * enough space for the kernel's known extended fields, but for inodes
720 * created with an old kernel this might not have been the case. None of
721 * the extended inode fields is critical for correct filesystem operation.
722 * This macro checks if a certain field fits in the inode. Note that
723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
724 */
725#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
726 ((offsetof(typeof(*ext4_inode), field) + \
727 sizeof((ext4_inode)->field)) \
728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
729 (einode)->i_extra_isize)) \
730
731static inline __le32 ext4_encode_extra_time(struct timespec *time)
732{
733 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
734 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
735 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
736}
737
738static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
739{
740 if (sizeof(time->tv_sec) > 4)
741 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
742 << 32;
743 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
744}
745
746#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
747do { \
748 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
749 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
750 (raw_inode)->xtime ## _extra = \
751 ext4_encode_extra_time(&(inode)->xtime); \
752} while (0)
753
754#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
755do { \
756 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
757 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
758 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
759 (raw_inode)->xtime ## _extra = \
760 ext4_encode_extra_time(&(einode)->xtime); \
761} while (0)
762
763#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
764do { \
765 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
766 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
767 ext4_decode_extra_time(&(inode)->xtime, \
768 raw_inode->xtime ## _extra); \
769 else \
770 (inode)->xtime.tv_nsec = 0; \
771} while (0)
772
773#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
774do { \
775 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
776 (einode)->xtime.tv_sec = \
777 (signed)le32_to_cpu((raw_inode)->xtime); \
778 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
779 ext4_decode_extra_time(&(einode)->xtime, \
780 raw_inode->xtime ## _extra); \
781 else \
782 (einode)->xtime.tv_nsec = 0; \
783} while (0)
784
785#define i_disk_version osd1.linux1.l_i_version
786
787#if defined(__KERNEL__) || defined(__linux__)
788#define i_reserved1 osd1.linux1.l_i_reserved1
789#define i_file_acl_high osd2.linux2.l_i_file_acl_high
790#define i_blocks_high osd2.linux2.l_i_blocks_high
791#define i_uid_low i_uid
792#define i_gid_low i_gid
793#define i_uid_high osd2.linux2.l_i_uid_high
794#define i_gid_high osd2.linux2.l_i_gid_high
795#define i_checksum_lo osd2.linux2.l_i_checksum_lo
796
797#elif defined(__GNU__)
798
799#define i_translator osd1.hurd1.h_i_translator
800#define i_uid_high osd2.hurd2.h_i_uid_high
801#define i_gid_high osd2.hurd2.h_i_gid_high
802#define i_author osd2.hurd2.h_i_author
803
804#elif defined(__masix__)
805
806#define i_reserved1 osd1.masix1.m_i_reserved1
807#define i_file_acl_high osd2.masix2.m_i_file_acl_high
808#define i_reserved2 osd2.masix2.m_i_reserved2
809
810#endif /* defined(__KERNEL__) || defined(__linux__) */
811
812#include "extents_status.h"
813
814/*
815 * fourth extended file system inode data in memory
816 */
817struct ext4_inode_info {
818 __le32 i_data[15]; /* unconverted */
819 __u32 i_dtime;
820 ext4_fsblk_t i_file_acl;
821
822 /*
823 * i_block_group is the number of the block group which contains
824 * this file's inode. Constant across the lifetime of the inode,
825 * it is ued for making block allocation decisions - we try to
826 * place a file's data blocks near its inode block, and new inodes
827 * near to their parent directory's inode.
828 */
829 ext4_group_t i_block_group;
830 ext4_lblk_t i_dir_start_lookup;
831#if (BITS_PER_LONG < 64)
832 unsigned long i_state_flags; /* Dynamic state flags */
833#endif
834 unsigned long i_flags;
835
836 /*
837 * Extended attributes can be read independently of the main file
838 * data. Taking i_mutex even when reading would cause contention
839 * between readers of EAs and writers of regular file data, so
840 * instead we synchronize on xattr_sem when reading or changing
841 * EAs.
842 */
843 struct rw_semaphore xattr_sem;
844
845 struct list_head i_orphan; /* unlinked but open inodes */
846
847 /*
848 * i_disksize keeps track of what the inode size is ON DISK, not
849 * in memory. During truncate, i_size is set to the new size by
850 * the VFS prior to calling ext4_truncate(), but the filesystem won't
851 * set i_disksize to 0 until the truncate is actually under way.
852 *
853 * The intent is that i_disksize always represents the blocks which
854 * are used by this file. This allows recovery to restart truncate
855 * on orphans if we crash during truncate. We actually write i_disksize
856 * into the on-disk inode when writing inodes out, instead of i_size.
857 *
858 * The only time when i_disksize and i_size may be different is when
859 * a truncate is in progress. The only things which change i_disksize
860 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
861 */
862 loff_t i_disksize;
863
864 /*
865 * i_data_sem is for serialising ext4_truncate() against
866 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
867 * data tree are chopped off during truncate. We can't do that in
868 * ext4 because whenever we perform intermediate commits during
869 * truncate, the inode and all the metadata blocks *must* be in a
870 * consistent state which allows truncation of the orphans to restart
871 * during recovery. Hence we must fix the get_block-vs-truncate race
872 * by other means, so we have i_data_sem.
873 */
874 struct rw_semaphore i_data_sem;
875 struct inode vfs_inode;
876 struct jbd2_inode *jinode;
877
878 /*
879 * File creation time. Its function is same as that of
880 * struct timespec i_{a,c,m}time in the generic inode.
881 */
882 struct timespec i_crtime;
883
884 /* mballoc */
885 struct list_head i_prealloc_list;
886 spinlock_t i_prealloc_lock;
887
888 /* extents status tree */
889 struct ext4_es_tree i_es_tree;
890 rwlock_t i_es_lock;
891 struct list_head i_es_lru;
892 unsigned int i_es_lru_nr; /* protected by i_es_lock */
893
894 /* ialloc */
895 ext4_group_t i_last_alloc_group;
896
897 /* allocation reservation info for delalloc */
898 /* In case of bigalloc, these refer to clusters rather than blocks */
899 unsigned int i_reserved_data_blocks;
900 unsigned int i_reserved_meta_blocks;
901 unsigned int i_allocated_meta_blocks;
902 ext4_lblk_t i_da_metadata_calc_last_lblock;
903 int i_da_metadata_calc_len;
904
905 /* on-disk additional length */
906 __u16 i_extra_isize;
907
908 /* Indicate the inline data space. */
909 u16 i_inline_off;
910 u16 i_inline_size;
911
912#ifdef CONFIG_QUOTA
913 /* quota space reservation, managed internally by quota code */
914 qsize_t i_reserved_quota;
915#endif
916
917 /* completed IOs that might need unwritten extents handling */
918 struct list_head i_completed_io_list;
919 spinlock_t i_completed_io_lock;
920 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
921 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
922 struct work_struct i_unwritten_work; /* deferred extent conversion */
923
924 spinlock_t i_block_reservation_lock;
925
926 /*
927 * Transactions that contain inode's metadata needed to complete
928 * fsync and fdatasync, respectively.
929 */
930 tid_t i_sync_tid;
931 tid_t i_datasync_tid;
932
933 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
934 __u32 i_csum_seed;
935};
936
937/*
938 * File system states
939 */
940#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
941#define EXT4_ERROR_FS 0x0002 /* Errors detected */
942#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
943
944/*
945 * Misc. filesystem flags
946 */
947#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
948#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
949#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
950
951/*
952 * Mount flags
953 */
954#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
955#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
956#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
957#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
958#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
959#define EXT4_MOUNT_ERRORS_MASK 0x00070
960#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
961#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
962#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
963#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
964#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
965#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
966#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
967#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
968#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
969#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
970#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
971#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
972#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
973#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
974#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
975#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
976#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
977#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
978#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
979#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
980#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
981#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
982#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
983
984#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
985 specified delalloc */
986
987#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
988 ~EXT4_MOUNT_##opt
989#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
990 EXT4_MOUNT_##opt
991#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
992 EXT4_MOUNT_##opt)
993
994#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
995 ~EXT4_MOUNT2_##opt
996#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
997 EXT4_MOUNT2_##opt
998#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
999 EXT4_MOUNT2_##opt)
1000
1001#define ext4_test_and_set_bit __test_and_set_bit_le
1002#define ext4_set_bit __set_bit_le
1003#define ext4_set_bit_atomic ext2_set_bit_atomic
1004#define ext4_test_and_clear_bit __test_and_clear_bit_le
1005#define ext4_clear_bit __clear_bit_le
1006#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1007#define ext4_test_bit test_bit_le
1008#define ext4_find_next_zero_bit find_next_zero_bit_le
1009#define ext4_find_next_bit find_next_bit_le
1010
1011extern void ext4_set_bits(void *bm, int cur, int len);
1012
1013/*
1014 * Maximal mount counts between two filesystem checks
1015 */
1016#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1017#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1018
1019/*
1020 * Behaviour when detecting errors
1021 */
1022#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1023#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1024#define EXT4_ERRORS_PANIC 3 /* Panic */
1025#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1026
1027/* Metadata checksum algorithm codes */
1028#define EXT4_CRC32C_CHKSUM 1
1029
1030/*
1031 * Structure of the super block
1032 */
1033struct ext4_super_block {
1034/*00*/ __le32 s_inodes_count; /* Inodes count */
1035 __le32 s_blocks_count_lo; /* Blocks count */
1036 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1037 __le32 s_free_blocks_count_lo; /* Free blocks count */
1038/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1039 __le32 s_first_data_block; /* First Data Block */
1040 __le32 s_log_block_size; /* Block size */
1041 __le32 s_log_cluster_size; /* Allocation cluster size */
1042/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1043 __le32 s_clusters_per_group; /* # Clusters per group */
1044 __le32 s_inodes_per_group; /* # Inodes per group */
1045 __le32 s_mtime; /* Mount time */
1046/*30*/ __le32 s_wtime; /* Write time */
1047 __le16 s_mnt_count; /* Mount count */
1048 __le16 s_max_mnt_count; /* Maximal mount count */
1049 __le16 s_magic; /* Magic signature */
1050 __le16 s_state; /* File system state */
1051 __le16 s_errors; /* Behaviour when detecting errors */
1052 __le16 s_minor_rev_level; /* minor revision level */
1053/*40*/ __le32 s_lastcheck; /* time of last check */
1054 __le32 s_checkinterval; /* max. time between checks */
1055 __le32 s_creator_os; /* OS */
1056 __le32 s_rev_level; /* Revision level */
1057/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1058 __le16 s_def_resgid; /* Default gid for reserved blocks */
1059 /*
1060 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1061 *
1062 * Note: the difference between the compatible feature set and
1063 * the incompatible feature set is that if there is a bit set
1064 * in the incompatible feature set that the kernel doesn't
1065 * know about, it should refuse to mount the filesystem.
1066 *
1067 * e2fsck's requirements are more strict; if it doesn't know
1068 * about a feature in either the compatible or incompatible
1069 * feature set, it must abort and not try to meddle with
1070 * things it doesn't understand...
1071 */
1072 __le32 s_first_ino; /* First non-reserved inode */
1073 __le16 s_inode_size; /* size of inode structure */
1074 __le16 s_block_group_nr; /* block group # of this superblock */
1075 __le32 s_feature_compat; /* compatible feature set */
1076/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1077 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1078/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1079/*78*/ char s_volume_name[16]; /* volume name */
1080/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1081/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1082 /*
1083 * Performance hints. Directory preallocation should only
1084 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1085 */
1086 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1087 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1088 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1089 /*
1090 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1091 */
1092/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1093/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1094 __le32 s_journal_dev; /* device number of journal file */
1095 __le32 s_last_orphan; /* start of list of inodes to delete */
1096 __le32 s_hash_seed[4]; /* HTREE hash seed */
1097 __u8 s_def_hash_version; /* Default hash version to use */
1098 __u8 s_jnl_backup_type;
1099 __le16 s_desc_size; /* size of group descriptor */
1100/*100*/ __le32 s_default_mount_opts;
1101 __le32 s_first_meta_bg; /* First metablock block group */
1102 __le32 s_mkfs_time; /* When the filesystem was created */
1103 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1104 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1105/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1106 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1107 __le32 s_free_blocks_count_hi; /* Free blocks count */
1108 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1109 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1110 __le32 s_flags; /* Miscellaneous flags */
1111 __le16 s_raid_stride; /* RAID stride */
1112 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1113 __le64 s_mmp_block; /* Block for multi-mount protection */
1114 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1115 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1116 __u8 s_checksum_type; /* metadata checksum algorithm used */
1117 __le16 s_reserved_pad;
1118 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1119 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1120 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1121 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1122 snapshot's future use */
1123 __le32 s_snapshot_list; /* inode number of the head of the
1124 on-disk snapshot list */
1125#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1126 __le32 s_error_count; /* number of fs errors */
1127 __le32 s_first_error_time; /* first time an error happened */
1128 __le32 s_first_error_ino; /* inode involved in first error */
1129 __le64 s_first_error_block; /* block involved of first error */
1130 __u8 s_first_error_func[32]; /* function where the error happened */
1131 __le32 s_first_error_line; /* line number where error happened */
1132 __le32 s_last_error_time; /* most recent time of an error */
1133 __le32 s_last_error_ino; /* inode involved in last error */
1134 __le32 s_last_error_line; /* line number where error happened */
1135 __le64 s_last_error_block; /* block involved of last error */
1136 __u8 s_last_error_func[32]; /* function where the error happened */
1137#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1138 __u8 s_mount_opts[64];
1139 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1140 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1141 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1142 __le32 s_reserved[108]; /* Padding to the end of the block */
1143 __le32 s_checksum; /* crc32c(superblock) */
1144};
1145
1146#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1147
1148#ifdef __KERNEL__
1149
1150/*
1151 * run-time mount flags
1152 */
1153#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1154#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1155
1156/*
1157 * fourth extended-fs super-block data in memory
1158 */
1159struct ext4_sb_info {
1160 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1161 unsigned long s_inodes_per_block;/* Number of inodes per block */
1162 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1163 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1164 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1165 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1166 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1167 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1168 ext4_group_t s_groups_count; /* Number of groups in the fs */
1169 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1170 unsigned long s_overhead; /* # of fs overhead clusters */
1171 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1172 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1173 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1174 struct buffer_head * s_sbh; /* Buffer containing the super block */
1175 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1176 struct buffer_head **s_group_desc;
1177 unsigned int s_mount_opt;
1178 unsigned int s_mount_opt2;
1179 unsigned int s_mount_flags;
1180 unsigned int s_def_mount_opt;
1181 ext4_fsblk_t s_sb_block;
1182 kuid_t s_resuid;
1183 kgid_t s_resgid;
1184 unsigned short s_mount_state;
1185 unsigned short s_pad;
1186 int s_addr_per_block_bits;
1187 int s_desc_per_block_bits;
1188 int s_inode_size;
1189 int s_first_ino;
1190 unsigned int s_inode_readahead_blks;
1191 unsigned int s_inode_goal;
1192 spinlock_t s_next_gen_lock;
1193 u32 s_next_generation;
1194 u32 s_hash_seed[4];
1195 int s_def_hash_version;
1196 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1197 struct percpu_counter s_freeclusters_counter;
1198 struct percpu_counter s_freeinodes_counter;
1199 struct percpu_counter s_dirs_counter;
1200 struct percpu_counter s_dirtyclusters_counter;
1201 struct blockgroup_lock *s_blockgroup_lock;
1202 struct proc_dir_entry *s_proc;
1203 struct kobject s_kobj;
1204 struct completion s_kobj_unregister;
1205 struct super_block *s_sb;
1206
1207 /* Journaling */
1208 struct journal_s *s_journal;
1209 struct list_head s_orphan;
1210 struct mutex s_orphan_lock;
1211 unsigned long s_resize_flags; /* Flags indicating if there
1212 is a resizer */
1213 unsigned long s_commit_interval;
1214 u32 s_max_batch_time;
1215 u32 s_min_batch_time;
1216 struct block_device *journal_bdev;
1217#ifdef CONFIG_QUOTA
1218 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1219 int s_jquota_fmt; /* Format of quota to use */
1220#endif
1221 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1222 struct rb_root system_blks;
1223
1224#ifdef EXTENTS_STATS
1225 /* ext4 extents stats */
1226 unsigned long s_ext_min;
1227 unsigned long s_ext_max;
1228 unsigned long s_depth_max;
1229 spinlock_t s_ext_stats_lock;
1230 unsigned long s_ext_blocks;
1231 unsigned long s_ext_extents;
1232#endif
1233
1234 /* for buddy allocator */
1235 struct ext4_group_info ***s_group_info;
1236 struct inode *s_buddy_cache;
1237 spinlock_t s_md_lock;
1238 unsigned short *s_mb_offsets;
1239 unsigned int *s_mb_maxs;
1240 unsigned int s_group_info_size;
1241
1242 /* tunables */
1243 unsigned long s_stripe;
1244 unsigned int s_mb_stream_request;
1245 unsigned int s_mb_max_to_scan;
1246 unsigned int s_mb_min_to_scan;
1247 unsigned int s_mb_stats;
1248 unsigned int s_mb_order2_reqs;
1249 unsigned int s_mb_group_prealloc;
1250 unsigned int s_max_writeback_mb_bump;
1251 unsigned int s_max_dir_size_kb;
1252 /* where last allocation was done - for stream allocation */
1253 unsigned long s_mb_last_group;
1254 unsigned long s_mb_last_start;
1255
1256 /* stats for buddy allocator */
1257 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1258 atomic_t s_bal_success; /* we found long enough chunks */
1259 atomic_t s_bal_allocated; /* in blocks */
1260 atomic_t s_bal_ex_scanned; /* total extents scanned */
1261 atomic_t s_bal_goals; /* goal hits */
1262 atomic_t s_bal_breaks; /* too long searches */
1263 atomic_t s_bal_2orders; /* 2^order hits */
1264 spinlock_t s_bal_lock;
1265 unsigned long s_mb_buddies_generated;
1266 unsigned long long s_mb_generation_time;
1267 atomic_t s_mb_lost_chunks;
1268 atomic_t s_mb_preallocated;
1269 atomic_t s_mb_discarded;
1270 atomic_t s_lock_busy;
1271
1272 /* locality groups */
1273 struct ext4_locality_group __percpu *s_locality_groups;
1274
1275 /* for write statistics */
1276 unsigned long s_sectors_written_start;
1277 u64 s_kbytes_written;
1278
1279 /* the size of zero-out chunk */
1280 unsigned int s_extent_max_zeroout_kb;
1281
1282 unsigned int s_log_groups_per_flex;
1283 struct flex_groups *s_flex_groups;
1284 ext4_group_t s_flex_groups_allocated;
1285
1286 /* workqueue for dio unwritten */
1287 struct workqueue_struct *dio_unwritten_wq;
1288
1289 /* timer for periodic error stats printing */
1290 struct timer_list s_err_report;
1291
1292 /* Lazy inode table initialization info */
1293 struct ext4_li_request *s_li_request;
1294 /* Wait multiplier for lazy initialization thread */
1295 unsigned int s_li_wait_mult;
1296
1297 /* Kernel thread for multiple mount protection */
1298 struct task_struct *s_mmp_tsk;
1299
1300 /* record the last minlen when FITRIM is called. */
1301 atomic_t s_last_trim_minblks;
1302
1303 /* Reference to checksum algorithm driver via cryptoapi */
1304 struct crypto_shash *s_chksum_driver;
1305
1306 /* Precomputed FS UUID checksum for seeding other checksums */
1307 __u32 s_csum_seed;
1308
1309 /* Reclaim extents from extent status tree */
1310 struct shrinker s_es_shrinker;
1311 struct list_head s_es_lru;
1312 struct percpu_counter s_extent_cache_cnt;
1313 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1314};
1315
1316static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1317{
1318 return sb->s_fs_info;
1319}
1320static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1321{
1322 return container_of(inode, struct ext4_inode_info, vfs_inode);
1323}
1324
1325static inline struct timespec ext4_current_time(struct inode *inode)
1326{
1327 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1328 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1329}
1330
1331static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1332{
1333 return ino == EXT4_ROOT_INO ||
1334 ino == EXT4_USR_QUOTA_INO ||
1335 ino == EXT4_GRP_QUOTA_INO ||
1336 ino == EXT4_JOURNAL_INO ||
1337 ino == EXT4_RESIZE_INO ||
1338 (ino >= EXT4_FIRST_INO(sb) &&
1339 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1340}
1341
1342static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1343 struct ext4_io_end *io_end)
1344{
1345 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1346 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1347 atomic_inc(&EXT4_I(inode)->i_unwritten);
1348 }
1349}
1350
1351static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1352{
1353 return inode->i_private;
1354}
1355
1356static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1357{
1358 inode->i_private = io;
1359}
1360
1361/*
1362 * Inode dynamic state flags
1363 */
1364enum {
1365 EXT4_STATE_JDATA, /* journaled data exists */
1366 EXT4_STATE_NEW, /* inode is newly created */
1367 EXT4_STATE_XATTR, /* has in-inode xattrs */
1368 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1369 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1370 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1371 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1372 EXT4_STATE_NEWENTRY, /* File just added to dir */
1373 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1374 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1375 nolocking */
1376 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1377};
1378
1379#define EXT4_INODE_BIT_FNS(name, field, offset) \
1380static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1381{ \
1382 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1383} \
1384static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1385{ \
1386 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1387} \
1388static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1389{ \
1390 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1391}
1392
1393EXT4_INODE_BIT_FNS(flag, flags, 0)
1394#if (BITS_PER_LONG < 64)
1395EXT4_INODE_BIT_FNS(state, state_flags, 0)
1396
1397static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1398{
1399 (ei)->i_state_flags = 0;
1400}
1401#else
1402EXT4_INODE_BIT_FNS(state, flags, 32)
1403
1404static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1405{
1406 /* We depend on the fact that callers will set i_flags */
1407}
1408#endif
1409#else
1410/* Assume that user mode programs are passing in an ext4fs superblock, not
1411 * a kernel struct super_block. This will allow us to call the feature-test
1412 * macros from user land. */
1413#define EXT4_SB(sb) (sb)
1414#endif
1415
1416#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1417
1418/*
1419 * Codes for operating systems
1420 */
1421#define EXT4_OS_LINUX 0
1422#define EXT4_OS_HURD 1
1423#define EXT4_OS_MASIX 2
1424#define EXT4_OS_FREEBSD 3
1425#define EXT4_OS_LITES 4
1426
1427/*
1428 * Revision levels
1429 */
1430#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1431#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1432
1433#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1434#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1435
1436#define EXT4_GOOD_OLD_INODE_SIZE 128
1437
1438/*
1439 * Feature set definitions
1440 */
1441
1442#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1443 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1444#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1445 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1446#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1447 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1448#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1449 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1450#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1451 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1452#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1453 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1454#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1455 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1456#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1457 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1458#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1459 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1460
1461#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1462#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1463#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1464#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1465#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1466#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1467
1468#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1469#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1470#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1471#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1472#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1473#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1474#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1475#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1476#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1477/*
1478 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1479 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1480 * all other data structures' checksums. However, the METADATA_CSUM and
1481 * GDT_CSUM bits are mutually exclusive.
1482 */
1483#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1484
1485#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1486#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1487#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1488#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1489#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1490#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1491#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1492#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1493#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1494#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1495#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1496#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1497#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1498#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1499
1500#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1501#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1502 EXT4_FEATURE_INCOMPAT_META_BG)
1503#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1504 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1505 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1506
1507#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1508#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1509 EXT4_FEATURE_INCOMPAT_RECOVER| \
1510 EXT4_FEATURE_INCOMPAT_META_BG)
1511#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1512 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1513 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1514
1515#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1516#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1517 EXT4_FEATURE_INCOMPAT_RECOVER| \
1518 EXT4_FEATURE_INCOMPAT_META_BG| \
1519 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1520 EXT4_FEATURE_INCOMPAT_64BIT| \
1521 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1522 EXT4_FEATURE_INCOMPAT_MMP | \
1523 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1524#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1525 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1526 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1527 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1528 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1529 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1530 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1531 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1532 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1533 EXT4_FEATURE_RO_COMPAT_QUOTA)
1534
1535/*
1536 * Default values for user and/or group using reserved blocks
1537 */
1538#define EXT4_DEF_RESUID 0
1539#define EXT4_DEF_RESGID 0
1540
1541#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1542
1543/*
1544 * Default mount options
1545 */
1546#define EXT4_DEFM_DEBUG 0x0001
1547#define EXT4_DEFM_BSDGROUPS 0x0002
1548#define EXT4_DEFM_XATTR_USER 0x0004
1549#define EXT4_DEFM_ACL 0x0008
1550#define EXT4_DEFM_UID16 0x0010
1551#define EXT4_DEFM_JMODE 0x0060
1552#define EXT4_DEFM_JMODE_DATA 0x0020
1553#define EXT4_DEFM_JMODE_ORDERED 0x0040
1554#define EXT4_DEFM_JMODE_WBACK 0x0060
1555#define EXT4_DEFM_NOBARRIER 0x0100
1556#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1557#define EXT4_DEFM_DISCARD 0x0400
1558#define EXT4_DEFM_NODELALLOC 0x0800
1559
1560/*
1561 * Default journal batch times
1562 */
1563#define EXT4_DEF_MIN_BATCH_TIME 0
1564#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1565
1566/*
1567 * Minimum number of groups in a flexgroup before we separate out
1568 * directories into the first block group of a flexgroup
1569 */
1570#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1571
1572/*
1573 * Structure of a directory entry
1574 */
1575#define EXT4_NAME_LEN 255
1576
1577struct ext4_dir_entry {
1578 __le32 inode; /* Inode number */
1579 __le16 rec_len; /* Directory entry length */
1580 __le16 name_len; /* Name length */
1581 char name[EXT4_NAME_LEN]; /* File name */
1582};
1583
1584/*
1585 * The new version of the directory entry. Since EXT4 structures are
1586 * stored in intel byte order, and the name_len field could never be
1587 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1588 * file_type field.
1589 */
1590struct ext4_dir_entry_2 {
1591 __le32 inode; /* Inode number */
1592 __le16 rec_len; /* Directory entry length */
1593 __u8 name_len; /* Name length */
1594 __u8 file_type;
1595 char name[EXT4_NAME_LEN]; /* File name */
1596};
1597
1598/*
1599 * This is a bogus directory entry at the end of each leaf block that
1600 * records checksums.
1601 */
1602struct ext4_dir_entry_tail {
1603 __le32 det_reserved_zero1; /* Pretend to be unused */
1604 __le16 det_rec_len; /* 12 */
1605 __u8 det_reserved_zero2; /* Zero name length */
1606 __u8 det_reserved_ft; /* 0xDE, fake file type */
1607 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1608};
1609
1610#define EXT4_DIRENT_TAIL(block, blocksize) \
1611 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1612 ((blocksize) - \
1613 sizeof(struct ext4_dir_entry_tail))))
1614
1615/*
1616 * Ext4 directory file types. Only the low 3 bits are used. The
1617 * other bits are reserved for now.
1618 */
1619#define EXT4_FT_UNKNOWN 0
1620#define EXT4_FT_REG_FILE 1
1621#define EXT4_FT_DIR 2
1622#define EXT4_FT_CHRDEV 3
1623#define EXT4_FT_BLKDEV 4
1624#define EXT4_FT_FIFO 5
1625#define EXT4_FT_SOCK 6
1626#define EXT4_FT_SYMLINK 7
1627
1628#define EXT4_FT_MAX 8
1629
1630#define EXT4_FT_DIR_CSUM 0xDE
1631
1632/*
1633 * EXT4_DIR_PAD defines the directory entries boundaries
1634 *
1635 * NOTE: It must be a multiple of 4
1636 */
1637#define EXT4_DIR_PAD 4
1638#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1639#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1640 ~EXT4_DIR_ROUND)
1641#define EXT4_MAX_REC_LEN ((1<<16)-1)
1642
1643/*
1644 * If we ever get support for fs block sizes > page_size, we'll need
1645 * to remove the #if statements in the next two functions...
1646 */
1647static inline unsigned int
1648ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1649{
1650 unsigned len = le16_to_cpu(dlen);
1651
1652#if (PAGE_CACHE_SIZE >= 65536)
1653 if (len == EXT4_MAX_REC_LEN || len == 0)
1654 return blocksize;
1655 return (len & 65532) | ((len & 3) << 16);
1656#else
1657 return len;
1658#endif
1659}
1660
1661static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1662{
1663 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1664 BUG();
1665#if (PAGE_CACHE_SIZE >= 65536)
1666 if (len < 65536)
1667 return cpu_to_le16(len);
1668 if (len == blocksize) {
1669 if (blocksize == 65536)
1670 return cpu_to_le16(EXT4_MAX_REC_LEN);
1671 else
1672 return cpu_to_le16(0);
1673 }
1674 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1675#else
1676 return cpu_to_le16(len);
1677#endif
1678}
1679
1680/*
1681 * Hash Tree Directory indexing
1682 * (c) Daniel Phillips, 2001
1683 */
1684
1685#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1686 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1687 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1688#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1689#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1690
1691/* Legal values for the dx_root hash_version field: */
1692
1693#define DX_HASH_LEGACY 0
1694#define DX_HASH_HALF_MD4 1
1695#define DX_HASH_TEA 2
1696#define DX_HASH_LEGACY_UNSIGNED 3
1697#define DX_HASH_HALF_MD4_UNSIGNED 4
1698#define DX_HASH_TEA_UNSIGNED 5
1699
1700static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1701 const void *address, unsigned int length)
1702{
1703 struct {
1704 struct shash_desc shash;
1705 char ctx[4];
1706 } desc;
1707 int err;
1708
1709 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1710
1711 desc.shash.tfm = sbi->s_chksum_driver;
1712 desc.shash.flags = 0;
1713 *(u32 *)desc.ctx = crc;
1714
1715 err = crypto_shash_update(&desc.shash, address, length);
1716 BUG_ON(err);
1717
1718 return *(u32 *)desc.ctx;
1719}
1720
1721#ifdef __KERNEL__
1722
1723/* hash info structure used by the directory hash */
1724struct dx_hash_info
1725{
1726 u32 hash;
1727 u32 minor_hash;
1728 int hash_version;
1729 u32 *seed;
1730};
1731
1732
1733/* 32 and 64 bit signed EOF for dx directories */
1734#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1735#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1736
1737
1738/*
1739 * Control parameters used by ext4_htree_next_block
1740 */
1741#define HASH_NB_ALWAYS 1
1742
1743
1744/*
1745 * Describe an inode's exact location on disk and in memory
1746 */
1747struct ext4_iloc
1748{
1749 struct buffer_head *bh;
1750 unsigned long offset;
1751 ext4_group_t block_group;
1752};
1753
1754static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1755{
1756 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1757}
1758
1759/*
1760 * This structure is stuffed into the struct file's private_data field
1761 * for directories. It is where we put information so that we can do
1762 * readdir operations in hash tree order.
1763 */
1764struct dir_private_info {
1765 struct rb_root root;
1766 struct rb_node *curr_node;
1767 struct fname *extra_fname;
1768 loff_t last_pos;
1769 __u32 curr_hash;
1770 __u32 curr_minor_hash;
1771 __u32 next_hash;
1772};
1773
1774/* calculate the first block number of the group */
1775static inline ext4_fsblk_t
1776ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1777{
1778 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1779 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1780}
1781
1782/*
1783 * Special error return code only used by dx_probe() and its callers.
1784 */
1785#define ERR_BAD_DX_DIR -75000
1786
1787void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1788 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1789
1790/*
1791 * Timeout and state flag for lazy initialization inode thread.
1792 */
1793#define EXT4_DEF_LI_WAIT_MULT 10
1794#define EXT4_DEF_LI_MAX_START_DELAY 5
1795#define EXT4_LAZYINIT_QUIT 0x0001
1796#define EXT4_LAZYINIT_RUNNING 0x0002
1797
1798/*
1799 * Lazy inode table initialization info
1800 */
1801struct ext4_lazy_init {
1802 unsigned long li_state;
1803 struct list_head li_request_list;
1804 struct mutex li_list_mtx;
1805};
1806
1807struct ext4_li_request {
1808 struct super_block *lr_super;
1809 struct ext4_sb_info *lr_sbi;
1810 ext4_group_t lr_next_group;
1811 struct list_head lr_request;
1812 unsigned long lr_next_sched;
1813 unsigned long lr_timeout;
1814};
1815
1816struct ext4_features {
1817 struct kobject f_kobj;
1818 struct completion f_kobj_unregister;
1819};
1820
1821/*
1822 * This structure will be used for multiple mount protection. It will be
1823 * written into the block number saved in the s_mmp_block field in the
1824 * superblock. Programs that check MMP should assume that if
1825 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1826 * to use the filesystem, regardless of how old the timestamp is.
1827 */
1828#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1829#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1830#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1831#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1832
1833struct mmp_struct {
1834 __le32 mmp_magic; /* Magic number for MMP */
1835 __le32 mmp_seq; /* Sequence no. updated periodically */
1836
1837 /*
1838 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1839 * purposes and do not affect the correctness of the algorithm
1840 */
1841 __le64 mmp_time; /* Time last updated */
1842 char mmp_nodename[64]; /* Node which last updated MMP block */
1843 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1844
1845 /*
1846 * mmp_check_interval is used to verify if the MMP block has been
1847 * updated on the block device. The value is updated based on the
1848 * maximum time to write the MMP block during an update cycle.
1849 */
1850 __le16 mmp_check_interval;
1851
1852 __le16 mmp_pad1;
1853 __le32 mmp_pad2[226];
1854 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1855};
1856
1857/* arguments passed to the mmp thread */
1858struct mmpd_data {
1859 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1860 struct super_block *sb; /* super block of the fs */
1861};
1862
1863/*
1864 * Check interval multiplier
1865 * The MMP block is written every update interval and initially checked every
1866 * update interval x the multiplier (the value is then adapted based on the
1867 * write latency). The reason is that writes can be delayed under load and we
1868 * don't want readers to incorrectly assume that the filesystem is no longer
1869 * in use.
1870 */
1871#define EXT4_MMP_CHECK_MULT 2UL
1872
1873/*
1874 * Minimum interval for MMP checking in seconds.
1875 */
1876#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1877
1878/*
1879 * Maximum interval for MMP checking in seconds.
1880 */
1881#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1882
1883/*
1884 * Function prototypes
1885 */
1886
1887/*
1888 * Ok, these declarations are also in <linux/kernel.h> but none of the
1889 * ext4 source programs needs to include it so they are duplicated here.
1890 */
1891# define NORET_TYPE /**/
1892# define ATTRIB_NORET __attribute__((noreturn))
1893# define NORET_AND noreturn,
1894
1895/* bitmap.c */
1896extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1897void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1898 struct ext4_group_desc *gdp,
1899 struct buffer_head *bh, int sz);
1900int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1901 struct ext4_group_desc *gdp,
1902 struct buffer_head *bh, int sz);
1903void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1904 struct ext4_group_desc *gdp,
1905 struct buffer_head *bh);
1906int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1907 struct ext4_group_desc *gdp,
1908 struct buffer_head *bh);
1909
1910/* balloc.c */
1911extern void ext4_validate_block_bitmap(struct super_block *sb,
1912 struct ext4_group_desc *desc,
1913 unsigned int block_group,
1914 struct buffer_head *bh);
1915extern unsigned int ext4_block_group(struct super_block *sb,
1916 ext4_fsblk_t blocknr);
1917extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1918 ext4_fsblk_t blocknr);
1919extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1920extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1921 ext4_group_t group);
1922extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1923 ext4_fsblk_t goal,
1924 unsigned int flags,
1925 unsigned long *count,
1926 int *errp);
1927extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1928 s64 nclusters, unsigned int flags);
1929extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1930extern void ext4_check_blocks_bitmap(struct super_block *);
1931extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1932 ext4_group_t block_group,
1933 struct buffer_head ** bh);
1934extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1935
1936extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1937 ext4_group_t block_group);
1938extern int ext4_wait_block_bitmap(struct super_block *sb,
1939 ext4_group_t block_group,
1940 struct buffer_head *bh);
1941extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1942 ext4_group_t block_group);
1943extern void ext4_init_block_bitmap(struct super_block *sb,
1944 struct buffer_head *bh,
1945 ext4_group_t group,
1946 struct ext4_group_desc *desc);
1947extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1948 ext4_group_t block_group,
1949 struct ext4_group_desc *gdp);
1950extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1951 ext4_group_t block_group,
1952 struct ext4_group_desc *gdp);
1953ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1954
1955/* dir.c */
1956extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1957 struct file *,
1958 struct ext4_dir_entry_2 *,
1959 struct buffer_head *, char *, int,
1960 unsigned int);
1961#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1962 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1963 (de), (bh), (buf), (size), (offset)))
1964extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1965 __u32 minor_hash,
1966 struct ext4_dir_entry_2 *dirent);
1967extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1968extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1969 struct buffer_head *bh,
1970 void *buf, int buf_size,
1971 const char *name, int namelen,
1972 struct ext4_dir_entry_2 **dest_de);
1973void ext4_insert_dentry(struct inode *inode,
1974 struct ext4_dir_entry_2 *de,
1975 int buf_size,
1976 const char *name, int namelen);
1977static inline void ext4_update_dx_flag(struct inode *inode)
1978{
1979 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1980 EXT4_FEATURE_COMPAT_DIR_INDEX))
1981 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1982}
1983static unsigned char ext4_filetype_table[] = {
1984 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1985};
1986
1987static inline unsigned char get_dtype(struct super_block *sb, int filetype)
1988{
1989 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
1990 (filetype >= EXT4_FT_MAX))
1991 return DT_UNKNOWN;
1992
1993 return ext4_filetype_table[filetype];
1994}
1995
1996/* fsync.c */
1997extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1998extern int ext4_flush_unwritten_io(struct inode *);
1999
2000/* hash.c */
2001extern int ext4fs_dirhash(const char *name, int len, struct
2002 dx_hash_info *hinfo);
2003
2004/* ialloc.c */
2005extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2006 const struct qstr *qstr, __u32 goal,
2007 uid_t *owner, int handle_type,
2008 unsigned int line_no, int nblocks);
2009
2010#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2011 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2012 0, 0, 0)
2013#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2014 type, nblocks) \
2015 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2016 (type), __LINE__, (nblocks))
2017
2018
2019extern void ext4_free_inode(handle_t *, struct inode *);
2020extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2021extern unsigned long ext4_count_free_inodes(struct super_block *);
2022extern unsigned long ext4_count_dirs(struct super_block *);
2023extern void ext4_check_inodes_bitmap(struct super_block *);
2024extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2025extern int ext4_init_inode_table(struct super_block *sb,
2026 ext4_group_t group, int barrier);
2027extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2028
2029/* mballoc.c */
2030extern long ext4_mb_stats;
2031extern long ext4_mb_max_to_scan;
2032extern int ext4_mb_init(struct super_block *);
2033extern int ext4_mb_release(struct super_block *);
2034extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2035 struct ext4_allocation_request *, int *);
2036extern int ext4_mb_reserve_blocks(struct super_block *, int);
2037extern void ext4_discard_preallocations(struct inode *);
2038extern int __init ext4_init_mballoc(void);
2039extern void ext4_exit_mballoc(void);
2040extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2041 struct buffer_head *bh, ext4_fsblk_t block,
2042 unsigned long count, int flags);
2043extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2044 ext4_group_t ngroups);
2045extern int ext4_mb_add_groupinfo(struct super_block *sb,
2046 ext4_group_t i, struct ext4_group_desc *desc);
2047extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2048 ext4_fsblk_t block, unsigned long count);
2049extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2050
2051/* inode.c */
2052struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2053 ext4_lblk_t, int, int *);
2054struct buffer_head *ext4_bread(handle_t *, struct inode *,
2055 ext4_lblk_t, int, int *);
2056int ext4_get_block_write(struct inode *inode, sector_t iblock,
2057 struct buffer_head *bh_result, int create);
2058int ext4_get_block(struct inode *inode, sector_t iblock,
2059 struct buffer_head *bh_result, int create);
2060int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2061 struct buffer_head *bh, int create);
2062int ext4_walk_page_buffers(handle_t *handle,
2063 struct buffer_head *head,
2064 unsigned from,
2065 unsigned to,
2066 int *partial,
2067 int (*fn)(handle_t *handle,
2068 struct buffer_head *bh));
2069int do_journal_get_write_access(handle_t *handle,
2070 struct buffer_head *bh);
2071#define FALL_BACK_TO_NONDELALLOC 1
2072#define CONVERT_INLINE_DATA 2
2073
2074extern struct inode *ext4_iget(struct super_block *, unsigned long);
2075extern int ext4_write_inode(struct inode *, struct writeback_control *);
2076extern int ext4_setattr(struct dentry *, struct iattr *);
2077extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2078 struct kstat *stat);
2079extern void ext4_evict_inode(struct inode *);
2080extern void ext4_clear_inode(struct inode *);
2081extern int ext4_sync_inode(handle_t *, struct inode *);
2082extern void ext4_dirty_inode(struct inode *, int);
2083extern int ext4_change_inode_journal_flag(struct inode *, int);
2084extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2085extern int ext4_can_truncate(struct inode *inode);
2086extern void ext4_truncate(struct inode *);
2087extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2088extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2089extern void ext4_set_inode_flags(struct inode *);
2090extern void ext4_get_inode_flags(struct ext4_inode_info *);
2091extern int ext4_alloc_da_blocks(struct inode *inode);
2092extern void ext4_set_aops(struct inode *inode);
2093extern int ext4_writepage_trans_blocks(struct inode *);
2094extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2095extern int ext4_discard_partial_page_buffers(handle_t *handle,
2096 struct address_space *mapping, loff_t from,
2097 loff_t length, int flags);
2098extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2099extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2100extern void ext4_da_update_reserve_space(struct inode *inode,
2101 int used, int quota_claim);
2102
2103/* indirect.c */
2104extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2105 struct ext4_map_blocks *map, int flags);
2106extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2107 const struct iovec *iov, loff_t offset,
2108 unsigned long nr_segs);
2109extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2110extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2111extern void ext4_ind_truncate(struct inode *inode);
2112extern int ext4_ind_punch_hole(struct file *file, loff_t offset, loff_t length);
2113
2114/* ioctl.c */
2115extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2116extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2117
2118/* migrate.c */
2119extern int ext4_ext_migrate(struct inode *);
2120
2121/* namei.c */
2122extern int ext4_dirent_csum_verify(struct inode *inode,
2123 struct ext4_dir_entry *dirent);
2124extern int ext4_orphan_add(handle_t *, struct inode *);
2125extern int ext4_orphan_del(handle_t *, struct inode *);
2126extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2127 __u32 start_minor_hash, __u32 *next_hash);
2128extern int search_dir(struct buffer_head *bh,
2129 char *search_buf,
2130 int buf_size,
2131 struct inode *dir,
2132 const struct qstr *d_name,
2133 unsigned int offset,
2134 struct ext4_dir_entry_2 **res_dir);
2135extern int ext4_generic_delete_entry(handle_t *handle,
2136 struct inode *dir,
2137 struct ext4_dir_entry_2 *de_del,
2138 struct buffer_head *bh,
2139 void *entry_buf,
2140 int buf_size,
2141 int csum_size);
2142
2143/* resize.c */
2144extern int ext4_group_add(struct super_block *sb,
2145 struct ext4_new_group_data *input);
2146extern int ext4_group_extend(struct super_block *sb,
2147 struct ext4_super_block *es,
2148 ext4_fsblk_t n_blocks_count);
2149extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2150
2151/* super.c */
2152extern int ext4_calculate_overhead(struct super_block *sb);
2153extern int ext4_superblock_csum_verify(struct super_block *sb,
2154 struct ext4_super_block *es);
2155extern void ext4_superblock_csum_set(struct super_block *sb);
2156extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2157extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2158extern void ext4_kvfree(void *ptr);
2159extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2160 ext4_group_t ngroup);
2161extern const char *ext4_decode_error(struct super_block *sb, int errno,
2162 char nbuf[16]);
2163extern __printf(4, 5)
2164void __ext4_error(struct super_block *, const char *, unsigned int,
2165 const char *, ...);
2166#define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2167 __LINE__, ## message)
2168extern __printf(5, 6)
2169void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2170 const char *, ...);
2171extern __printf(5, 6)
2172void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2173 const char *, ...);
2174extern void __ext4_std_error(struct super_block *, const char *,
2175 unsigned int, int);
2176extern __printf(4, 5)
2177void __ext4_abort(struct super_block *, const char *, unsigned int,
2178 const char *, ...);
2179#define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2180 __LINE__, ## message)
2181extern __printf(4, 5)
2182void __ext4_warning(struct super_block *, const char *, unsigned int,
2183 const char *, ...);
2184#define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2185 __LINE__, ## message)
2186extern __printf(3, 4)
2187void ext4_msg(struct super_block *, const char *, const char *, ...);
2188extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2189 const char *, unsigned int, const char *);
2190#define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2191 __LINE__, msg)
2192extern __printf(7, 8)
2193void __ext4_grp_locked_error(const char *, unsigned int,
2194 struct super_block *, ext4_group_t,
2195 unsigned long, ext4_fsblk_t,
2196 const char *, ...);
2197#define ext4_grp_locked_error(sb, grp, message...) \
2198 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2199extern void ext4_update_dynamic_rev(struct super_block *sb);
2200extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2201 __u32 compat);
2202extern int ext4_update_rocompat_feature(handle_t *handle,
2203 struct super_block *sb, __u32 rocompat);
2204extern int ext4_update_incompat_feature(handle_t *handle,
2205 struct super_block *sb, __u32 incompat);
2206extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2207 struct ext4_group_desc *bg);
2208extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2209 struct ext4_group_desc *bg);
2210extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2211 struct ext4_group_desc *bg);
2212extern __u32 ext4_free_group_clusters(struct super_block *sb,
2213 struct ext4_group_desc *bg);
2214extern __u32 ext4_free_inodes_count(struct super_block *sb,
2215 struct ext4_group_desc *bg);
2216extern __u32 ext4_used_dirs_count(struct super_block *sb,
2217 struct ext4_group_desc *bg);
2218extern __u32 ext4_itable_unused_count(struct super_block *sb,
2219 struct ext4_group_desc *bg);
2220extern void ext4_block_bitmap_set(struct super_block *sb,
2221 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2222extern void ext4_inode_bitmap_set(struct super_block *sb,
2223 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2224extern void ext4_inode_table_set(struct super_block *sb,
2225 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2226extern void ext4_free_group_clusters_set(struct super_block *sb,
2227 struct ext4_group_desc *bg,
2228 __u32 count);
2229extern void ext4_free_inodes_set(struct super_block *sb,
2230 struct ext4_group_desc *bg, __u32 count);
2231extern void ext4_used_dirs_set(struct super_block *sb,
2232 struct ext4_group_desc *bg, __u32 count);
2233extern void ext4_itable_unused_set(struct super_block *sb,
2234 struct ext4_group_desc *bg, __u32 count);
2235extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2236 struct ext4_group_desc *gdp);
2237extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2238 struct ext4_group_desc *gdp);
2239extern int ext4_register_li_request(struct super_block *sb,
2240 ext4_group_t first_not_zeroed);
2241
2242static inline int ext4_has_group_desc_csum(struct super_block *sb)
2243{
2244 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2245 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2246 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2247}
2248
2249static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2250{
2251 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2252 le32_to_cpu(es->s_blocks_count_lo);
2253}
2254
2255static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2256{
2257 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2258 le32_to_cpu(es->s_r_blocks_count_lo);
2259}
2260
2261static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2262{
2263 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2264 le32_to_cpu(es->s_free_blocks_count_lo);
2265}
2266
2267static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2268 ext4_fsblk_t blk)
2269{
2270 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2271 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2272}
2273
2274static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2275 ext4_fsblk_t blk)
2276{
2277 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2278 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2279}
2280
2281static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2282 ext4_fsblk_t blk)
2283{
2284 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2285 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2286}
2287
2288static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2289{
2290 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2291 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2292 le32_to_cpu(raw_inode->i_size_lo);
2293 else
2294 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2295}
2296
2297static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2298{
2299 raw_inode->i_size_lo = cpu_to_le32(i_size);
2300 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2301}
2302
2303static inline
2304struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2305 ext4_group_t group)
2306{
2307 struct ext4_group_info ***grp_info;
2308 long indexv, indexh;
2309 grp_info = EXT4_SB(sb)->s_group_info;
2310 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2311 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2312 return grp_info[indexv][indexh];
2313}
2314
2315/*
2316 * Reading s_groups_count requires using smp_rmb() afterwards. See
2317 * the locking protocol documented in the comments of ext4_group_add()
2318 * in resize.c
2319 */
2320static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2321{
2322 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2323
2324 smp_rmb();
2325 return ngroups;
2326}
2327
2328static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2329 ext4_group_t block_group)
2330{
2331 return block_group >> sbi->s_log_groups_per_flex;
2332}
2333
2334static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2335{
2336 return 1 << sbi->s_log_groups_per_flex;
2337}
2338
2339#define ext4_std_error(sb, errno) \
2340do { \
2341 if ((errno)) \
2342 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2343} while (0)
2344
2345#ifdef CONFIG_SMP
2346/* Each CPU can accumulate percpu_counter_batch clusters in their local
2347 * counters. So we need to make sure we have free clusters more
2348 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2349 */
2350#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2351#else
2352#define EXT4_FREECLUSTERS_WATERMARK 0
2353#endif
2354
2355static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2356{
2357 /*
2358 * XXX: replace with spinlock if seen contended -bzzz
2359 */
2360 down_write(&EXT4_I(inode)->i_data_sem);
2361 if (newsize > EXT4_I(inode)->i_disksize)
2362 EXT4_I(inode)->i_disksize = newsize;
2363 up_write(&EXT4_I(inode)->i_data_sem);
2364 return ;
2365}
2366
2367struct ext4_group_info {
2368 unsigned long bb_state;
2369 struct rb_root bb_free_root;
2370 ext4_grpblk_t bb_first_free; /* first free block */
2371 ext4_grpblk_t bb_free; /* total free blocks */
2372 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2373 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2374 struct list_head bb_prealloc_list;
2375#ifdef DOUBLE_CHECK
2376 void *bb_bitmap;
2377#endif
2378 struct rw_semaphore alloc_sem;
2379 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2380 * regions, index is order.
2381 * bb_counters[3] = 5 means
2382 * 5 free 8-block regions. */
2383};
2384
2385#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2386#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2387
2388#define EXT4_MB_GRP_NEED_INIT(grp) \
2389 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2390
2391#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2392 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2393#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2394 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2395#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2396 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2397
2398#define EXT4_MAX_CONTENTION 8
2399#define EXT4_CONTENTION_THRESHOLD 2
2400
2401static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2402 ext4_group_t group)
2403{
2404 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2405}
2406
2407/*
2408 * Returns true if the filesystem is busy enough that attempts to
2409 * access the block group locks has run into contention.
2410 */
2411static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2412{
2413 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2414}
2415
2416static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2417{
2418 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2419 if (spin_trylock(lock))
2420 /*
2421 * We're able to grab the lock right away, so drop the
2422 * lock contention counter.
2423 */
2424 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2425 else {
2426 /*
2427 * The lock is busy, so bump the contention counter,
2428 * and then wait on the spin lock.
2429 */
2430 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2431 EXT4_MAX_CONTENTION);
2432 spin_lock(lock);
2433 }
2434}
2435
2436static inline void ext4_unlock_group(struct super_block *sb,
2437 ext4_group_t group)
2438{
2439 spin_unlock(ext4_group_lock_ptr(sb, group));
2440}
2441
2442/*
2443 * Block validity checking
2444 */
2445#define ext4_check_indirect_blockref(inode, bh) \
2446 ext4_check_blockref(__func__, __LINE__, inode, \
2447 (__le32 *)(bh)->b_data, \
2448 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2449
2450#define ext4_ind_check_inode(inode) \
2451 ext4_check_blockref(__func__, __LINE__, inode, \
2452 EXT4_I(inode)->i_data, \
2453 EXT4_NDIR_BLOCKS)
2454
2455/*
2456 * Inodes and files operations
2457 */
2458
2459/* dir.c */
2460extern const struct file_operations ext4_dir_operations;
2461
2462/* file.c */
2463extern const struct inode_operations ext4_file_inode_operations;
2464extern const struct file_operations ext4_file_operations;
2465extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2466extern void ext4_unwritten_wait(struct inode *inode);
2467
2468/* inline.c */
2469extern int ext4_has_inline_data(struct inode *inode);
2470extern int ext4_get_inline_size(struct inode *inode);
2471extern int ext4_get_max_inline_size(struct inode *inode);
2472extern int ext4_find_inline_data_nolock(struct inode *inode);
2473extern void ext4_write_inline_data(struct inode *inode,
2474 struct ext4_iloc *iloc,
2475 void *buffer, loff_t pos,
2476 unsigned int len);
2477extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2478 unsigned int len);
2479extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2480 unsigned int len);
2481extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2482
2483extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2484extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2485 struct inode *inode,
2486 loff_t pos, unsigned len,
2487 unsigned flags,
2488 struct page **pagep);
2489extern int ext4_write_inline_data_end(struct inode *inode,
2490 loff_t pos, unsigned len,
2491 unsigned copied,
2492 struct page *page);
2493extern struct buffer_head *
2494ext4_journalled_write_inline_data(struct inode *inode,
2495 unsigned len,
2496 struct page *page);
2497extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2498 struct inode *inode,
2499 loff_t pos, unsigned len,
2500 unsigned flags,
2501 struct page **pagep,
2502 void **fsdata);
2503extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2504 unsigned len, unsigned copied,
2505 struct page *page);
2506extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2507 struct inode *inode);
2508extern int ext4_try_create_inline_dir(handle_t *handle,
2509 struct inode *parent,
2510 struct inode *inode);
2511extern int ext4_read_inline_dir(struct file *filp,
2512 void *dirent, filldir_t filldir,
2513 int *has_inline_data);
2514extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2515 const struct qstr *d_name,
2516 struct ext4_dir_entry_2 **res_dir,
2517 int *has_inline_data);
2518extern int ext4_delete_inline_entry(handle_t *handle,
2519 struct inode *dir,
2520 struct ext4_dir_entry_2 *de_del,
2521 struct buffer_head *bh,
2522 int *has_inline_data);
2523extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2524extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2525 struct ext4_dir_entry_2 **parent_de,
2526 int *retval);
2527extern int ext4_inline_data_fiemap(struct inode *inode,
2528 struct fiemap_extent_info *fieinfo,
2529 int *has_inline);
2530extern int ext4_try_to_evict_inline_data(handle_t *handle,
2531 struct inode *inode,
2532 int needed);
2533extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2534
2535extern int ext4_convert_inline_data(struct inode *inode);
2536
2537/* namei.c */
2538extern const struct inode_operations ext4_dir_inode_operations;
2539extern const struct inode_operations ext4_special_inode_operations;
2540extern struct dentry *ext4_get_parent(struct dentry *child);
2541extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2542 struct ext4_dir_entry_2 *de,
2543 int blocksize, int csum_size,
2544 unsigned int parent_ino, int dotdot_real_len);
2545extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2546 unsigned int blocksize);
2547extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2548 struct inode *inode,
2549 struct buffer_head *bh);
2550
2551/* symlink.c */
2552extern const struct inode_operations ext4_symlink_inode_operations;
2553extern const struct inode_operations ext4_fast_symlink_inode_operations;
2554
2555/* block_validity */
2556extern void ext4_release_system_zone(struct super_block *sb);
2557extern int ext4_setup_system_zone(struct super_block *sb);
2558extern int __init ext4_init_system_zone(void);
2559extern void ext4_exit_system_zone(void);
2560extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2561 ext4_fsblk_t start_blk,
2562 unsigned int count);
2563extern int ext4_check_blockref(const char *, unsigned int,
2564 struct inode *, __le32 *, unsigned int);
2565
2566/* extents.c */
2567struct ext4_ext_path;
2568struct ext4_extent;
2569
2570extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2571extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2572extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2573 int chunk);
2574extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2575 struct ext4_map_blocks *map, int flags);
2576extern void ext4_ext_truncate(struct inode *);
2577extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2578 loff_t length);
2579extern void ext4_ext_init(struct super_block *);
2580extern void ext4_ext_release(struct super_block *);
2581extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2582 loff_t len);
2583extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2584 ssize_t len);
2585extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2586 struct ext4_map_blocks *map, int flags);
2587extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2588 ext4_lblk_t lblocks);
2589extern int ext4_extent_tree_init(handle_t *, struct inode *);
2590extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2591 int num,
2592 struct ext4_ext_path *path);
2593extern int ext4_can_extents_be_merged(struct inode *inode,
2594 struct ext4_extent *ex1,
2595 struct ext4_extent *ex2);
2596extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2597 struct ext4_ext_path *,
2598 struct ext4_extent *, int);
2599extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2600 struct ext4_ext_path *);
2601extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2602extern int ext4_ext_check_inode(struct inode *inode);
2603extern int ext4_find_delalloc_range(struct inode *inode,
2604 ext4_lblk_t lblk_start,
2605 ext4_lblk_t lblk_end);
2606extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2607extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2608 __u64 start, __u64 len);
2609
2610
2611/* move_extent.c */
2612extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2613 __u64 start_orig, __u64 start_donor,
2614 __u64 len, __u64 *moved_len);
2615
2616/* page-io.c */
2617extern int __init ext4_init_pageio(void);
2618extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2619extern void ext4_exit_pageio(void);
2620extern void ext4_ioend_wait(struct inode *);
2621extern void ext4_free_io_end(ext4_io_end_t *io);
2622extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2623extern void ext4_end_io_work(struct work_struct *work);
2624extern void ext4_io_submit(struct ext4_io_submit *io);
2625extern int ext4_bio_write_page(struct ext4_io_submit *io,
2626 struct page *page,
2627 int len,
2628 struct writeback_control *wbc);
2629
2630/* mmp.c */
2631extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2632extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2633extern int ext4_mmp_csum_verify(struct super_block *sb,
2634 struct mmp_struct *mmp);
2635
2636/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2637enum ext4_state_bits {
2638 BH_Uninit /* blocks are allocated but uninitialized on disk */
2639 = BH_JBDPrivateStart,
2640 BH_AllocFromCluster, /* allocated blocks were part of already
2641 * allocated cluster. Note that this flag will
2642 * never, ever appear in a buffer_head's state
2643 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2644 * this is used. */
2645};
2646
2647BUFFER_FNS(Uninit, uninit)
2648TAS_BUFFER_FNS(Uninit, uninit)
2649
2650/*
2651 * Add new method to test whether block and inode bitmaps are properly
2652 * initialized. With uninit_bg reading the block from disk is not enough
2653 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2654 */
2655#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2656
2657static inline int bitmap_uptodate(struct buffer_head *bh)
2658{
2659 return (buffer_uptodate(bh) &&
2660 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2661}
2662static inline void set_bitmap_uptodate(struct buffer_head *bh)
2663{
2664 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2665}
2666
2667/*
2668 * Disable DIO read nolock optimization, so new dioreaders will be forced
2669 * to grab i_mutex
2670 */
2671static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2672{
2673 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2674 smp_mb();
2675}
2676static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2677{
2678 smp_mb();
2679 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2680}
2681
2682#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2683
2684/* For ioend & aio unwritten conversion wait queues */
2685#define EXT4_WQ_HASH_SZ 37
2686#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2687 EXT4_WQ_HASH_SZ])
2688#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2689 EXT4_WQ_HASH_SZ])
2690extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2691extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2692
2693#define EXT4_RESIZING 0
2694extern int ext4_resize_begin(struct super_block *sb);
2695extern void ext4_resize_end(struct super_block *sb);
2696
2697#endif /* __KERNEL__ */
2698
2699#endif /* _EXT4_H */