]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - fs/ext4/ext4.h
ext4: Fix quota accounting error with fallocate
[mirror_ubuntu-artful-kernel.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 */
5f524950 325#define EXT4_STATE_DIO_UNWRITTEN 0x00000040 /* need convert on dio done*/
ac27a0ec
DK
326
327/* Used to pass group descriptor data when online resize is done */
617ba13b 328struct ext4_new_group_input {
8c55e204
DK
329 __u32 group; /* Group number for this data */
330 __u64 block_bitmap; /* Absolute block number of block bitmap */
331 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
332 __u64 inode_table; /* Absolute block number of inode table start */
333 __u32 blocks_count; /* Total number of blocks in this group */
334 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
335 __u16 unused;
336};
337
617ba13b
MC
338/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
339struct ext4_new_group_data {
ac27a0ec 340 __u32 group;
bd81d8ee
LV
341 __u64 block_bitmap;
342 __u64 inode_bitmap;
343 __u64 inode_table;
ac27a0ec
DK
344 __u32 blocks_count;
345 __u16 reserved_blocks;
346 __u16 unused;
347 __u32 free_blocks_count;
348};
349
a2df2a63 350/*
c2177057 351 * Flags used by ext4_get_blocks()
a2df2a63 352 */
c2177057
TT
353 /* Allocate any needed blocks and/or convert an unitialized
354 extent to be an initialized ext4 */
2ac3b6e0 355#define EXT4_GET_BLOCKS_CREATE 0x0001
c2177057 356 /* Request the creation of an unitialized extent */
2ac3b6e0 357#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
c2177057
TT
358#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
359 EXT4_GET_BLOCKS_CREATE)
c2177057 360 /* Caller is from the delayed allocation writeout path,
2ac3b6e0
TT
361 so set the magic i_delalloc_reserve_flag after taking the
362 inode allocation semaphore for */
03f5d8bc 363#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
2ac3b6e0
TT
364 /* Call ext4_da_update_reserve_space() after successfully
365 allocating the blocks */
03f5d8bc 366#define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008
0031462b
MC
367 /* caller is from the direct IO path, request to creation of an
368 unitialized extents if not allocated, split the uninitialized
369 extent if blocks has been preallocated already*/
370#define EXT4_GET_BLOCKS_DIO 0x0010
371#define EXT4_GET_BLOCKS_CONVERT 0x0020
372#define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
373 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
374 /* Convert extent to initialized after direct IO complete */
375#define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
376 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
ac27a0ec 377
e6362609
TT
378/*
379 * Flags used by ext4_free_blocks
380 */
381#define EXT4_FREE_BLOCKS_METADATA 0x0001
382#define EXT4_FREE_BLOCKS_FORGET 0x0002
383
ac27a0ec
DK
384/*
385 * ioctl commands
386 */
617ba13b
MC
387#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
388#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
389#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
390#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
391#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
392#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
e23291b9 393#ifdef CONFIG_JBD2_DEBUG
617ba13b 394#define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
ac27a0ec 395#endif
617ba13b
MC
396#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
397#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
398#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
399#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
400#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 401 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 402 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 403#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 404#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
ac27a0ec
DK
405
406/*
407 * ioctl commands in 32 bit emulation
408 */
617ba13b
MC
409#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
410#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
411#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
412#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
413#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
414#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
415#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
e23291b9 416#ifdef CONFIG_JBD2_DEBUG
617ba13b 417#define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
ac27a0ec 418#endif
617ba13b
MC
419#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
420#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
ac27a0ec
DK
421
422
423/*
424 * Mount options
425 */
617ba13b 426struct ext4_mount_options {
ac27a0ec
DK
427 unsigned long s_mount_opt;
428 uid_t s_resuid;
429 gid_t s_resgid;
430 unsigned long s_commit_interval;
30773840 431 u32 s_min_batch_time, s_max_batch_time;
ac27a0ec
DK
432#ifdef CONFIG_QUOTA
433 int s_jquota_fmt;
434 char *s_qf_names[MAXQUOTAS];
435#endif
436};
437
fb0a387d
ES
438/* Max physical block we can addres w/o extents */
439#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
440
ac27a0ec
DK
441/*
442 * Structure of an inode on the disk
443 */
617ba13b 444struct ext4_inode {
ac27a0ec
DK
445 __le16 i_mode; /* File mode */
446 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 447 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 448 __le32 i_atime; /* Access time */
ef7f3835 449 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
450 __le32 i_mtime; /* Modification time */
451 __le32 i_dtime; /* Deletion Time */
452 __le16 i_gid; /* Low 16 bits of Group Id */
453 __le16 i_links_count; /* Links count */
0fc1b451 454 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
455 __le32 i_flags; /* File flags */
456 union {
457 struct {
25ec56b5 458 __le32 l_i_version;
ac27a0ec
DK
459 } linux1;
460 struct {
461 __u32 h_i_translator;
462 } hurd1;
463 struct {
464 __u32 m_i_reserved1;
465 } masix1;
466 } osd1; /* OS dependent 1 */
617ba13b 467 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 468 __le32 i_generation; /* File version (for NFS) */
7973c0c1 469 __le32 i_file_acl_lo; /* File ACL */
a48380f7 470 __le32 i_size_high;
f077d0d7 471 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
472 union {
473 struct {
0fc1b451 474 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 475 __le16 l_i_file_acl_high;
8c55e204 476 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec
DK
477 __le16 l_i_gid_high; /* were reserved2[0] */
478 __u32 l_i_reserved2;
479 } linux2;
480 struct {
f077d0d7 481 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
482 __u16 h_i_mode_high;
483 __u16 h_i_uid_high;
484 __u16 h_i_gid_high;
485 __u32 h_i_author;
486 } hurd2;
487 struct {
f077d0d7 488 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 489 __le16 m_i_file_acl_high;
ac27a0ec
DK
490 __u32 m_i_reserved2[2];
491 } masix2;
492 } osd2; /* OS dependent 2 */
493 __le16 i_extra_isize;
494 __le16 i_pad1;
ef7f3835
KS
495 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
496 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
497 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
498 __le32 i_crtime; /* File Creation time */
499 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 500 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
501};
502
748de673
AF
503struct move_extent {
504 __u32 reserved; /* should be zero */
505 __u32 donor_fd; /* donor file descriptor */
506 __u64 orig_start; /* logical start offset in block for orig */
507 __u64 donor_start; /* logical start offset in block for donor */
508 __u64 len; /* block length to be moved */
509 __u64 moved_len; /* moved block length */
510};
ac27a0ec 511
ef7f3835
KS
512#define EXT4_EPOCH_BITS 2
513#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
514#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
515
516/*
517 * Extended fields will fit into an inode if the filesystem was formatted
518 * with large inodes (-I 256 or larger) and there are not currently any EAs
519 * consuming all of the available space. For new inodes we always reserve
520 * enough space for the kernel's known extended fields, but for inodes
521 * created with an old kernel this might not have been the case. None of
522 * the extended inode fields is critical for correct filesystem operation.
523 * This macro checks if a certain field fits in the inode. Note that
524 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
525 */
526#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
527 ((offsetof(typeof(*ext4_inode), field) + \
528 sizeof((ext4_inode)->field)) \
529 <= (EXT4_GOOD_OLD_INODE_SIZE + \
530 (einode)->i_extra_isize)) \
531
532static inline __le32 ext4_encode_extra_time(struct timespec *time)
533{
534 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
535 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
536 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
537}
538
539static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
540{
541 if (sizeof(time->tv_sec) > 4)
542 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
543 << 32;
c1fccc06 544 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
545}
546
547#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
548do { \
549 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
550 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
551 (raw_inode)->xtime ## _extra = \
552 ext4_encode_extra_time(&(inode)->xtime); \
553} while (0)
554
555#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
556do { \
557 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
558 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
559 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
560 (raw_inode)->xtime ## _extra = \
561 ext4_encode_extra_time(&(einode)->xtime); \
562} while (0)
563
564#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
565do { \
566 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
567 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
568 ext4_decode_extra_time(&(inode)->xtime, \
569 raw_inode->xtime ## _extra); \
570} while (0)
571
572#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
573do { \
574 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
575 (einode)->xtime.tv_sec = \
576 (signed)le32_to_cpu((raw_inode)->xtime); \
577 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
578 ext4_decode_extra_time(&(einode)->xtime, \
579 raw_inode->xtime ## _extra); \
580} while (0)
581
25ec56b5
JNC
582#define i_disk_version osd1.linux1.l_i_version
583
ac27a0ec
DK
584#if defined(__KERNEL__) || defined(__linux__)
585#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 586#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 587#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
588#define i_uid_low i_uid
589#define i_gid_low i_gid
590#define i_uid_high osd2.linux2.l_i_uid_high
591#define i_gid_high osd2.linux2.l_i_gid_high
592#define i_reserved2 osd2.linux2.l_i_reserved2
593
594#elif defined(__GNU__)
595
596#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
597#define i_uid_high osd2.hurd2.h_i_uid_high
598#define i_gid_high osd2.hurd2.h_i_gid_high
599#define i_author osd2.hurd2.h_i_author
600
601#elif defined(__masix__)
602
603#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 604#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
605#define i_reserved2 osd2.masix2.m_i_reserved2
606
607#endif /* defined(__KERNEL__) || defined(__linux__) */
608
d444c3c3
TT
609/*
610 * storage for cached extent
611 */
612struct ext4_ext_cache {
613 ext4_fsblk_t ec_start;
614 ext4_lblk_t ec_block;
615 __u32 ec_len; /* must be 32bit to return holes */
616 __u32 ec_type;
617};
618
619/*
620 * fourth extended file system inode data in memory
621 */
622struct ext4_inode_info {
623 __le32 i_data[15]; /* unconverted */
624 __u32 i_flags;
625 ext4_fsblk_t i_file_acl;
626 __u32 i_dtime;
627
628 /*
629 * i_block_group is the number of the block group which contains
630 * this file's inode. Constant across the lifetime of the inode,
631 * it is ued for making block allocation decisions - we try to
632 * place a file's data blocks near its inode block, and new inodes
633 * near to their parent directory's inode.
634 */
635 ext4_group_t i_block_group;
636 __u32 i_state; /* Dynamic state flags for ext4 */
637
638 ext4_lblk_t i_dir_start_lookup;
639#ifdef CONFIG_EXT4_FS_XATTR
640 /*
641 * Extended attributes can be read independently of the main file
642 * data. Taking i_mutex even when reading would cause contention
643 * between readers of EAs and writers of regular file data, so
644 * instead we synchronize on xattr_sem when reading or changing
645 * EAs.
646 */
647 struct rw_semaphore xattr_sem;
648#endif
d444c3c3
TT
649
650 struct list_head i_orphan; /* unlinked but open inodes */
651
652 /*
653 * i_disksize keeps track of what the inode size is ON DISK, not
654 * in memory. During truncate, i_size is set to the new size by
655 * the VFS prior to calling ext4_truncate(), but the filesystem won't
656 * set i_disksize to 0 until the truncate is actually under way.
657 *
658 * The intent is that i_disksize always represents the blocks which
659 * are used by this file. This allows recovery to restart truncate
660 * on orphans if we crash during truncate. We actually write i_disksize
661 * into the on-disk inode when writing inodes out, instead of i_size.
662 *
663 * The only time when i_disksize and i_size may be different is when
664 * a truncate is in progress. The only things which change i_disksize
665 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
666 */
667 loff_t i_disksize;
668
669 /*
670 * i_data_sem is for serialising ext4_truncate() against
671 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
672 * data tree are chopped off during truncate. We can't do that in
673 * ext4 because whenever we perform intermediate commits during
674 * truncate, the inode and all the metadata blocks *must* be in a
675 * consistent state which allows truncation of the orphans to restart
676 * during recovery. Hence we must fix the get_block-vs-truncate race
677 * by other means, so we have i_data_sem.
678 */
679 struct rw_semaphore i_data_sem;
680 struct inode vfs_inode;
681 struct jbd2_inode jinode;
682
683 struct ext4_ext_cache i_cached_extent;
684 /*
685 * File creation time. Its function is same as that of
686 * struct timespec i_{a,c,m}time in the generic inode.
687 */
688 struct timespec i_crtime;
689
690 /* mballoc */
691 struct list_head i_prealloc_list;
692 spinlock_t i_prealloc_lock;
693
694 /* ialloc */
695 ext4_group_t i_last_alloc_group;
696
697 /* allocation reservation info for delalloc */
698 unsigned int i_reserved_data_blocks;
699 unsigned int i_reserved_meta_blocks;
700 unsigned int i_allocated_meta_blocks;
701 unsigned short i_delalloc_reserved_flag;
9d0be502
TT
702 sector_t i_da_metadata_calc_last_lblock;
703 int i_da_metadata_calc_len;
d444c3c3
TT
704
705 /* on-disk additional length */
706 __u16 i_extra_isize;
707
708 spinlock_t i_block_reservation_lock;
a9e7f447
DM
709#ifdef CONFIG_QUOTA
710 /* quota space reservation, managed internally by quota code */
711 qsize_t i_reserved_quota;
712#endif
8d5d02e6
MC
713
714 /* completed async DIOs that might need unwritten extents handling */
715 struct list_head i_aio_dio_complete_list;
716 /* current io_end structure for async DIO write*/
717 ext4_io_end_t *cur_aio_dio;
b436b9be
JK
718
719 /*
720 * Transactions that contain inode's metadata needed to complete
721 * fsync and fdatasync, respectively.
722 */
723 tid_t i_sync_tid;
724 tid_t i_datasync_tid;
d444c3c3
TT
725};
726
ac27a0ec
DK
727/*
728 * File system states
729 */
617ba13b
MC
730#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
731#define EXT4_ERROR_FS 0x0002 /* Errors detected */
732#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 733
469108ff
TT
734/*
735 * Misc. filesystem flags
736 */
737#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
738#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
739#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
740
ac27a0ec
DK
741/*
742 * Mount flags
743 */
617ba13b
MC
744#define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
745#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
746#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
747#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
748#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
749#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
750#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
751#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
752#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
753#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
754#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
755#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
756#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
757#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
758#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
759#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 760#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b
MC
761#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
762#define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
763#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
764#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
765#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
d4da6c9c 766#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 767#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
25ec56b5 768#define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
64769240 769#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 770#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 771#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 772#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
5bf5683a 773
617ba13b
MC
774#define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
775#define set_opt(o, opt) o |= EXT4_MOUNT_##opt
776#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
777 EXT4_MOUNT_##opt)
ac27a0ec 778
617ba13b
MC
779#define ext4_set_bit ext2_set_bit
780#define ext4_set_bit_atomic ext2_set_bit_atomic
781#define ext4_clear_bit ext2_clear_bit
782#define ext4_clear_bit_atomic ext2_clear_bit_atomic
783#define ext4_test_bit ext2_test_bit
784#define ext4_find_first_zero_bit ext2_find_first_zero_bit
785#define ext4_find_next_zero_bit ext2_find_next_zero_bit
aa02ad67 786#define ext4_find_next_bit ext2_find_next_bit
ac27a0ec
DK
787
788/*
789 * Maximal mount counts between two filesystem checks
790 */
617ba13b
MC
791#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
792#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
793
794/*
795 * Behaviour when detecting errors
796 */
617ba13b
MC
797#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
798#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
799#define EXT4_ERRORS_PANIC 3 /* Panic */
800#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec
DK
801
802/*
803 * Structure of the super block
804 */
617ba13b 805struct ext4_super_block {
ac27a0ec 806/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 807 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
808 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
809 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
810/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
811 __le32 s_first_data_block; /* First Data Block */
812 __le32 s_log_block_size; /* Block size */
f077d0d7 813 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
ac27a0ec 814/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
f077d0d7 815 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
ac27a0ec
DK
816 __le32 s_inodes_per_group; /* # Inodes per group */
817 __le32 s_mtime; /* Mount time */
818/*30*/ __le32 s_wtime; /* Write time */
819 __le16 s_mnt_count; /* Mount count */
820 __le16 s_max_mnt_count; /* Maximal mount count */
821 __le16 s_magic; /* Magic signature */
822 __le16 s_state; /* File system state */
823 __le16 s_errors; /* Behaviour when detecting errors */
824 __le16 s_minor_rev_level; /* minor revision level */
825/*40*/ __le32 s_lastcheck; /* time of last check */
826 __le32 s_checkinterval; /* max. time between checks */
827 __le32 s_creator_os; /* OS */
828 __le32 s_rev_level; /* Revision level */
829/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
830 __le16 s_def_resgid; /* Default gid for reserved blocks */
831 /*
617ba13b 832 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
833 *
834 * Note: the difference between the compatible feature set and
835 * the incompatible feature set is that if there is a bit set
836 * in the incompatible feature set that the kernel doesn't
837 * know about, it should refuse to mount the filesystem.
838 *
839 * e2fsck's requirements are more strict; if it doesn't know
840 * about a feature in either the compatible or incompatible
841 * feature set, it must abort and not try to meddle with
842 * things it doesn't understand...
843 */
844 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 845 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
846 __le16 s_block_group_nr; /* block group # of this superblock */
847 __le32 s_feature_compat; /* compatible feature set */
848/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
849 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
850/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
851/*78*/ char s_volume_name[16]; /* volume name */
852/*88*/ char s_last_mounted[64]; /* directory where last mounted */
853/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
854 /*
855 * Performance hints. Directory preallocation should only
617ba13b 856 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
857 */
858 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
859 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
860 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
861 /*
617ba13b 862 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
863 */
864/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
865/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
866 __le32 s_journal_dev; /* device number of journal file */
867 __le32 s_last_orphan; /* start of list of inodes to delete */
868 __le32 s_hash_seed[4]; /* HTREE hash seed */
869 __u8 s_def_hash_version; /* Default hash version to use */
870 __u8 s_reserved_char_pad;
0d1ee42f 871 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 872/*100*/ __le32 s_default_mount_opts;
ac27a0ec 873 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
874 __le32 s_mkfs_time; /* When the filesystem was created */
875 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
876 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
877/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
878 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
879 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
880 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
881 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
882 __le32 s_flags; /* Miscellaneous flags */
883 __le16 s_raid_stride; /* RAID stride */
884 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */
885 __le64 s_mmp_block; /* Block for multi-mount protection */
886 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8
JS
887 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
888 __u8 s_reserved_char_pad2;
889 __le16 s_reserved_pad;
afc32f7e
TT
890 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
891 __u32 s_reserved[160]; /* Padding to the end of the block */
ac27a0ec
DK
892};
893
894#ifdef __KERNEL__
bc0b0d6d
TT
895
896/*
4ab2f15b 897 * run-time mount flags
bc0b0d6d
TT
898 */
899#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 900#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 901
ca0faba0
TT
902/*
903 * fourth extended-fs super-block data in memory
904 */
905struct ext4_sb_info {
906 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
907 unsigned long s_inodes_per_block;/* Number of inodes per block */
908 unsigned long s_blocks_per_group;/* Number of blocks in a group */
909 unsigned long s_inodes_per_group;/* Number of inodes in a group */
910 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
911 unsigned long s_gdb_count; /* Number of group descriptor blocks */
912 unsigned long s_desc_per_block; /* Number of group descriptors per block */
913 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 914 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
ca0faba0
TT
915 unsigned long s_overhead_last; /* Last calculated overhead */
916 unsigned long s_blocks_last; /* Last seen block count */
917 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
918 struct buffer_head * s_sbh; /* Buffer containing the super block */
919 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
920 struct buffer_head **s_group_desc;
7f4520cc 921 unsigned int s_mount_opt;
bc0b0d6d 922 unsigned int s_mount_flags;
ca0faba0
TT
923 ext4_fsblk_t s_sb_block;
924 uid_t s_resuid;
925 gid_t s_resgid;
926 unsigned short s_mount_state;
927 unsigned short s_pad;
928 int s_addr_per_block_bits;
929 int s_desc_per_block_bits;
930 int s_inode_size;
931 int s_first_ino;
932 unsigned int s_inode_readahead_blks;
11013911 933 unsigned int s_inode_goal;
ca0faba0
TT
934 spinlock_t s_next_gen_lock;
935 u32 s_next_generation;
936 u32 s_hash_seed[4];
937 int s_def_hash_version;
938 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
939 struct percpu_counter s_freeblocks_counter;
940 struct percpu_counter s_freeinodes_counter;
941 struct percpu_counter s_dirs_counter;
942 struct percpu_counter s_dirtyblocks_counter;
943 struct blockgroup_lock *s_blockgroup_lock;
944 struct proc_dir_entry *s_proc;
945 struct kobject s_kobj;
946 struct completion s_kobj_unregister;
947
948 /* Journaling */
949 struct inode *s_journal_inode;
950 struct journal_s *s_journal;
951 struct list_head s_orphan;
952 struct mutex s_orphan_lock;
953 struct mutex s_resize_lock;
954 unsigned long s_commit_interval;
955 u32 s_max_batch_time;
956 u32 s_min_batch_time;
957 struct block_device *journal_bdev;
958#ifdef CONFIG_JBD2_DEBUG
959 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
960 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
961#endif
962#ifdef CONFIG_QUOTA
963 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
964 int s_jquota_fmt; /* Format of quota to use */
965#endif
966 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 967 struct rb_root system_blks;
ca0faba0
TT
968
969#ifdef EXTENTS_STATS
970 /* ext4 extents stats */
971 unsigned long s_ext_min;
972 unsigned long s_ext_max;
973 unsigned long s_depth_max;
974 spinlock_t s_ext_stats_lock;
975 unsigned long s_ext_blocks;
976 unsigned long s_ext_extents;
977#endif
978
979 /* for buddy allocator */
980 struct ext4_group_info ***s_group_info;
981 struct inode *s_buddy_cache;
982 long s_blocks_reserved;
983 spinlock_t s_reserve_lock;
984 spinlock_t s_md_lock;
985 tid_t s_last_transaction;
986 unsigned short *s_mb_offsets;
987 unsigned int *s_mb_maxs;
988
989 /* tunables */
990 unsigned long s_stripe;
991 unsigned int s_mb_stream_request;
992 unsigned int s_mb_max_to_scan;
993 unsigned int s_mb_min_to_scan;
994 unsigned int s_mb_stats;
995 unsigned int s_mb_order2_reqs;
996 unsigned int s_mb_group_prealloc;
55138e0b 997 unsigned int s_max_writeback_mb_bump;
ca0faba0
TT
998 /* where last allocation was done - for stream allocation */
999 unsigned long s_mb_last_group;
1000 unsigned long s_mb_last_start;
1001
ca0faba0
TT
1002 /* stats for buddy allocator */
1003 spinlock_t s_mb_pa_lock;
1004 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1005 atomic_t s_bal_success; /* we found long enough chunks */
1006 atomic_t s_bal_allocated; /* in blocks */
1007 atomic_t s_bal_ex_scanned; /* total extents scanned */
1008 atomic_t s_bal_goals; /* goal hits */
1009 atomic_t s_bal_breaks; /* too long searches */
1010 atomic_t s_bal_2orders; /* 2^order hits */
1011 spinlock_t s_bal_lock;
1012 unsigned long s_mb_buddies_generated;
1013 unsigned long long s_mb_generation_time;
1014 atomic_t s_mb_lost_chunks;
1015 atomic_t s_mb_preallocated;
1016 atomic_t s_mb_discarded;
50797481 1017 atomic_t s_lock_busy;
ca0faba0
TT
1018
1019 /* locality groups */
1020 struct ext4_locality_group *s_locality_groups;
1021
1022 /* for write statistics */
1023 unsigned long s_sectors_written_start;
1024 u64 s_kbytes_written;
1025
1026 unsigned int s_log_groups_per_flex;
1027 struct flex_groups *s_flex_groups;
4c0425ff
MC
1028
1029 /* workqueue for dio unwritten */
1030 struct workqueue_struct *dio_unwritten_wq;
ca0faba0
TT
1031};
1032
af5bc92d 1033static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1034{
1035 return sb->s_fs_info;
1036}
617ba13b 1037static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1038{
617ba13b 1039 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1040}
1041
ef7f3835
KS
1042static inline struct timespec ext4_current_time(struct inode *inode)
1043{
1044 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1045 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1046}
1047
617ba13b 1048static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1049{
617ba13b
MC
1050 return ino == EXT4_ROOT_INO ||
1051 ino == EXT4_JOURNAL_INO ||
1052 ino == EXT4_RESIZE_INO ||
1053 (ino >= EXT4_FIRST_INO(sb) &&
1054 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec
DK
1055}
1056#else
617ba13b 1057/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1058 * a kernel struct super_block. This will allow us to call the feature-test
1059 * macros from user land. */
617ba13b 1060#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1061#endif
1062
617ba13b 1063#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1064
1065/*
1066 * Codes for operating systems
1067 */
617ba13b
MC
1068#define EXT4_OS_LINUX 0
1069#define EXT4_OS_HURD 1
1070#define EXT4_OS_MASIX 2
1071#define EXT4_OS_FREEBSD 3
1072#define EXT4_OS_LITES 4
ac27a0ec
DK
1073
1074/*
1075 * Revision levels
1076 */
617ba13b
MC
1077#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1078#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1079
617ba13b
MC
1080#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1081#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1082
617ba13b 1083#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1084
1085/*
1086 * Feature set definitions
1087 */
1088
617ba13b 1089#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1090 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1091#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1092 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1093#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1094 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1095#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1096 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1097#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1098 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1099#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1100 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1101#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1102 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1103#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1104 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1105#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1106 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1107
1108#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1109#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1110#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1111#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1112#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1113#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1114
1115#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1116#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1117#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1118#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1119#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1120#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1121#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
617ba13b
MC
1122
1123#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1124#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1125#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1126#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1127#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1128#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1129#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1130#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1131#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
617ba13b
MC
1132
1133#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1134#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1135 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1136 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1137 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581
JS
1138 EXT4_FEATURE_INCOMPAT_64BIT| \
1139 EXT4_FEATURE_INCOMPAT_FLEX_BG)
617ba13b
MC
1140#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1141 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1142 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1143 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1144 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451
AK
1145 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1146 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
ac27a0ec
DK
1147
1148/*
1149 * Default values for user and/or group using reserved blocks
1150 */
617ba13b
MC
1151#define EXT4_DEF_RESUID 0
1152#define EXT4_DEF_RESGID 0
ac27a0ec 1153
240799cd
TT
1154#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1155
ac27a0ec
DK
1156/*
1157 * Default mount options
1158 */
617ba13b
MC
1159#define EXT4_DEFM_DEBUG 0x0001
1160#define EXT4_DEFM_BSDGROUPS 0x0002
1161#define EXT4_DEFM_XATTR_USER 0x0004
1162#define EXT4_DEFM_ACL 0x0008
1163#define EXT4_DEFM_UID16 0x0010
1164#define EXT4_DEFM_JMODE 0x0060
1165#define EXT4_DEFM_JMODE_DATA 0x0020
1166#define EXT4_DEFM_JMODE_ORDERED 0x0040
1167#define EXT4_DEFM_JMODE_WBACK 0x0060
ac27a0ec 1168
30773840
TT
1169/*
1170 * Default journal batch times
1171 */
1172#define EXT4_DEF_MIN_BATCH_TIME 0
1173#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1174
a4912123
TT
1175/*
1176 * Minimum number of groups in a flexgroup before we separate out
1177 * directories into the first block group of a flexgroup
1178 */
1179#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1180
ac27a0ec
DK
1181/*
1182 * Structure of a directory entry
1183 */
617ba13b 1184#define EXT4_NAME_LEN 255
ac27a0ec 1185
617ba13b 1186struct ext4_dir_entry {
ac27a0ec
DK
1187 __le32 inode; /* Inode number */
1188 __le16 rec_len; /* Directory entry length */
1189 __le16 name_len; /* Name length */
617ba13b 1190 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1191};
1192
1193/*
617ba13b 1194 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1195 * stored in intel byte order, and the name_len field could never be
1196 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1197 * file_type field.
1198 */
617ba13b 1199struct ext4_dir_entry_2 {
ac27a0ec
DK
1200 __le32 inode; /* Inode number */
1201 __le16 rec_len; /* Directory entry length */
1202 __u8 name_len; /* Name length */
1203 __u8 file_type;
617ba13b 1204 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1205};
1206
1207/*
617ba13b 1208 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1209 * other bits are reserved for now.
1210 */
617ba13b
MC
1211#define EXT4_FT_UNKNOWN 0
1212#define EXT4_FT_REG_FILE 1
1213#define EXT4_FT_DIR 2
1214#define EXT4_FT_CHRDEV 3
1215#define EXT4_FT_BLKDEV 4
1216#define EXT4_FT_FIFO 5
1217#define EXT4_FT_SOCK 6
1218#define EXT4_FT_SYMLINK 7
ac27a0ec 1219
617ba13b 1220#define EXT4_FT_MAX 8
ac27a0ec
DK
1221
1222/*
617ba13b 1223 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1224 *
1225 * NOTE: It must be a multiple of 4
1226 */
617ba13b
MC
1227#define EXT4_DIR_PAD 4
1228#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1229#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1230 ~EXT4_DIR_ROUND)
a72d7f83
JK
1231#define EXT4_MAX_REC_LEN ((1<<16)-1)
1232
ac27a0ec
DK
1233/*
1234 * Hash Tree Directory indexing
1235 * (c) Daniel Phillips, 2001
1236 */
1237
4074fe37
ES
1238#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1239 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
617ba13b
MC
1240 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1241#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1242#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1243
1244/* Legal values for the dx_root hash_version field: */
1245
1246#define DX_HASH_LEGACY 0
1247#define DX_HASH_HALF_MD4 1
1248#define DX_HASH_TEA 2
f99b2589
TT
1249#define DX_HASH_LEGACY_UNSIGNED 3
1250#define DX_HASH_HALF_MD4_UNSIGNED 4
1251#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec
DK
1252
1253#ifdef __KERNEL__
1254
1255/* hash info structure used by the directory hash */
1256struct dx_hash_info
1257{
1258 u32 hash;
1259 u32 minor_hash;
1260 int hash_version;
1261 u32 *seed;
1262};
1263
617ba13b 1264#define EXT4_HTREE_EOF 0x7fffffff
ac27a0ec
DK
1265
1266/*
617ba13b 1267 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1268 */
1269#define HASH_NB_ALWAYS 1
1270
1271
1272/*
1273 * Describe an inode's exact location on disk and in memory
1274 */
617ba13b 1275struct ext4_iloc
ac27a0ec
DK
1276{
1277 struct buffer_head *bh;
1278 unsigned long offset;
fd2d4291 1279 ext4_group_t block_group;
ac27a0ec
DK
1280};
1281
617ba13b 1282static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1283{
617ba13b 1284 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1285}
1286
1287/*
1288 * This structure is stuffed into the struct file's private_data field
1289 * for directories. It is where we put information so that we can do
1290 * readdir operations in hash tree order.
1291 */
1292struct dir_private_info {
1293 struct rb_root root;
1294 struct rb_node *curr_node;
1295 struct fname *extra_fname;
1296 loff_t last_pos;
1297 __u32 curr_hash;
1298 __u32 curr_minor_hash;
1299 __u32 next_hash;
1300};
1301
1302/* calculate the first block number of the group */
617ba13b 1303static inline ext4_fsblk_t
fd2d4291 1304ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1305{
617ba13b
MC
1306 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1307 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1308}
1309
1310/*
1311 * Special error return code only used by dx_probe() and its callers.
1312 */
1313#define ERR_BAD_DX_DIR -75000
1314
72b64b59 1315void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
a9df9a49 1316 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
3a5b2ecd 1317
9f6200bb
TT
1318extern struct proc_dir_entry *ext4_proc_root;
1319
ac27a0ec
DK
1320/*
1321 * Function prototypes
1322 */
1323
1324/*
1325 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1326 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1327 */
8c55e204
DK
1328# define NORET_TYPE /**/
1329# define ATTRIB_NORET __attribute__((noreturn))
1330# define NORET_AND noreturn,
ac27a0ec 1331
498e5f24
TT
1332/* bitmap.c */
1333extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1334
ac27a0ec 1335/* balloc.c */
3a5b2ecd
MC
1336extern unsigned int ext4_block_group(struct super_block *sb,
1337 ext4_fsblk_t blocknr);
1338extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1339 ext4_fsblk_t blocknr);
fd2d4291
AM
1340extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1341extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1342 ext4_group_t group);
7061eba7 1343extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
617ba13b 1344 ext4_fsblk_t goal, unsigned long *count, int *errp);
5c791616 1345extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
8c3bf8a0 1346extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
e21675d4
AK
1347extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1348 ext4_fsblk_t block, unsigned long count);
af5bc92d
TT
1349extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1350extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1351extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1352 ext4_group_t block_group,
ac27a0ec 1353 struct buffer_head ** bh);
617ba13b 1354extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
bb23c20a
TT
1355struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1356 ext4_group_t block_group);
1357extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1358 struct buffer_head *bh,
1359 ext4_group_t group,
1360 struct ext4_group_desc *desc);
1361#define ext4_free_blocks_after_init(sb, group, desc) \
1362 ext4_init_block_bitmap(sb, NULL, group, desc)
ac27a0ec
DK
1363
1364/* dir.c */
617ba13b
MC
1365extern int ext4_check_dir_entry(const char *, struct inode *,
1366 struct ext4_dir_entry_2 *,
498e5f24 1367 struct buffer_head *, unsigned int);
617ba13b 1368extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 1369 __u32 minor_hash,
617ba13b
MC
1370 struct ext4_dir_entry_2 *dirent);
1371extern void ext4_htree_free_dir_info(struct dir_private_info *p);
ac27a0ec
DK
1372
1373/* fsync.c */
af5bc92d 1374extern int ext4_sync_file(struct file *, struct dentry *, int);
ac27a0ec
DK
1375
1376/* hash.c */
617ba13b 1377extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
1378 dx_hash_info *hinfo);
1379
1380/* ialloc.c */
f157a4aa 1381extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
11013911 1382 const struct qstr *qstr, __u32 goal);
af5bc92d
TT
1383extern void ext4_free_inode(handle_t *, struct inode *);
1384extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1385extern unsigned long ext4_count_free_inodes(struct super_block *);
1386extern unsigned long ext4_count_dirs(struct super_block *);
1387extern void ext4_check_inodes_bitmap(struct super_block *);
bb23c20a
TT
1388extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1389 struct buffer_head *bh,
1390 ext4_group_t group,
1391 struct ext4_group_desc *desc);
1392extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
ac27a0ec 1393
c9de560d
AT
1394/* mballoc.c */
1395extern long ext4_mb_stats;
1396extern long ext4_mb_max_to_scan;
1397extern int ext4_mb_init(struct super_block *, int);
1398extern int ext4_mb_release(struct super_block *);
1399extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1400 struct ext4_allocation_request *, int *);
1401extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 1402extern void ext4_discard_preallocations(struct inode *);
c9de560d
AT
1403extern int __init init_ext4_mballoc(void);
1404extern void exit_ext4_mballoc(void);
44338711 1405extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
1406 struct buffer_head *bh, ext4_fsblk_t block,
1407 unsigned long count, int flags);
920313a7 1408extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 1409 ext4_group_t i, struct ext4_group_desc *desc);
920313a7
AK
1410extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1411extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1412 ext4_group_t, int);
ac27a0ec 1413/* inode.c */
725d26d3
AK
1414struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1415 ext4_lblk_t, int, int *);
1416struct buffer_head *ext4_bread(handle_t *, struct inode *,
1417 ext4_lblk_t, int, int *);
6873fa0d
ES
1418int ext4_get_block(struct inode *inode, sector_t iblock,
1419 struct buffer_head *bh_result, int create);
ac27a0ec 1420
1d1fe1ee 1421extern struct inode *ext4_iget(struct super_block *, unsigned long);
af5bc92d
TT
1422extern int ext4_write_inode(struct inode *, int);
1423extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
1424extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1425 struct kstat *stat);
af5bc92d
TT
1426extern void ext4_delete_inode(struct inode *);
1427extern int ext4_sync_inode(handle_t *, struct inode *);
617ba13b
MC
1428extern void ext4_dirty_inode(struct inode *);
1429extern int ext4_change_inode_journal_flag(struct inode *, int);
1430extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
91ef4caf 1431extern int ext4_can_truncate(struct inode *inode);
af5bc92d 1432extern void ext4_truncate(struct inode *);
487caeef 1433extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 1434extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 1435extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 1436extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 1437extern void ext4_set_aops(struct inode *inode);
a86c6181 1438extern int ext4_writepage_trans_blocks(struct inode *);
a02908f1 1439extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
f3bd1f3f 1440extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
cf108bca 1441extern int ext4_block_truncate_page(handle_t *handle,
a86c6181 1442 struct address_space *mapping, loff_t from);
c2ec175c 1443extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
a9e7f447 1444extern qsize_t *ext4_get_reserved_space(struct inode *inode);
8d5d02e6 1445extern int flush_aio_dio_completed_IO(struct inode *inode);
5f634d06
AK
1446extern void ext4_da_update_reserve_space(struct inode *inode,
1447 int used, int quota_claim);
ac27a0ec 1448/* ioctl.c */
5cdd7b2d 1449extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 1450extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 1451
c14c6fd5 1452/* migrate.c */
2a43a878 1453extern int ext4_ext_migrate(struct inode *);
3d0518f4 1454
ac27a0ec 1455/* namei.c */
3d0518f4
WY
1456extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1457extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
617ba13b
MC
1458extern int ext4_orphan_add(handle_t *, struct inode *);
1459extern int ext4_orphan_del(handle_t *, struct inode *);
1460extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec
DK
1461 __u32 start_minor_hash, __u32 *next_hash);
1462
1463/* resize.c */
617ba13b
MC
1464extern int ext4_group_add(struct super_block *sb,
1465 struct ext4_new_group_data *input);
1466extern int ext4_group_extend(struct super_block *sb,
1467 struct ext4_super_block *es,
1468 ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
1469
1470/* super.c */
af5bc92d 1471extern void ext4_error(struct super_block *, const char *, const char *, ...)
ac27a0ec 1472 __attribute__ ((format (printf, 3, 4)));
af5bc92d
TT
1473extern void __ext4_std_error(struct super_block *, const char *, int);
1474extern void ext4_abort(struct super_block *, const char *, const char *, ...)
ac27a0ec 1475 __attribute__ ((format (printf, 3, 4)));
af5bc92d 1476extern void ext4_warning(struct super_block *, const char *, const char *, ...)
ac27a0ec 1477 __attribute__ ((format (printf, 3, 4)));
b31e1552
ES
1478extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1479 __attribute__ ((format (printf, 3, 4)));
5d1b1b3f
AK
1480extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1481 const char *, const char *, ...)
1482 __attribute__ ((format (printf, 4, 5)));
af5bc92d 1483extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
1484extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1485 __u32 compat);
1486extern int ext4_update_rocompat_feature(handle_t *handle,
1487 struct super_block *sb, __u32 rocompat);
1488extern int ext4_update_incompat_feature(handle_t *handle,
1489 struct super_block *sb, __u32 incompat);
8fadc143
AR
1490extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1491 struct ext4_group_desc *bg);
1492extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1493 struct ext4_group_desc *bg);
1494extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1495 struct ext4_group_desc *bg);
560671a0
AK
1496extern __u32 ext4_free_blks_count(struct super_block *sb,
1497 struct ext4_group_desc *bg);
1498extern __u32 ext4_free_inodes_count(struct super_block *sb,
1499 struct ext4_group_desc *bg);
1500extern __u32 ext4_used_dirs_count(struct super_block *sb,
1501 struct ext4_group_desc *bg);
1502extern __u32 ext4_itable_unused_count(struct super_block *sb,
1503 struct ext4_group_desc *bg);
8fadc143
AR
1504extern void ext4_block_bitmap_set(struct super_block *sb,
1505 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1506extern void ext4_inode_bitmap_set(struct super_block *sb,
1507 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1508extern void ext4_inode_table_set(struct super_block *sb,
1509 struct ext4_group_desc *bg, ext4_fsblk_t blk);
560671a0
AK
1510extern void ext4_free_blks_set(struct super_block *sb,
1511 struct ext4_group_desc *bg, __u32 count);
1512extern void ext4_free_inodes_set(struct super_block *sb,
1513 struct ext4_group_desc *bg, __u32 count);
1514extern void ext4_used_dirs_set(struct super_block *sb,
1515 struct ext4_group_desc *bg, __u32 count);
1516extern void ext4_itable_unused_set(struct super_block *sb,
1517 struct ext4_group_desc *bg, __u32 count);
bb23c20a
TT
1518extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1519 struct ext4_group_desc *gdp);
1520extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1521 struct ext4_group_desc *gdp);
bd81d8ee
LV
1522
1523static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1524{
1525 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 1526 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
1527}
1528
1529static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1530{
1531 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 1532 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
1533}
1534
1535static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1536{
1537 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 1538 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
1539}
1540
1541static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1542 ext4_fsblk_t blk)
1543{
6bc9feff 1544 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1545 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1546}
1547
1548static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1549 ext4_fsblk_t blk)
1550{
308ba3ec 1551 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1552 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1553}
1554
1555static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1556 ext4_fsblk_t blk)
1557{
308ba3ec 1558 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1559 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1560}
1561
a48380f7
AK
1562static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1563{
06a279d6
TT
1564 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1565 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1566 le32_to_cpu(raw_inode->i_size_lo);
1567 else
1568 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 1569}
bd81d8ee 1570
a48380f7
AK
1571static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1572{
1573 raw_inode->i_size_lo = cpu_to_le32(i_size);
1574 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1575}
ac27a0ec 1576
c9de560d
AT
1577static inline
1578struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 1579 ext4_group_t group)
c9de560d
AT
1580{
1581 struct ext4_group_info ***grp_info;
1582 long indexv, indexh;
1583 grp_info = EXT4_SB(sb)->s_group_info;
1584 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1585 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1586 return grp_info[indexv][indexh];
1587}
1588
8df9675f
TT
1589/*
1590 * Reading s_groups_count requires using smp_rmb() afterwards. See
1591 * the locking protocol documented in the comments of ext4_group_add()
1592 * in resize.c
1593 */
1594static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1595{
1596 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
1597
1598 smp_rmb();
1599 return ngroups;
1600}
c9de560d 1601
772cb7c8
JS
1602static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1603 ext4_group_t block_group)
1604{
1605 return block_group >> sbi->s_log_groups_per_flex;
1606}
1607
1608static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1609{
1610 return 1 << sbi->s_log_groups_per_flex;
1611}
1612
617ba13b 1613#define ext4_std_error(sb, errno) \
ac27a0ec
DK
1614do { \
1615 if ((errno)) \
4db9c54a 1616 __ext4_std_error((sb), __func__, (errno)); \
ac27a0ec
DK
1617} while (0)
1618
a30d542a 1619#ifdef CONFIG_SMP
179f7ebf 1620/* Each CPU can accumulate percpu_counter_batch blocks in their local
a30d542a 1621 * counters. So we need to make sure we have free blocks more
179f7ebf 1622 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 1623 */
179f7ebf 1624#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a
AK
1625#else
1626#define EXT4_FREEBLOCKS_WATERMARK 0
1627#endif
1628
cf17fea6
AK
1629static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1630{
1631 /*
1632 * XXX: replace with spinlock if seen contended -bzzz
1633 */
1634 down_write(&EXT4_I(inode)->i_data_sem);
1635 if (newsize > EXT4_I(inode)->i_disksize)
1636 EXT4_I(inode)->i_disksize = newsize;
1637 up_write(&EXT4_I(inode)->i_data_sem);
1638 return ;
1639}
a30d542a 1640
5d1b1b3f
AK
1641struct ext4_group_info {
1642 unsigned long bb_state;
1643 struct rb_root bb_free_root;
a36b4498
ES
1644 ext4_grpblk_t bb_first_free; /* first free block */
1645 ext4_grpblk_t bb_free; /* total free blocks */
1646 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
5d1b1b3f
AK
1647 struct list_head bb_prealloc_list;
1648#ifdef DOUBLE_CHECK
1649 void *bb_bitmap;
1650#endif
1651 struct rw_semaphore alloc_sem;
a36b4498
ES
1652 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
1653 * regions, index is order.
1654 * bb_counters[3] = 5 means
1655 * 5 free 8-block regions. */
5d1b1b3f
AK
1656};
1657
1658#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
5d1b1b3f
AK
1659
1660#define EXT4_MB_GRP_NEED_INIT(grp) \
1661 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1662
50797481
TT
1663#define EXT4_MAX_CONTENTION 8
1664#define EXT4_CONTENTION_THRESHOLD 2
1665
955ce5f5
AK
1666static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1667 ext4_group_t group)
5d1b1b3f 1668{
955ce5f5
AK
1669 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1670}
5d1b1b3f 1671
50797481
TT
1672/*
1673 * Returns true if the filesystem is busy enough that attempts to
1674 * access the block group locks has run into contention.
1675 */
1676static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1677{
1678 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1679}
1680
955ce5f5
AK
1681static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1682{
50797481
TT
1683 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1684 if (spin_trylock(lock))
1685 /*
1686 * We're able to grab the lock right away, so drop the
1687 * lock contention counter.
1688 */
1689 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1690 else {
1691 /*
1692 * The lock is busy, so bump the contention counter,
1693 * and then wait on the spin lock.
1694 */
1695 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1696 EXT4_MAX_CONTENTION);
1697 spin_lock(lock);
1698 }
5d1b1b3f
AK
1699}
1700
1701static inline void ext4_unlock_group(struct super_block *sb,
1702 ext4_group_t group)
1703{
955ce5f5 1704 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
1705}
1706
ac27a0ec
DK
1707/*
1708 * Inodes and files operations
1709 */
1710
1711/* dir.c */
617ba13b 1712extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
1713
1714/* file.c */
c5ef1c42 1715extern const struct inode_operations ext4_file_inode_operations;
617ba13b 1716extern const struct file_operations ext4_file_operations;
ac27a0ec
DK
1717
1718/* namei.c */
c5ef1c42
AV
1719extern const struct inode_operations ext4_dir_inode_operations;
1720extern const struct inode_operations ext4_special_inode_operations;
596397b7 1721extern struct dentry *ext4_get_parent(struct dentry *child);
ac27a0ec
DK
1722
1723/* symlink.c */
c5ef1c42
AV
1724extern const struct inode_operations ext4_symlink_inode_operations;
1725extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 1726
6fd058f7
TT
1727/* block_validity */
1728extern void ext4_release_system_zone(struct super_block *sb);
1729extern int ext4_setup_system_zone(struct super_block *sb);
1730extern int __init init_ext4_system_zone(void);
1731extern void exit_ext4_system_zone(void);
1732extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1733 ext4_fsblk_t start_blk,
1734 unsigned int count);
1735
a86c6181
AT
1736/* extents.c */
1737extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1738extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
a02908f1
MC
1739extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1740 int chunk);
a86c6181 1741extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
498e5f24 1742 ext4_lblk_t iblock, unsigned int max_blocks,
c2177057 1743 struct buffer_head *bh_result, int flags);
cf108bca 1744extern void ext4_ext_truncate(struct inode *);
a86c6181
AT
1745extern void ext4_ext_init(struct super_block *);
1746extern void ext4_ext_release(struct super_block *);
a2df2a63
AA
1747extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1748 loff_t len);
0031462b
MC
1749extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1750 loff_t len);
12b7ac17
TT
1751extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1752 sector_t block, unsigned int max_blocks,
c2177057 1753 struct buffer_head *bh, int flags);
3a06d778
AK
1754extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1755 __u64 start, __u64 len);
748de673
AF
1756/* move_extent.c */
1757extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1758 __u64 start_orig, __u64 start_donor,
1759 __u64 len, __u64 *moved_len);
1760
3a06d778 1761
2ccb5fb9
AK
1762/*
1763 * Add new method to test wether block and inode bitmaps are properly
1764 * initialized. With uninit_bg reading the block from disk is not enough
1765 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1766 */
1767#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1768
1769static inline int bitmap_uptodate(struct buffer_head *bh)
1770{
1771 return (buffer_uptodate(bh) &&
1772 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1773}
1774static inline void set_bitmap_uptodate(struct buffer_head *bh)
1775{
1776 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1777}
1778
ac27a0ec
DK
1779#endif /* __KERNEL__ */
1780
3dcf5451 1781#endif /* _EXT4_H */