ext4: fix ext4_ext_direct_IO()'s return value after converting uninit extents
[deliverable/linux.git] / fs / ext4 / ext4.h
CommitLineData
ac27a0ec 1/*
3dcf5451 2 * ext4.h
ac27a0ec
DK
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
3dcf5451
CH
16#ifndef _EXT4_H
17#define _EXT4_H
ac27a0ec
DK
18
19#include <linux/types.h>
3a5b2ecd 20#include <linux/blkdev.h>
ac27a0ec 21#include <linux/magic.h>
2ccb5fb9 22#include <linux/jbd2.h>
60e58e0f 23#include <linux/quota.h>
d444c3c3
TT
24#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
ca0faba0
TT
28#include <linux/timer.h>
29#include <linux/wait.h>
30#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h>
c9de560d 32
ac27a0ec 33/*
8a35694e 34 * The fourth extended filesystem constants/structures
ac27a0ec
DK
35 */
36
37/*
617ba13b 38 * Define EXT4FS_DEBUG to produce debug messages
ac27a0ec 39 */
617ba13b 40#undef EXT4FS_DEBUG
ac27a0ec 41
ac27a0ec
DK
42/*
43 * Debug code
44 */
617ba13b
MC
45#ifdef EXT4FS_DEBUG
46#define ext4_debug(f, a...) \
ac27a0ec 47 do { \
af5bc92d 48 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
4db9c54a 49 __FILE__, __LINE__, __func__); \
af5bc92d 50 printk(KERN_DEBUG f, ## a); \
ac27a0ec
DK
51 } while (0)
52#else
617ba13b 53#define ext4_debug(f, a...) do {} while (0)
ac27a0ec
DK
54#endif
55
d444c3c3
TT
56/* data type for block offset of block group */
57typedef int ext4_grpblk_t;
58
59/* data type for filesystem-wide blocks number */
60typedef unsigned long long ext4_fsblk_t;
61
62/* data type for file logical block number */
63typedef __u32 ext4_lblk_t;
64
65/* data type for block group number */
66typedef unsigned int ext4_group_t;
67
296c355c
TT
68/*
69 * Flags used in mballoc's allocation_context flags field.
70 *
71 * Also used to show what's going on for debugging purposes when the
72 * flag field is exported via the traceport interface
73 */
d444c3c3 74
c9de560d 75/* prefer goal again. length */
0ef90db9 76#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 77/* blocks already reserved */
0ef90db9 78#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 79/* metadata is being allocated */
0ef90db9 80#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 81/* first blocks in the file */
0ef90db9 82#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 83/* search for the best chunk */
0ef90db9 84#define EXT4_MB_HINT_BEST 0x0010
c9de560d 85/* data is being allocated */
0ef90db9 86#define EXT4_MB_HINT_DATA 0x0020
c9de560d 87/* don't preallocate (for tails) */
0ef90db9 88#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 89/* allocate for locality group */
0ef90db9 90#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 91/* allocate goal blocks or none */
0ef90db9 92#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 93/* goal is meaningful */
0ef90db9 94#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 95/* blocks already pre-reserved by delayed allocation */
0ef90db9 96#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
97/* We are doing stream allocation */
98#define EXT4_MB_STREAM_ALLOC 0x0800
d2a17637 99
c9de560d
AT
100
101struct ext4_allocation_request {
102 /* target inode for block we're allocating */
103 struct inode *inode;
726447d8
ES
104 /* how many blocks we want to allocate */
105 unsigned int len;
c9de560d
AT
106 /* logical block in target inode */
107 ext4_lblk_t logical;
c9de560d
AT
108 /* the closest logical allocated block to the left */
109 ext4_lblk_t lleft;
c9de560d
AT
110 /* the closest logical allocated block to the right */
111 ext4_lblk_t lright;
726447d8
ES
112 /* phys. target (a hint) */
113 ext4_fsblk_t goal;
114 /* phys. block for the closest logical allocated block to the left */
115 ext4_fsblk_t pleft;
116 /* phys. block for the closest logical allocated block to the right */
c9de560d 117 ext4_fsblk_t pright;
c9de560d 118 /* flags. see above EXT4_MB_HINT_* */
498e5f24 119 unsigned int flags;
c9de560d
AT
120};
121
b3a3ca8c
TT
122/*
123 * For delayed allocation tracking
124 */
125struct mpage_da_data {
126 struct inode *inode;
127 sector_t b_blocknr; /* start block number of extent */
128 size_t b_size; /* size of extent */
129 unsigned long b_state; /* state of the extent */
130 unsigned long first_page, next_page; /* extent of pages */
131 struct writeback_control *wbc;
132 int io_done;
133 int pages_written;
134 int retval;
135};
8d5d02e6 136#define DIO_AIO_UNWRITTEN 0x1
0031462b 137typedef struct ext4_io_end {
8d5d02e6 138 struct list_head list; /* per-file finished AIO list */
0031462b 139 struct inode *inode; /* file being written to */
8d5d02e6 140 unsigned int flag; /* unwritten or not */
0031462b
MC
141 int error; /* I/O error code */
142 ext4_lblk_t offset; /* offset in the file */
143 size_t size; /* size of the extent */
144 struct work_struct work; /* data work queue */
145} ext4_io_end_t;
146
ac27a0ec
DK
147/*
148 * Special inodes numbers
149 */
617ba13b
MC
150#define EXT4_BAD_INO 1 /* Bad blocks inode */
151#define EXT4_ROOT_INO 2 /* Root inode */
152#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
153#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
154#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
155#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 156
617ba13b
MC
157/* First non-reserved inode for old ext4 filesystems */
158#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
159
160/*
161 * Maximal count of links to a file
162 */
f8628a14 163#define EXT4_LINK_MAX 65000
ac27a0ec
DK
164
165/*
166 * Macro-instructions used to manage several block sizes
167 */
617ba13b 168#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
169#define EXT4_MAX_BLOCK_SIZE 65536
170#define EXT4_MIN_BLOCK_LOG_SIZE 10
ac27a0ec 171#ifdef __KERNEL__
617ba13b 172# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 173#else
617ba13b 174# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 175#endif
af5bc92d 176#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
ac27a0ec 177#ifdef __KERNEL__
617ba13b 178# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
ac27a0ec 179#else
617ba13b 180# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
181#endif
182#ifdef __KERNEL__
617ba13b
MC
183#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
184#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
185#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 186#else
617ba13b
MC
187#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
188 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 189 (s)->s_inode_size)
617ba13b
MC
190#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
191 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
192 (s)->s_first_ino)
193#endif
a2df2a63 194#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 195
ac27a0ec
DK
196/*
197 * Structure of a blocks group descriptor
198 */
617ba13b 199struct ext4_group_desc
ac27a0ec 200{
3a14589c 201 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
202 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
203 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
204 __le16 bg_free_blocks_count_lo;/* Free blocks count */
205 __le16 bg_free_inodes_count_lo;/* Free inodes count */
206 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4
AD
207 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
208 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
560671a0 209 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 210 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
211 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
212 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
213 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
214 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
215 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
216 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 217 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
91b51a01 218 __u32 bg_reserved2[3];
ac27a0ec
DK
219};
220
772cb7c8
JS
221/*
222 * Structure of a flex block group info
223 */
224
225struct flex_groups {
9f24e420
TT
226 atomic_t free_inodes;
227 atomic_t free_blocks;
7d39db14 228 atomic_t used_dirs;
772cb7c8
JS
229};
230
717d50e4
AD
231#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
232#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
233#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
234
ac27a0ec
DK
235/*
236 * Macro-instructions used to manage group descriptors
237 */
0d1ee42f 238#define EXT4_MIN_DESC_SIZE 32
8fadc143 239#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
240#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
241#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 242#ifdef __KERNEL__
617ba13b
MC
243# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
244# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
245# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
246# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 247#else
617ba13b 248# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 249# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 250# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
251#endif
252
253/*
254 * Constants relative to the data blocks
255 */
617ba13b
MC
256#define EXT4_NDIR_BLOCKS 12
257#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
258#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
259#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
260#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
261
262/*
263 * Inode flags
264 */
617ba13b
MC
265#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
266#define EXT4_UNRM_FL 0x00000002 /* Undelete */
267#define EXT4_COMPR_FL 0x00000004 /* Compress file */
268#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
269#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
270#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
271#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
272#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 273/* Reserved for compression usage... */
617ba13b
MC
274#define EXT4_DIRTY_FL 0x00000100
275#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
276#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
277#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
ac27a0ec 278/* End compression flags --- maybe not all used */
617ba13b
MC
279#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
280#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
281#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
282#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
283#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
284#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 285#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 286#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
8180a562 287#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 288
a86c6181 289#define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
4db46fc2 290#define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
ac27a0ec 291
8fa43a81
DG
292/* Flags that should be inherited by new inodes from their parent. */
293#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
294 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
295 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
296 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
297 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
298
2dc6b0d4
DG
299/* Flags that are appropriate for regular files (all but dir-specific ones). */
300#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
301
302/* Flags that are appropriate for non-directories/regular files. */
303#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
304
305/* Mask out flags that are inappropriate for the given type of inode. */
306static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
307{
308 if (S_ISDIR(mode))
309 return flags;
310 else if (S_ISREG(mode))
311 return flags & EXT4_REG_FLMASK;
312 else
313 return flags & EXT4_OTHER_FLMASK;
314}
315
ac27a0ec
DK
316/*
317 * Inode dynamic state flags
318 */
617ba13b
MC
319#define EXT4_STATE_JDATA 0x00000001 /* journaled data exists */
320#define EXT4_STATE_NEW 0x00000002 /* inode is newly created */
321#define EXT4_STATE_XATTR 0x00000004 /* has in-inode xattrs */
6dd4ee7c 322#define EXT4_STATE_NO_EXPAND 0x00000008 /* No space for expansion */
7d8f9f7d 323#define EXT4_STATE_DA_ALLOC_CLOSE 0x00000010 /* Alloc DA blks on close */
1b9c12f4 324#define EXT4_STATE_EXT_MIGRATE 0x00000020 /* Inode is migrating */
ac27a0ec
DK
325
326/* Used to pass group descriptor data when online resize is done */
617ba13b 327struct ext4_new_group_input {
8c55e204
DK
328 __u32 group; /* Group number for this data */
329 __u64 block_bitmap; /* Absolute block number of block bitmap */
330 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
331 __u64 inode_table; /* Absolute block number of inode table start */
332 __u32 blocks_count; /* Total number of blocks in this group */
333 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
334 __u16 unused;
335};
336
617ba13b
MC
337/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
338struct ext4_new_group_data {
ac27a0ec 339 __u32 group;
bd81d8ee
LV
340 __u64 block_bitmap;
341 __u64 inode_bitmap;
342 __u64 inode_table;
ac27a0ec
DK
343 __u32 blocks_count;
344 __u16 reserved_blocks;
345 __u16 unused;
346 __u32 free_blocks_count;
347};
348
a2df2a63 349/*
c2177057 350 * Flags used by ext4_get_blocks()
a2df2a63 351 */
c2177057
TT
352 /* Allocate any needed blocks and/or convert an unitialized
353 extent to be an initialized ext4 */
2ac3b6e0 354#define EXT4_GET_BLOCKS_CREATE 0x0001
c2177057 355 /* Request the creation of an unitialized extent */
2ac3b6e0 356#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
c2177057
TT
357#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
358 EXT4_GET_BLOCKS_CREATE)
c2177057 359 /* Caller is from the delayed allocation writeout path,
2ac3b6e0
TT
360 so set the magic i_delalloc_reserve_flag after taking the
361 inode allocation semaphore for */
03f5d8bc 362#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
2ac3b6e0
TT
363 /* Call ext4_da_update_reserve_space() after successfully
364 allocating the blocks */
03f5d8bc 365#define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008
0031462b
MC
366 /* caller is from the direct IO path, request to creation of an
367 unitialized extents if not allocated, split the uninitialized
368 extent if blocks has been preallocated already*/
369#define EXT4_GET_BLOCKS_DIO 0x0010
370#define EXT4_GET_BLOCKS_CONVERT 0x0020
371#define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
372 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
373 /* Convert extent to initialized after direct IO complete */
374#define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
375 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
ac27a0ec
DK
376
377/*
378 * ioctl commands
379 */
617ba13b
MC
380#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
381#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
382#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
383#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
384#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
385#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
e23291b9 386#ifdef CONFIG_JBD2_DEBUG
617ba13b 387#define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
ac27a0ec 388#endif
617ba13b
MC
389#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
390#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
391#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
392#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
393#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 394 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 395 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 396#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 397#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
ac27a0ec
DK
398
399/*
400 * ioctl commands in 32 bit emulation
401 */
617ba13b
MC
402#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
403#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
404#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
405#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
406#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
407#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
408#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
e23291b9 409#ifdef CONFIG_JBD2_DEBUG
617ba13b 410#define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
ac27a0ec 411#endif
617ba13b
MC
412#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
413#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
ac27a0ec
DK
414
415
416/*
417 * Mount options
418 */
617ba13b 419struct ext4_mount_options {
ac27a0ec
DK
420 unsigned long s_mount_opt;
421 uid_t s_resuid;
422 gid_t s_resgid;
423 unsigned long s_commit_interval;
30773840 424 u32 s_min_batch_time, s_max_batch_time;
ac27a0ec
DK
425#ifdef CONFIG_QUOTA
426 int s_jquota_fmt;
427 char *s_qf_names[MAXQUOTAS];
428#endif
429};
430
fb0a387d
ES
431/* Max physical block we can addres w/o extents */
432#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
433
ac27a0ec
DK
434/*
435 * Structure of an inode on the disk
436 */
617ba13b 437struct ext4_inode {
ac27a0ec
DK
438 __le16 i_mode; /* File mode */
439 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 440 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 441 __le32 i_atime; /* Access time */
ef7f3835 442 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
443 __le32 i_mtime; /* Modification time */
444 __le32 i_dtime; /* Deletion Time */
445 __le16 i_gid; /* Low 16 bits of Group Id */
446 __le16 i_links_count; /* Links count */
0fc1b451 447 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
448 __le32 i_flags; /* File flags */
449 union {
450 struct {
25ec56b5 451 __le32 l_i_version;
ac27a0ec
DK
452 } linux1;
453 struct {
454 __u32 h_i_translator;
455 } hurd1;
456 struct {
457 __u32 m_i_reserved1;
458 } masix1;
459 } osd1; /* OS dependent 1 */
617ba13b 460 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 461 __le32 i_generation; /* File version (for NFS) */
7973c0c1 462 __le32 i_file_acl_lo; /* File ACL */
a48380f7 463 __le32 i_size_high;
f077d0d7 464 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
465 union {
466 struct {
0fc1b451 467 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 468 __le16 l_i_file_acl_high;
8c55e204 469 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec
DK
470 __le16 l_i_gid_high; /* were reserved2[0] */
471 __u32 l_i_reserved2;
472 } linux2;
473 struct {
f077d0d7 474 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
475 __u16 h_i_mode_high;
476 __u16 h_i_uid_high;
477 __u16 h_i_gid_high;
478 __u32 h_i_author;
479 } hurd2;
480 struct {
f077d0d7 481 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 482 __le16 m_i_file_acl_high;
ac27a0ec
DK
483 __u32 m_i_reserved2[2];
484 } masix2;
485 } osd2; /* OS dependent 2 */
486 __le16 i_extra_isize;
487 __le16 i_pad1;
ef7f3835
KS
488 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
489 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
490 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
491 __le32 i_crtime; /* File Creation time */
492 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 493 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
494};
495
748de673
AF
496struct move_extent {
497 __u32 reserved; /* should be zero */
498 __u32 donor_fd; /* donor file descriptor */
499 __u64 orig_start; /* logical start offset in block for orig */
500 __u64 donor_start; /* logical start offset in block for donor */
501 __u64 len; /* block length to be moved */
502 __u64 moved_len; /* moved block length */
503};
ac27a0ec 504
ef7f3835
KS
505#define EXT4_EPOCH_BITS 2
506#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
507#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
508
509/*
510 * Extended fields will fit into an inode if the filesystem was formatted
511 * with large inodes (-I 256 or larger) and there are not currently any EAs
512 * consuming all of the available space. For new inodes we always reserve
513 * enough space for the kernel's known extended fields, but for inodes
514 * created with an old kernel this might not have been the case. None of
515 * the extended inode fields is critical for correct filesystem operation.
516 * This macro checks if a certain field fits in the inode. Note that
517 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
518 */
519#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
520 ((offsetof(typeof(*ext4_inode), field) + \
521 sizeof((ext4_inode)->field)) \
522 <= (EXT4_GOOD_OLD_INODE_SIZE + \
523 (einode)->i_extra_isize)) \
524
525static inline __le32 ext4_encode_extra_time(struct timespec *time)
526{
527 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
528 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
529 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
530}
531
532static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
533{
534 if (sizeof(time->tv_sec) > 4)
535 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
536 << 32;
c1fccc06 537 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
538}
539
540#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
541do { \
542 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
543 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
544 (raw_inode)->xtime ## _extra = \
545 ext4_encode_extra_time(&(inode)->xtime); \
546} while (0)
547
548#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
549do { \
550 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
551 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
552 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
553 (raw_inode)->xtime ## _extra = \
554 ext4_encode_extra_time(&(einode)->xtime); \
555} while (0)
556
557#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
558do { \
559 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
560 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
561 ext4_decode_extra_time(&(inode)->xtime, \
562 raw_inode->xtime ## _extra); \
563} while (0)
564
565#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
566do { \
567 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
568 (einode)->xtime.tv_sec = \
569 (signed)le32_to_cpu((raw_inode)->xtime); \
570 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
571 ext4_decode_extra_time(&(einode)->xtime, \
572 raw_inode->xtime ## _extra); \
573} while (0)
574
25ec56b5
JNC
575#define i_disk_version osd1.linux1.l_i_version
576
ac27a0ec
DK
577#if defined(__KERNEL__) || defined(__linux__)
578#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 579#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 580#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
581#define i_uid_low i_uid
582#define i_gid_low i_gid
583#define i_uid_high osd2.linux2.l_i_uid_high
584#define i_gid_high osd2.linux2.l_i_gid_high
585#define i_reserved2 osd2.linux2.l_i_reserved2
586
587#elif defined(__GNU__)
588
589#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
590#define i_uid_high osd2.hurd2.h_i_uid_high
591#define i_gid_high osd2.hurd2.h_i_gid_high
592#define i_author osd2.hurd2.h_i_author
593
594#elif defined(__masix__)
595
596#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 597#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
598#define i_reserved2 osd2.masix2.m_i_reserved2
599
600#endif /* defined(__KERNEL__) || defined(__linux__) */
601
d444c3c3
TT
602/*
603 * storage for cached extent
604 */
605struct ext4_ext_cache {
606 ext4_fsblk_t ec_start;
607 ext4_lblk_t ec_block;
608 __u32 ec_len; /* must be 32bit to return holes */
609 __u32 ec_type;
610};
611
612/*
613 * fourth extended file system inode data in memory
614 */
615struct ext4_inode_info {
616 __le32 i_data[15]; /* unconverted */
617 __u32 i_flags;
618 ext4_fsblk_t i_file_acl;
619 __u32 i_dtime;
620
621 /*
622 * i_block_group is the number of the block group which contains
623 * this file's inode. Constant across the lifetime of the inode,
624 * it is ued for making block allocation decisions - we try to
625 * place a file's data blocks near its inode block, and new inodes
626 * near to their parent directory's inode.
627 */
628 ext4_group_t i_block_group;
629 __u32 i_state; /* Dynamic state flags for ext4 */
630
631 ext4_lblk_t i_dir_start_lookup;
632#ifdef CONFIG_EXT4_FS_XATTR
633 /*
634 * Extended attributes can be read independently of the main file
635 * data. Taking i_mutex even when reading would cause contention
636 * between readers of EAs and writers of regular file data, so
637 * instead we synchronize on xattr_sem when reading or changing
638 * EAs.
639 */
640 struct rw_semaphore xattr_sem;
641#endif
d444c3c3
TT
642
643 struct list_head i_orphan; /* unlinked but open inodes */
644
645 /*
646 * i_disksize keeps track of what the inode size is ON DISK, not
647 * in memory. During truncate, i_size is set to the new size by
648 * the VFS prior to calling ext4_truncate(), but the filesystem won't
649 * set i_disksize to 0 until the truncate is actually under way.
650 *
651 * The intent is that i_disksize always represents the blocks which
652 * are used by this file. This allows recovery to restart truncate
653 * on orphans if we crash during truncate. We actually write i_disksize
654 * into the on-disk inode when writing inodes out, instead of i_size.
655 *
656 * The only time when i_disksize and i_size may be different is when
657 * a truncate is in progress. The only things which change i_disksize
658 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
659 */
660 loff_t i_disksize;
661
662 /*
663 * i_data_sem is for serialising ext4_truncate() against
664 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
665 * data tree are chopped off during truncate. We can't do that in
666 * ext4 because whenever we perform intermediate commits during
667 * truncate, the inode and all the metadata blocks *must* be in a
668 * consistent state which allows truncation of the orphans to restart
669 * during recovery. Hence we must fix the get_block-vs-truncate race
670 * by other means, so we have i_data_sem.
671 */
672 struct rw_semaphore i_data_sem;
673 struct inode vfs_inode;
674 struct jbd2_inode jinode;
675
676 struct ext4_ext_cache i_cached_extent;
677 /*
678 * File creation time. Its function is same as that of
679 * struct timespec i_{a,c,m}time in the generic inode.
680 */
681 struct timespec i_crtime;
682
683 /* mballoc */
684 struct list_head i_prealloc_list;
685 spinlock_t i_prealloc_lock;
686
687 /* ialloc */
688 ext4_group_t i_last_alloc_group;
689
690 /* allocation reservation info for delalloc */
691 unsigned int i_reserved_data_blocks;
692 unsigned int i_reserved_meta_blocks;
693 unsigned int i_allocated_meta_blocks;
694 unsigned short i_delalloc_reserved_flag;
695
696 /* on-disk additional length */
697 __u16 i_extra_isize;
698
699 spinlock_t i_block_reservation_lock;
8d5d02e6
MC
700
701 /* completed async DIOs that might need unwritten extents handling */
702 struct list_head i_aio_dio_complete_list;
703 /* current io_end structure for async DIO write*/
704 ext4_io_end_t *cur_aio_dio;
d444c3c3
TT
705};
706
ac27a0ec
DK
707/*
708 * File system states
709 */
617ba13b
MC
710#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
711#define EXT4_ERROR_FS 0x0002 /* Errors detected */
712#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 713
469108ff
TT
714/*
715 * Misc. filesystem flags
716 */
717#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
718#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
719#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
720
ac27a0ec
DK
721/*
722 * Mount flags
723 */
617ba13b
MC
724#define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
725#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
726#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
727#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
728#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
729#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
730#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
731#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
732#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
733#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
734#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
735#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
736#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
737#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
738#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
739#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 740#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b
MC
741#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
742#define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
743#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
744#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
745#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
d4da6c9c 746#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 747#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
25ec56b5 748#define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
64769240 749#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 750#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 751#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5bf5683a 752
617ba13b
MC
753#define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
754#define set_opt(o, opt) o |= EXT4_MOUNT_##opt
755#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
756 EXT4_MOUNT_##opt)
ac27a0ec 757
617ba13b
MC
758#define ext4_set_bit ext2_set_bit
759#define ext4_set_bit_atomic ext2_set_bit_atomic
760#define ext4_clear_bit ext2_clear_bit
761#define ext4_clear_bit_atomic ext2_clear_bit_atomic
762#define ext4_test_bit ext2_test_bit
763#define ext4_find_first_zero_bit ext2_find_first_zero_bit
764#define ext4_find_next_zero_bit ext2_find_next_zero_bit
aa02ad67 765#define ext4_find_next_bit ext2_find_next_bit
ac27a0ec
DK
766
767/*
768 * Maximal mount counts between two filesystem checks
769 */
617ba13b
MC
770#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
771#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
772
773/*
774 * Behaviour when detecting errors
775 */
617ba13b
MC
776#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
777#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
778#define EXT4_ERRORS_PANIC 3 /* Panic */
779#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec
DK
780
781/*
782 * Structure of the super block
783 */
617ba13b 784struct ext4_super_block {
ac27a0ec 785/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 786 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
787 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
788 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
789/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
790 __le32 s_first_data_block; /* First Data Block */
791 __le32 s_log_block_size; /* Block size */
f077d0d7 792 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
ac27a0ec 793/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
f077d0d7 794 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
ac27a0ec
DK
795 __le32 s_inodes_per_group; /* # Inodes per group */
796 __le32 s_mtime; /* Mount time */
797/*30*/ __le32 s_wtime; /* Write time */
798 __le16 s_mnt_count; /* Mount count */
799 __le16 s_max_mnt_count; /* Maximal mount count */
800 __le16 s_magic; /* Magic signature */
801 __le16 s_state; /* File system state */
802 __le16 s_errors; /* Behaviour when detecting errors */
803 __le16 s_minor_rev_level; /* minor revision level */
804/*40*/ __le32 s_lastcheck; /* time of last check */
805 __le32 s_checkinterval; /* max. time between checks */
806 __le32 s_creator_os; /* OS */
807 __le32 s_rev_level; /* Revision level */
808/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
809 __le16 s_def_resgid; /* Default gid for reserved blocks */
810 /*
617ba13b 811 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
812 *
813 * Note: the difference between the compatible feature set and
814 * the incompatible feature set is that if there is a bit set
815 * in the incompatible feature set that the kernel doesn't
816 * know about, it should refuse to mount the filesystem.
817 *
818 * e2fsck's requirements are more strict; if it doesn't know
819 * about a feature in either the compatible or incompatible
820 * feature set, it must abort and not try to meddle with
821 * things it doesn't understand...
822 */
823 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 824 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
825 __le16 s_block_group_nr; /* block group # of this superblock */
826 __le32 s_feature_compat; /* compatible feature set */
827/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
828 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
829/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
830/*78*/ char s_volume_name[16]; /* volume name */
831/*88*/ char s_last_mounted[64]; /* directory where last mounted */
832/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
833 /*
834 * Performance hints. Directory preallocation should only
617ba13b 835 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
836 */
837 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
838 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
839 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
840 /*
617ba13b 841 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
842 */
843/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
844/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
845 __le32 s_journal_dev; /* device number of journal file */
846 __le32 s_last_orphan; /* start of list of inodes to delete */
847 __le32 s_hash_seed[4]; /* HTREE hash seed */
848 __u8 s_def_hash_version; /* Default hash version to use */
849 __u8 s_reserved_char_pad;
0d1ee42f 850 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 851/*100*/ __le32 s_default_mount_opts;
ac27a0ec 852 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
853 __le32 s_mkfs_time; /* When the filesystem was created */
854 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
855 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
856/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
857 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
858 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
859 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
860 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
861 __le32 s_flags; /* Miscellaneous flags */
862 __le16 s_raid_stride; /* RAID stride */
863 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */
864 __le64 s_mmp_block; /* Block for multi-mount protection */
865 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8
JS
866 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
867 __u8 s_reserved_char_pad2;
868 __le16 s_reserved_pad;
afc32f7e
TT
869 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
870 __u32 s_reserved[160]; /* Padding to the end of the block */
ac27a0ec
DK
871};
872
873#ifdef __KERNEL__
bc0b0d6d
TT
874
875/*
4ab2f15b 876 * run-time mount flags
bc0b0d6d
TT
877 */
878#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 879#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 880
ca0faba0
TT
881/*
882 * fourth extended-fs super-block data in memory
883 */
884struct ext4_sb_info {
885 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
886 unsigned long s_inodes_per_block;/* Number of inodes per block */
887 unsigned long s_blocks_per_group;/* Number of blocks in a group */
888 unsigned long s_inodes_per_group;/* Number of inodes in a group */
889 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
890 unsigned long s_gdb_count; /* Number of group descriptor blocks */
891 unsigned long s_desc_per_block; /* Number of group descriptors per block */
892 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 893 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
ca0faba0
TT
894 unsigned long s_overhead_last; /* Last calculated overhead */
895 unsigned long s_blocks_last; /* Last seen block count */
896 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
897 struct buffer_head * s_sbh; /* Buffer containing the super block */
898 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
899 struct buffer_head **s_group_desc;
7f4520cc 900 unsigned int s_mount_opt;
bc0b0d6d 901 unsigned int s_mount_flags;
ca0faba0
TT
902 ext4_fsblk_t s_sb_block;
903 uid_t s_resuid;
904 gid_t s_resgid;
905 unsigned short s_mount_state;
906 unsigned short s_pad;
907 int s_addr_per_block_bits;
908 int s_desc_per_block_bits;
909 int s_inode_size;
910 int s_first_ino;
911 unsigned int s_inode_readahead_blks;
11013911 912 unsigned int s_inode_goal;
ca0faba0
TT
913 spinlock_t s_next_gen_lock;
914 u32 s_next_generation;
915 u32 s_hash_seed[4];
916 int s_def_hash_version;
917 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
918 struct percpu_counter s_freeblocks_counter;
919 struct percpu_counter s_freeinodes_counter;
920 struct percpu_counter s_dirs_counter;
921 struct percpu_counter s_dirtyblocks_counter;
922 struct blockgroup_lock *s_blockgroup_lock;
923 struct proc_dir_entry *s_proc;
924 struct kobject s_kobj;
925 struct completion s_kobj_unregister;
926
927 /* Journaling */
928 struct inode *s_journal_inode;
929 struct journal_s *s_journal;
930 struct list_head s_orphan;
931 struct mutex s_orphan_lock;
932 struct mutex s_resize_lock;
933 unsigned long s_commit_interval;
934 u32 s_max_batch_time;
935 u32 s_min_batch_time;
936 struct block_device *journal_bdev;
937#ifdef CONFIG_JBD2_DEBUG
938 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
939 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
940#endif
941#ifdef CONFIG_QUOTA
942 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
943 int s_jquota_fmt; /* Format of quota to use */
944#endif
945 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 946 struct rb_root system_blks;
ca0faba0
TT
947
948#ifdef EXTENTS_STATS
949 /* ext4 extents stats */
950 unsigned long s_ext_min;
951 unsigned long s_ext_max;
952 unsigned long s_depth_max;
953 spinlock_t s_ext_stats_lock;
954 unsigned long s_ext_blocks;
955 unsigned long s_ext_extents;
956#endif
957
958 /* for buddy allocator */
959 struct ext4_group_info ***s_group_info;
960 struct inode *s_buddy_cache;
961 long s_blocks_reserved;
962 spinlock_t s_reserve_lock;
963 spinlock_t s_md_lock;
964 tid_t s_last_transaction;
965 unsigned short *s_mb_offsets;
966 unsigned int *s_mb_maxs;
967
968 /* tunables */
969 unsigned long s_stripe;
970 unsigned int s_mb_stream_request;
971 unsigned int s_mb_max_to_scan;
972 unsigned int s_mb_min_to_scan;
973 unsigned int s_mb_stats;
974 unsigned int s_mb_order2_reqs;
975 unsigned int s_mb_group_prealloc;
55138e0b 976 unsigned int s_max_writeback_mb_bump;
ca0faba0
TT
977 /* where last allocation was done - for stream allocation */
978 unsigned long s_mb_last_group;
979 unsigned long s_mb_last_start;
980
ca0faba0
TT
981 /* stats for buddy allocator */
982 spinlock_t s_mb_pa_lock;
983 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
984 atomic_t s_bal_success; /* we found long enough chunks */
985 atomic_t s_bal_allocated; /* in blocks */
986 atomic_t s_bal_ex_scanned; /* total extents scanned */
987 atomic_t s_bal_goals; /* goal hits */
988 atomic_t s_bal_breaks; /* too long searches */
989 atomic_t s_bal_2orders; /* 2^order hits */
990 spinlock_t s_bal_lock;
991 unsigned long s_mb_buddies_generated;
992 unsigned long long s_mb_generation_time;
993 atomic_t s_mb_lost_chunks;
994 atomic_t s_mb_preallocated;
995 atomic_t s_mb_discarded;
50797481 996 atomic_t s_lock_busy;
ca0faba0
TT
997
998 /* locality groups */
999 struct ext4_locality_group *s_locality_groups;
1000
1001 /* for write statistics */
1002 unsigned long s_sectors_written_start;
1003 u64 s_kbytes_written;
1004
1005 unsigned int s_log_groups_per_flex;
1006 struct flex_groups *s_flex_groups;
4c0425ff
MC
1007
1008 /* workqueue for dio unwritten */
1009 struct workqueue_struct *dio_unwritten_wq;
ca0faba0
TT
1010};
1011
af5bc92d 1012static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1013{
1014 return sb->s_fs_info;
1015}
617ba13b 1016static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1017{
617ba13b 1018 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1019}
1020
ef7f3835
KS
1021static inline struct timespec ext4_current_time(struct inode *inode)
1022{
1023 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1024 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1025}
1026
617ba13b 1027static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1028{
617ba13b
MC
1029 return ino == EXT4_ROOT_INO ||
1030 ino == EXT4_JOURNAL_INO ||
1031 ino == EXT4_RESIZE_INO ||
1032 (ino >= EXT4_FIRST_INO(sb) &&
1033 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec
DK
1034}
1035#else
617ba13b 1036/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1037 * a kernel struct super_block. This will allow us to call the feature-test
1038 * macros from user land. */
617ba13b 1039#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1040#endif
1041
617ba13b 1042#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1043
1044/*
1045 * Codes for operating systems
1046 */
617ba13b
MC
1047#define EXT4_OS_LINUX 0
1048#define EXT4_OS_HURD 1
1049#define EXT4_OS_MASIX 2
1050#define EXT4_OS_FREEBSD 3
1051#define EXT4_OS_LITES 4
ac27a0ec
DK
1052
1053/*
1054 * Revision levels
1055 */
617ba13b
MC
1056#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1057#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1058
617ba13b
MC
1059#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1060#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1061
617ba13b 1062#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1063
1064/*
1065 * Feature set definitions
1066 */
1067
617ba13b 1068#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1069 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1070#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1071 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1072#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1073 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1074#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1075 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1076#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1077 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1078#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1079 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1080#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1081 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1082#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1083 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1084#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1085 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1086
1087#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1088#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1089#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1090#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1091#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1092#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1093
1094#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1095#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1096#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1097#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1098#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1099#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1100#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
617ba13b
MC
1101
1102#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1103#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1104#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1105#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1106#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1107#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1108#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1109#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1110#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
617ba13b
MC
1111
1112#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1113#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1114 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1115 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1116 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581
JS
1117 EXT4_FEATURE_INCOMPAT_64BIT| \
1118 EXT4_FEATURE_INCOMPAT_FLEX_BG)
617ba13b
MC
1119#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1120 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1121 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1122 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1123 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451
AK
1124 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1125 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
ac27a0ec
DK
1126
1127/*
1128 * Default values for user and/or group using reserved blocks
1129 */
617ba13b
MC
1130#define EXT4_DEF_RESUID 0
1131#define EXT4_DEF_RESGID 0
ac27a0ec 1132
240799cd
TT
1133#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1134
ac27a0ec
DK
1135/*
1136 * Default mount options
1137 */
617ba13b
MC
1138#define EXT4_DEFM_DEBUG 0x0001
1139#define EXT4_DEFM_BSDGROUPS 0x0002
1140#define EXT4_DEFM_XATTR_USER 0x0004
1141#define EXT4_DEFM_ACL 0x0008
1142#define EXT4_DEFM_UID16 0x0010
1143#define EXT4_DEFM_JMODE 0x0060
1144#define EXT4_DEFM_JMODE_DATA 0x0020
1145#define EXT4_DEFM_JMODE_ORDERED 0x0040
1146#define EXT4_DEFM_JMODE_WBACK 0x0060
ac27a0ec 1147
30773840
TT
1148/*
1149 * Default journal batch times
1150 */
1151#define EXT4_DEF_MIN_BATCH_TIME 0
1152#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1153
a4912123
TT
1154/*
1155 * Minimum number of groups in a flexgroup before we separate out
1156 * directories into the first block group of a flexgroup
1157 */
1158#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1159
ac27a0ec
DK
1160/*
1161 * Structure of a directory entry
1162 */
617ba13b 1163#define EXT4_NAME_LEN 255
ac27a0ec 1164
617ba13b 1165struct ext4_dir_entry {
ac27a0ec
DK
1166 __le32 inode; /* Inode number */
1167 __le16 rec_len; /* Directory entry length */
1168 __le16 name_len; /* Name length */
617ba13b 1169 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1170};
1171
1172/*
617ba13b 1173 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1174 * stored in intel byte order, and the name_len field could never be
1175 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1176 * file_type field.
1177 */
617ba13b 1178struct ext4_dir_entry_2 {
ac27a0ec
DK
1179 __le32 inode; /* Inode number */
1180 __le16 rec_len; /* Directory entry length */
1181 __u8 name_len; /* Name length */
1182 __u8 file_type;
617ba13b 1183 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1184};
1185
1186/*
617ba13b 1187 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1188 * other bits are reserved for now.
1189 */
617ba13b
MC
1190#define EXT4_FT_UNKNOWN 0
1191#define EXT4_FT_REG_FILE 1
1192#define EXT4_FT_DIR 2
1193#define EXT4_FT_CHRDEV 3
1194#define EXT4_FT_BLKDEV 4
1195#define EXT4_FT_FIFO 5
1196#define EXT4_FT_SOCK 6
1197#define EXT4_FT_SYMLINK 7
ac27a0ec 1198
617ba13b 1199#define EXT4_FT_MAX 8
ac27a0ec
DK
1200
1201/*
617ba13b 1202 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1203 *
1204 * NOTE: It must be a multiple of 4
1205 */
617ba13b
MC
1206#define EXT4_DIR_PAD 4
1207#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1208#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1209 ~EXT4_DIR_ROUND)
a72d7f83
JK
1210#define EXT4_MAX_REC_LEN ((1<<16)-1)
1211
ac27a0ec
DK
1212/*
1213 * Hash Tree Directory indexing
1214 * (c) Daniel Phillips, 2001
1215 */
1216
4074fe37
ES
1217#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1218 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
617ba13b
MC
1219 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1220#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1221#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1222
1223/* Legal values for the dx_root hash_version field: */
1224
1225#define DX_HASH_LEGACY 0
1226#define DX_HASH_HALF_MD4 1
1227#define DX_HASH_TEA 2
f99b2589
TT
1228#define DX_HASH_LEGACY_UNSIGNED 3
1229#define DX_HASH_HALF_MD4_UNSIGNED 4
1230#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec
DK
1231
1232#ifdef __KERNEL__
1233
1234/* hash info structure used by the directory hash */
1235struct dx_hash_info
1236{
1237 u32 hash;
1238 u32 minor_hash;
1239 int hash_version;
1240 u32 *seed;
1241};
1242
617ba13b 1243#define EXT4_HTREE_EOF 0x7fffffff
ac27a0ec
DK
1244
1245/*
617ba13b 1246 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1247 */
1248#define HASH_NB_ALWAYS 1
1249
1250
1251/*
1252 * Describe an inode's exact location on disk and in memory
1253 */
617ba13b 1254struct ext4_iloc
ac27a0ec
DK
1255{
1256 struct buffer_head *bh;
1257 unsigned long offset;
fd2d4291 1258 ext4_group_t block_group;
ac27a0ec
DK
1259};
1260
617ba13b 1261static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1262{
617ba13b 1263 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1264}
1265
1266/*
1267 * This structure is stuffed into the struct file's private_data field
1268 * for directories. It is where we put information so that we can do
1269 * readdir operations in hash tree order.
1270 */
1271struct dir_private_info {
1272 struct rb_root root;
1273 struct rb_node *curr_node;
1274 struct fname *extra_fname;
1275 loff_t last_pos;
1276 __u32 curr_hash;
1277 __u32 curr_minor_hash;
1278 __u32 next_hash;
1279};
1280
1281/* calculate the first block number of the group */
617ba13b 1282static inline ext4_fsblk_t
fd2d4291 1283ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1284{
617ba13b
MC
1285 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1286 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1287}
1288
1289/*
1290 * Special error return code only used by dx_probe() and its callers.
1291 */
1292#define ERR_BAD_DX_DIR -75000
1293
72b64b59 1294void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
a9df9a49 1295 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
3a5b2ecd 1296
9f6200bb
TT
1297extern struct proc_dir_entry *ext4_proc_root;
1298
ac27a0ec
DK
1299/*
1300 * Function prototypes
1301 */
1302
1303/*
1304 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1305 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1306 */
8c55e204
DK
1307# define NORET_TYPE /**/
1308# define ATTRIB_NORET __attribute__((noreturn))
1309# define NORET_AND noreturn,
ac27a0ec 1310
498e5f24
TT
1311/* bitmap.c */
1312extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1313
ac27a0ec 1314/* balloc.c */
3a5b2ecd
MC
1315extern unsigned int ext4_block_group(struct super_block *sb,
1316 ext4_fsblk_t blocknr);
1317extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1318 ext4_fsblk_t blocknr);
fd2d4291
AM
1319extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1320extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1321 ext4_group_t group);
7061eba7 1322extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
617ba13b 1323 ext4_fsblk_t goal, unsigned long *count, int *errp);
5c791616 1324extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
8c3bf8a0 1325extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
af5bc92d 1326extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
c9de560d 1327 ext4_fsblk_t block, unsigned long count, int metadata);
e21675d4
AK
1328extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1329 ext4_fsblk_t block, unsigned long count);
af5bc92d
TT
1330extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1331extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1332extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1333 ext4_group_t block_group,
ac27a0ec 1334 struct buffer_head ** bh);
617ba13b 1335extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
bb23c20a
TT
1336struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1337 ext4_group_t block_group);
1338extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1339 struct buffer_head *bh,
1340 ext4_group_t group,
1341 struct ext4_group_desc *desc);
1342#define ext4_free_blocks_after_init(sb, group, desc) \
1343 ext4_init_block_bitmap(sb, NULL, group, desc)
ac27a0ec
DK
1344
1345/* dir.c */
617ba13b
MC
1346extern int ext4_check_dir_entry(const char *, struct inode *,
1347 struct ext4_dir_entry_2 *,
498e5f24 1348 struct buffer_head *, unsigned int);
617ba13b 1349extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 1350 __u32 minor_hash,
617ba13b
MC
1351 struct ext4_dir_entry_2 *dirent);
1352extern void ext4_htree_free_dir_info(struct dir_private_info *p);
ac27a0ec
DK
1353
1354/* fsync.c */
af5bc92d 1355extern int ext4_sync_file(struct file *, struct dentry *, int);
ac27a0ec
DK
1356
1357/* hash.c */
617ba13b 1358extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
1359 dx_hash_info *hinfo);
1360
1361/* ialloc.c */
f157a4aa 1362extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
11013911 1363 const struct qstr *qstr, __u32 goal);
af5bc92d
TT
1364extern void ext4_free_inode(handle_t *, struct inode *);
1365extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1366extern unsigned long ext4_count_free_inodes(struct super_block *);
1367extern unsigned long ext4_count_dirs(struct super_block *);
1368extern void ext4_check_inodes_bitmap(struct super_block *);
bb23c20a
TT
1369extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1370 struct buffer_head *bh,
1371 ext4_group_t group,
1372 struct ext4_group_desc *desc);
1373extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
ac27a0ec 1374
c9de560d
AT
1375/* mballoc.c */
1376extern long ext4_mb_stats;
1377extern long ext4_mb_max_to_scan;
1378extern int ext4_mb_init(struct super_block *, int);
1379extern int ext4_mb_release(struct super_block *);
1380extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1381 struct ext4_allocation_request *, int *);
1382extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 1383extern void ext4_discard_preallocations(struct inode *);
c9de560d
AT
1384extern int __init init_ext4_mballoc(void);
1385extern void exit_ext4_mballoc(void);
1386extern void ext4_mb_free_blocks(handle_t *, struct inode *,
0610b6e9 1387 ext4_fsblk_t, unsigned long, int, unsigned long *);
920313a7 1388extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 1389 ext4_group_t i, struct ext4_group_desc *desc);
920313a7
AK
1390extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1391extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1392 ext4_group_t, int);
ac27a0ec 1393/* inode.c */
617ba13b
MC
1394int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
1395 struct buffer_head *bh, ext4_fsblk_t blocknr);
725d26d3
AK
1396struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1397 ext4_lblk_t, int, int *);
1398struct buffer_head *ext4_bread(handle_t *, struct inode *,
1399 ext4_lblk_t, int, int *);
6873fa0d
ES
1400int ext4_get_block(struct inode *inode, sector_t iblock,
1401 struct buffer_head *bh_result, int create);
ac27a0ec 1402
1d1fe1ee 1403extern struct inode *ext4_iget(struct super_block *, unsigned long);
af5bc92d
TT
1404extern int ext4_write_inode(struct inode *, int);
1405extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
1406extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1407 struct kstat *stat);
af5bc92d
TT
1408extern void ext4_delete_inode(struct inode *);
1409extern int ext4_sync_inode(handle_t *, struct inode *);
617ba13b
MC
1410extern void ext4_dirty_inode(struct inode *);
1411extern int ext4_change_inode_journal_flag(struct inode *, int);
1412extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
91ef4caf 1413extern int ext4_can_truncate(struct inode *inode);
af5bc92d 1414extern void ext4_truncate(struct inode *);
487caeef 1415extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 1416extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 1417extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 1418extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 1419extern void ext4_set_aops(struct inode *inode);
a86c6181 1420extern int ext4_writepage_trans_blocks(struct inode *);
a02908f1 1421extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
f3bd1f3f 1422extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
cf108bca 1423extern int ext4_block_truncate_page(handle_t *handle,
a86c6181 1424 struct address_space *mapping, loff_t from);
c2ec175c 1425extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
60e58e0f 1426extern qsize_t ext4_get_reserved_space(struct inode *inode);
8d5d02e6 1427extern int flush_aio_dio_completed_IO(struct inode *inode);
ac27a0ec 1428/* ioctl.c */
5cdd7b2d 1429extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 1430extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 1431
c14c6fd5 1432/* migrate.c */
2a43a878 1433extern int ext4_ext_migrate(struct inode *);
3d0518f4 1434
ac27a0ec 1435/* namei.c */
3d0518f4
WY
1436extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1437extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
617ba13b
MC
1438extern int ext4_orphan_add(handle_t *, struct inode *);
1439extern int ext4_orphan_del(handle_t *, struct inode *);
1440extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec
DK
1441 __u32 start_minor_hash, __u32 *next_hash);
1442
1443/* resize.c */
617ba13b
MC
1444extern int ext4_group_add(struct super_block *sb,
1445 struct ext4_new_group_data *input);
1446extern int ext4_group_extend(struct super_block *sb,
1447 struct ext4_super_block *es,
1448 ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
1449
1450/* super.c */
af5bc92d 1451extern void ext4_error(struct super_block *, const char *, const char *, ...)
ac27a0ec 1452 __attribute__ ((format (printf, 3, 4)));
af5bc92d
TT
1453extern void __ext4_std_error(struct super_block *, const char *, int);
1454extern void ext4_abort(struct super_block *, const char *, const char *, ...)
ac27a0ec 1455 __attribute__ ((format (printf, 3, 4)));
af5bc92d 1456extern void ext4_warning(struct super_block *, const char *, const char *, ...)
ac27a0ec 1457 __attribute__ ((format (printf, 3, 4)));
b31e1552
ES
1458extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1459 __attribute__ ((format (printf, 3, 4)));
5d1b1b3f
AK
1460extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1461 const char *, const char *, ...)
1462 __attribute__ ((format (printf, 4, 5)));
af5bc92d 1463extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
1464extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1465 __u32 compat);
1466extern int ext4_update_rocompat_feature(handle_t *handle,
1467 struct super_block *sb, __u32 rocompat);
1468extern int ext4_update_incompat_feature(handle_t *handle,
1469 struct super_block *sb, __u32 incompat);
8fadc143
AR
1470extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1471 struct ext4_group_desc *bg);
1472extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1473 struct ext4_group_desc *bg);
1474extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1475 struct ext4_group_desc *bg);
560671a0
AK
1476extern __u32 ext4_free_blks_count(struct super_block *sb,
1477 struct ext4_group_desc *bg);
1478extern __u32 ext4_free_inodes_count(struct super_block *sb,
1479 struct ext4_group_desc *bg);
1480extern __u32 ext4_used_dirs_count(struct super_block *sb,
1481 struct ext4_group_desc *bg);
1482extern __u32 ext4_itable_unused_count(struct super_block *sb,
1483 struct ext4_group_desc *bg);
8fadc143
AR
1484extern void ext4_block_bitmap_set(struct super_block *sb,
1485 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1486extern void ext4_inode_bitmap_set(struct super_block *sb,
1487 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1488extern void ext4_inode_table_set(struct super_block *sb,
1489 struct ext4_group_desc *bg, ext4_fsblk_t blk);
560671a0
AK
1490extern void ext4_free_blks_set(struct super_block *sb,
1491 struct ext4_group_desc *bg, __u32 count);
1492extern void ext4_free_inodes_set(struct super_block *sb,
1493 struct ext4_group_desc *bg, __u32 count);
1494extern void ext4_used_dirs_set(struct super_block *sb,
1495 struct ext4_group_desc *bg, __u32 count);
1496extern void ext4_itable_unused_set(struct super_block *sb,
1497 struct ext4_group_desc *bg, __u32 count);
bb23c20a
TT
1498extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1499 struct ext4_group_desc *gdp);
1500extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1501 struct ext4_group_desc *gdp);
bd81d8ee
LV
1502
1503static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1504{
1505 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 1506 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
1507}
1508
1509static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1510{
1511 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 1512 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
1513}
1514
1515static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1516{
1517 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 1518 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
1519}
1520
1521static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1522 ext4_fsblk_t blk)
1523{
6bc9feff 1524 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1525 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1526}
1527
1528static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1529 ext4_fsblk_t blk)
1530{
308ba3ec 1531 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1532 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1533}
1534
1535static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1536 ext4_fsblk_t blk)
1537{
308ba3ec 1538 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1539 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1540}
1541
a48380f7
AK
1542static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1543{
06a279d6
TT
1544 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1545 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1546 le32_to_cpu(raw_inode->i_size_lo);
1547 else
1548 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 1549}
bd81d8ee 1550
a48380f7
AK
1551static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1552{
1553 raw_inode->i_size_lo = cpu_to_le32(i_size);
1554 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1555}
ac27a0ec 1556
c9de560d
AT
1557static inline
1558struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 1559 ext4_group_t group)
c9de560d
AT
1560{
1561 struct ext4_group_info ***grp_info;
1562 long indexv, indexh;
1563 grp_info = EXT4_SB(sb)->s_group_info;
1564 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1565 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1566 return grp_info[indexv][indexh];
1567}
1568
8df9675f
TT
1569/*
1570 * Reading s_groups_count requires using smp_rmb() afterwards. See
1571 * the locking protocol documented in the comments of ext4_group_add()
1572 * in resize.c
1573 */
1574static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1575{
1576 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
1577
1578 smp_rmb();
1579 return ngroups;
1580}
c9de560d 1581
772cb7c8
JS
1582static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1583 ext4_group_t block_group)
1584{
1585 return block_group >> sbi->s_log_groups_per_flex;
1586}
1587
1588static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1589{
1590 return 1 << sbi->s_log_groups_per_flex;
1591}
1592
617ba13b 1593#define ext4_std_error(sb, errno) \
ac27a0ec
DK
1594do { \
1595 if ((errno)) \
4db9c54a 1596 __ext4_std_error((sb), __func__, (errno)); \
ac27a0ec
DK
1597} while (0)
1598
a30d542a 1599#ifdef CONFIG_SMP
179f7ebf 1600/* Each CPU can accumulate percpu_counter_batch blocks in their local
a30d542a 1601 * counters. So we need to make sure we have free blocks more
179f7ebf 1602 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 1603 */
179f7ebf 1604#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a
AK
1605#else
1606#define EXT4_FREEBLOCKS_WATERMARK 0
1607#endif
1608
cf17fea6
AK
1609static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1610{
1611 /*
1612 * XXX: replace with spinlock if seen contended -bzzz
1613 */
1614 down_write(&EXT4_I(inode)->i_data_sem);
1615 if (newsize > EXT4_I(inode)->i_disksize)
1616 EXT4_I(inode)->i_disksize = newsize;
1617 up_write(&EXT4_I(inode)->i_data_sem);
1618 return ;
1619}
a30d542a 1620
5d1b1b3f
AK
1621struct ext4_group_info {
1622 unsigned long bb_state;
1623 struct rb_root bb_free_root;
a36b4498
ES
1624 ext4_grpblk_t bb_first_free; /* first free block */
1625 ext4_grpblk_t bb_free; /* total free blocks */
1626 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
5d1b1b3f
AK
1627 struct list_head bb_prealloc_list;
1628#ifdef DOUBLE_CHECK
1629 void *bb_bitmap;
1630#endif
1631 struct rw_semaphore alloc_sem;
a36b4498
ES
1632 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
1633 * regions, index is order.
1634 * bb_counters[3] = 5 means
1635 * 5 free 8-block regions. */
5d1b1b3f
AK
1636};
1637
1638#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
5d1b1b3f
AK
1639
1640#define EXT4_MB_GRP_NEED_INIT(grp) \
1641 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1642
50797481
TT
1643#define EXT4_MAX_CONTENTION 8
1644#define EXT4_CONTENTION_THRESHOLD 2
1645
955ce5f5
AK
1646static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1647 ext4_group_t group)
5d1b1b3f 1648{
955ce5f5
AK
1649 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1650}
5d1b1b3f 1651
50797481
TT
1652/*
1653 * Returns true if the filesystem is busy enough that attempts to
1654 * access the block group locks has run into contention.
1655 */
1656static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1657{
1658 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1659}
1660
955ce5f5
AK
1661static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1662{
50797481
TT
1663 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1664 if (spin_trylock(lock))
1665 /*
1666 * We're able to grab the lock right away, so drop the
1667 * lock contention counter.
1668 */
1669 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1670 else {
1671 /*
1672 * The lock is busy, so bump the contention counter,
1673 * and then wait on the spin lock.
1674 */
1675 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1676 EXT4_MAX_CONTENTION);
1677 spin_lock(lock);
1678 }
5d1b1b3f
AK
1679}
1680
1681static inline void ext4_unlock_group(struct super_block *sb,
1682 ext4_group_t group)
1683{
955ce5f5 1684 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
1685}
1686
ac27a0ec
DK
1687/*
1688 * Inodes and files operations
1689 */
1690
1691/* dir.c */
617ba13b 1692extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
1693
1694/* file.c */
c5ef1c42 1695extern const struct inode_operations ext4_file_inode_operations;
617ba13b 1696extern const struct file_operations ext4_file_operations;
ac27a0ec
DK
1697
1698/* namei.c */
c5ef1c42
AV
1699extern const struct inode_operations ext4_dir_inode_operations;
1700extern const struct inode_operations ext4_special_inode_operations;
596397b7 1701extern struct dentry *ext4_get_parent(struct dentry *child);
ac27a0ec
DK
1702
1703/* symlink.c */
c5ef1c42
AV
1704extern const struct inode_operations ext4_symlink_inode_operations;
1705extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 1706
6fd058f7
TT
1707/* block_validity */
1708extern void ext4_release_system_zone(struct super_block *sb);
1709extern int ext4_setup_system_zone(struct super_block *sb);
1710extern int __init init_ext4_system_zone(void);
1711extern void exit_ext4_system_zone(void);
1712extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1713 ext4_fsblk_t start_blk,
1714 unsigned int count);
1715
a86c6181
AT
1716/* extents.c */
1717extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1718extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
a02908f1
MC
1719extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1720 int chunk);
a86c6181 1721extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
498e5f24 1722 ext4_lblk_t iblock, unsigned int max_blocks,
c2177057 1723 struct buffer_head *bh_result, int flags);
cf108bca 1724extern void ext4_ext_truncate(struct inode *);
a86c6181
AT
1725extern void ext4_ext_init(struct super_block *);
1726extern void ext4_ext_release(struct super_block *);
a2df2a63
AA
1727extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1728 loff_t len);
0031462b
MC
1729extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1730 loff_t len);
12b7ac17
TT
1731extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1732 sector_t block, unsigned int max_blocks,
c2177057 1733 struct buffer_head *bh, int flags);
3a06d778
AK
1734extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1735 __u64 start, __u64 len);
748de673
AF
1736/* move_extent.c */
1737extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1738 __u64 start_orig, __u64 start_donor,
1739 __u64 len, __u64 *moved_len);
1740
3a06d778 1741
2ccb5fb9
AK
1742/*
1743 * Add new method to test wether block and inode bitmaps are properly
1744 * initialized. With uninit_bg reading the block from disk is not enough
1745 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1746 */
1747#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1748
1749static inline int bitmap_uptodate(struct buffer_head *bh)
1750{
1751 return (buffer_uptodate(bh) &&
1752 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1753}
1754static inline void set_bitmap_uptodate(struct buffer_head *bh)
1755{
1756 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1757}
1758
ac27a0ec
DK
1759#endif /* __KERNEL__ */
1760
3dcf5451 1761#endif /* _EXT4_H */
This page took 0.443948 seconds and 5 git commands to generate.