ext4: Use end_io callback to avoid direct I/O fallback to buffered I/O
[linux-2.6-block.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
68
c9de560d 69/* prefer goal again. length */
0ef90db9 70#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 71/* blocks already reserved */
0ef90db9 72#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 73/* metadata is being allocated */
0ef90db9 74#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 75/* first blocks in the file */
0ef90db9 76#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 77/* search for the best chunk */
0ef90db9 78#define EXT4_MB_HINT_BEST 0x0010
c9de560d 79/* data is being allocated */
0ef90db9 80#define EXT4_MB_HINT_DATA 0x0020
c9de560d 81/* don't preallocate (for tails) */
0ef90db9 82#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 83/* allocate for locality group */
0ef90db9 84#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 85/* allocate goal blocks or none */
0ef90db9 86#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 87/* goal is meaningful */
0ef90db9 88#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 89/* blocks already pre-reserved by delayed allocation */
0ef90db9 90#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
91/* We are doing stream allocation */
92#define EXT4_MB_STREAM_ALLOC 0x0800
d2a17637 93
c9de560d
AT
94
95struct ext4_allocation_request {
96 /* target inode for block we're allocating */
97 struct inode *inode;
726447d8
ES
98 /* how many blocks we want to allocate */
99 unsigned int len;
c9de560d
AT
100 /* logical block in target inode */
101 ext4_lblk_t logical;
c9de560d
AT
102 /* the closest logical allocated block to the left */
103 ext4_lblk_t lleft;
c9de560d
AT
104 /* the closest logical allocated block to the right */
105 ext4_lblk_t lright;
726447d8
ES
106 /* phys. target (a hint) */
107 ext4_fsblk_t goal;
108 /* phys. block for the closest logical allocated block to the left */
109 ext4_fsblk_t pleft;
110 /* phys. block for the closest logical allocated block to the right */
c9de560d 111 ext4_fsblk_t pright;
c9de560d 112 /* flags. see above EXT4_MB_HINT_* */
498e5f24 113 unsigned int flags;
c9de560d
AT
114};
115
b3a3ca8c
TT
116/*
117 * For delayed allocation tracking
118 */
119struct mpage_da_data {
120 struct inode *inode;
121 sector_t b_blocknr; /* start block number of extent */
122 size_t b_size; /* size of extent */
123 unsigned long b_state; /* state of the extent */
124 unsigned long first_page, next_page; /* extent of pages */
125 struct writeback_control *wbc;
126 int io_done;
127 int pages_written;
128 int retval;
129};
130
0031462b
MC
131typedef struct ext4_io_end {
132 struct inode *inode; /* file being written to */
133 unsigned int flag; /* sync IO or AIO */
134 int error; /* I/O error code */
135 ext4_lblk_t offset; /* offset in the file */
136 size_t size; /* size of the extent */
137 struct work_struct work; /* data work queue */
138} ext4_io_end_t;
139
ac27a0ec
DK
140/*
141 * Special inodes numbers
142 */
617ba13b
MC
143#define EXT4_BAD_INO 1 /* Bad blocks inode */
144#define EXT4_ROOT_INO 2 /* Root inode */
145#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
146#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
147#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
148#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 149
617ba13b
MC
150/* First non-reserved inode for old ext4 filesystems */
151#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
152
153/*
154 * Maximal count of links to a file
155 */
f8628a14 156#define EXT4_LINK_MAX 65000
ac27a0ec
DK
157
158/*
159 * Macro-instructions used to manage several block sizes
160 */
617ba13b 161#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
162#define EXT4_MAX_BLOCK_SIZE 65536
163#define EXT4_MIN_BLOCK_LOG_SIZE 10
ac27a0ec 164#ifdef __KERNEL__
617ba13b 165# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 166#else
617ba13b 167# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 168#endif
af5bc92d 169#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
ac27a0ec 170#ifdef __KERNEL__
617ba13b 171# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
ac27a0ec 172#else
617ba13b 173# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
174#endif
175#ifdef __KERNEL__
617ba13b
MC
176#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
177#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
178#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 179#else
617ba13b
MC
180#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
181 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 182 (s)->s_inode_size)
617ba13b
MC
183#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
184 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
185 (s)->s_first_ino)
186#endif
a2df2a63 187#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 188
ac27a0ec
DK
189/*
190 * Structure of a blocks group descriptor
191 */
617ba13b 192struct ext4_group_desc
ac27a0ec 193{
3a14589c 194 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
195 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
196 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
197 __le16 bg_free_blocks_count_lo;/* Free blocks count */
198 __le16 bg_free_inodes_count_lo;/* Free inodes count */
199 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4
AD
200 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
201 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
560671a0 202 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 203 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
204 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
205 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
206 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
207 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
208 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
209 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 210 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
91b51a01 211 __u32 bg_reserved2[3];
ac27a0ec
DK
212};
213
772cb7c8
JS
214/*
215 * Structure of a flex block group info
216 */
217
218struct flex_groups {
9f24e420
TT
219 atomic_t free_inodes;
220 atomic_t free_blocks;
7d39db14 221 atomic_t used_dirs;
772cb7c8
JS
222};
223
717d50e4
AD
224#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
225#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
226#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
227
ac27a0ec
DK
228/*
229 * Macro-instructions used to manage group descriptors
230 */
0d1ee42f 231#define EXT4_MIN_DESC_SIZE 32
8fadc143 232#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
233#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
234#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 235#ifdef __KERNEL__
617ba13b
MC
236# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
237# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
238# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
239# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 240#else
617ba13b 241# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 242# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 243# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
244#endif
245
246/*
247 * Constants relative to the data blocks
248 */
617ba13b
MC
249#define EXT4_NDIR_BLOCKS 12
250#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
251#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
252#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
253#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
254
255/*
256 * Inode flags
257 */
617ba13b
MC
258#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
259#define EXT4_UNRM_FL 0x00000002 /* Undelete */
260#define EXT4_COMPR_FL 0x00000004 /* Compress file */
261#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
262#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
263#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
264#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
265#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 266/* Reserved for compression usage... */
617ba13b
MC
267#define EXT4_DIRTY_FL 0x00000100
268#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
269#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
270#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
ac27a0ec 271/* End compression flags --- maybe not all used */
617ba13b
MC
272#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
273#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
274#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
275#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
276#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
277#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 278#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 279#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
8180a562 280#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 281
a86c6181 282#define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
4db46fc2 283#define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
ac27a0ec 284
8fa43a81
DG
285/* Flags that should be inherited by new inodes from their parent. */
286#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
287 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
288 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
289 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
290 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
291
2dc6b0d4
DG
292/* Flags that are appropriate for regular files (all but dir-specific ones). */
293#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
294
295/* Flags that are appropriate for non-directories/regular files. */
296#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
297
298/* Mask out flags that are inappropriate for the given type of inode. */
299static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
300{
301 if (S_ISDIR(mode))
302 return flags;
303 else if (S_ISREG(mode))
304 return flags & EXT4_REG_FLMASK;
305 else
306 return flags & EXT4_OTHER_FLMASK;
307}
308
ac27a0ec
DK
309/*
310 * Inode dynamic state flags
311 */
617ba13b
MC
312#define EXT4_STATE_JDATA 0x00000001 /* journaled data exists */
313#define EXT4_STATE_NEW 0x00000002 /* inode is newly created */
314#define EXT4_STATE_XATTR 0x00000004 /* has in-inode xattrs */
6dd4ee7c 315#define EXT4_STATE_NO_EXPAND 0x00000008 /* No space for expansion */
7d8f9f7d 316#define EXT4_STATE_DA_ALLOC_CLOSE 0x00000010 /* Alloc DA blks on close */
1b9c12f4 317#define EXT4_STATE_EXT_MIGRATE 0x00000020 /* Inode is migrating */
ac27a0ec
DK
318
319/* Used to pass group descriptor data when online resize is done */
617ba13b 320struct ext4_new_group_input {
8c55e204
DK
321 __u32 group; /* Group number for this data */
322 __u64 block_bitmap; /* Absolute block number of block bitmap */
323 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
324 __u64 inode_table; /* Absolute block number of inode table start */
325 __u32 blocks_count; /* Total number of blocks in this group */
326 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
327 __u16 unused;
328};
329
617ba13b
MC
330/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
331struct ext4_new_group_data {
ac27a0ec 332 __u32 group;
bd81d8ee
LV
333 __u64 block_bitmap;
334 __u64 inode_bitmap;
335 __u64 inode_table;
ac27a0ec
DK
336 __u32 blocks_count;
337 __u16 reserved_blocks;
338 __u16 unused;
339 __u32 free_blocks_count;
340};
341
a2df2a63 342/*
c2177057 343 * Flags used by ext4_get_blocks()
a2df2a63 344 */
c2177057
TT
345 /* Allocate any needed blocks and/or convert an unitialized
346 extent to be an initialized ext4 */
2ac3b6e0 347#define EXT4_GET_BLOCKS_CREATE 0x0001
c2177057 348 /* Request the creation of an unitialized extent */
2ac3b6e0 349#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
c2177057
TT
350#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
351 EXT4_GET_BLOCKS_CREATE)
c2177057 352 /* Caller is from the delayed allocation writeout path,
2ac3b6e0
TT
353 so set the magic i_delalloc_reserve_flag after taking the
354 inode allocation semaphore for */
03f5d8bc 355#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
2ac3b6e0
TT
356 /* Call ext4_da_update_reserve_space() after successfully
357 allocating the blocks */
03f5d8bc 358#define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008
0031462b
MC
359 /* caller is from the direct IO path, request to creation of an
360 unitialized extents if not allocated, split the uninitialized
361 extent if blocks has been preallocated already*/
362#define EXT4_GET_BLOCKS_DIO 0x0010
363#define EXT4_GET_BLOCKS_CONVERT 0x0020
364#define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
365 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
366 /* Convert extent to initialized after direct IO complete */
367#define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
368 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
ac27a0ec
DK
369
370/*
371 * ioctl commands
372 */
617ba13b
MC
373#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
374#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
375#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
376#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
377#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
378#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
e23291b9 379#ifdef CONFIG_JBD2_DEBUG
617ba13b 380#define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
ac27a0ec 381#endif
617ba13b
MC
382#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
383#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
384#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
385#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
386#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 387 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 388 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 389#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 390#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
ac27a0ec
DK
391
392/*
393 * ioctl commands in 32 bit emulation
394 */
617ba13b
MC
395#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
396#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
397#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
398#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
399#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
400#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
401#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
e23291b9 402#ifdef CONFIG_JBD2_DEBUG
617ba13b 403#define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
ac27a0ec 404#endif
617ba13b
MC
405#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
406#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
ac27a0ec
DK
407
408
409/*
410 * Mount options
411 */
617ba13b 412struct ext4_mount_options {
ac27a0ec
DK
413 unsigned long s_mount_opt;
414 uid_t s_resuid;
415 gid_t s_resgid;
416 unsigned long s_commit_interval;
30773840 417 u32 s_min_batch_time, s_max_batch_time;
ac27a0ec
DK
418#ifdef CONFIG_QUOTA
419 int s_jquota_fmt;
420 char *s_qf_names[MAXQUOTAS];
421#endif
422};
423
fb0a387d
ES
424/* Max physical block we can addres w/o extents */
425#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
426
ac27a0ec
DK
427/*
428 * Structure of an inode on the disk
429 */
617ba13b 430struct ext4_inode {
ac27a0ec
DK
431 __le16 i_mode; /* File mode */
432 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 433 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 434 __le32 i_atime; /* Access time */
ef7f3835 435 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
436 __le32 i_mtime; /* Modification time */
437 __le32 i_dtime; /* Deletion Time */
438 __le16 i_gid; /* Low 16 bits of Group Id */
439 __le16 i_links_count; /* Links count */
0fc1b451 440 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
441 __le32 i_flags; /* File flags */
442 union {
443 struct {
25ec56b5 444 __le32 l_i_version;
ac27a0ec
DK
445 } linux1;
446 struct {
447 __u32 h_i_translator;
448 } hurd1;
449 struct {
450 __u32 m_i_reserved1;
451 } masix1;
452 } osd1; /* OS dependent 1 */
617ba13b 453 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 454 __le32 i_generation; /* File version (for NFS) */
7973c0c1 455 __le32 i_file_acl_lo; /* File ACL */
a48380f7 456 __le32 i_size_high;
f077d0d7 457 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
458 union {
459 struct {
0fc1b451 460 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 461 __le16 l_i_file_acl_high;
8c55e204 462 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec
DK
463 __le16 l_i_gid_high; /* were reserved2[0] */
464 __u32 l_i_reserved2;
465 } linux2;
466 struct {
f077d0d7 467 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
468 __u16 h_i_mode_high;
469 __u16 h_i_uid_high;
470 __u16 h_i_gid_high;
471 __u32 h_i_author;
472 } hurd2;
473 struct {
f077d0d7 474 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 475 __le16 m_i_file_acl_high;
ac27a0ec
DK
476 __u32 m_i_reserved2[2];
477 } masix2;
478 } osd2; /* OS dependent 2 */
479 __le16 i_extra_isize;
480 __le16 i_pad1;
ef7f3835
KS
481 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
482 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
483 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
484 __le32 i_crtime; /* File Creation time */
485 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 486 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
487};
488
748de673
AF
489struct move_extent {
490 __u32 reserved; /* should be zero */
491 __u32 donor_fd; /* donor file descriptor */
492 __u64 orig_start; /* logical start offset in block for orig */
493 __u64 donor_start; /* logical start offset in block for donor */
494 __u64 len; /* block length to be moved */
495 __u64 moved_len; /* moved block length */
496};
ac27a0ec 497
ef7f3835
KS
498#define EXT4_EPOCH_BITS 2
499#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
500#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
501
502/*
503 * Extended fields will fit into an inode if the filesystem was formatted
504 * with large inodes (-I 256 or larger) and there are not currently any EAs
505 * consuming all of the available space. For new inodes we always reserve
506 * enough space for the kernel's known extended fields, but for inodes
507 * created with an old kernel this might not have been the case. None of
508 * the extended inode fields is critical for correct filesystem operation.
509 * This macro checks if a certain field fits in the inode. Note that
510 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
511 */
512#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
513 ((offsetof(typeof(*ext4_inode), field) + \
514 sizeof((ext4_inode)->field)) \
515 <= (EXT4_GOOD_OLD_INODE_SIZE + \
516 (einode)->i_extra_isize)) \
517
518static inline __le32 ext4_encode_extra_time(struct timespec *time)
519{
520 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
521 time->tv_sec >> 32 : 0) |
522 ((time->tv_nsec << 2) & EXT4_NSEC_MASK));
523}
524
525static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
526{
527 if (sizeof(time->tv_sec) > 4)
528 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
529 << 32;
530 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> 2;
531}
532
533#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
534do { \
535 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
536 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
537 (raw_inode)->xtime ## _extra = \
538 ext4_encode_extra_time(&(inode)->xtime); \
539} while (0)
540
541#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
542do { \
543 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
544 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
545 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
546 (raw_inode)->xtime ## _extra = \
547 ext4_encode_extra_time(&(einode)->xtime); \
548} while (0)
549
550#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
551do { \
552 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
553 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
554 ext4_decode_extra_time(&(inode)->xtime, \
555 raw_inode->xtime ## _extra); \
556} while (0)
557
558#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
559do { \
560 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
561 (einode)->xtime.tv_sec = \
562 (signed)le32_to_cpu((raw_inode)->xtime); \
563 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
564 ext4_decode_extra_time(&(einode)->xtime, \
565 raw_inode->xtime ## _extra); \
566} while (0)
567
25ec56b5
JNC
568#define i_disk_version osd1.linux1.l_i_version
569
ac27a0ec
DK
570#if defined(__KERNEL__) || defined(__linux__)
571#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 572#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 573#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
574#define i_uid_low i_uid
575#define i_gid_low i_gid
576#define i_uid_high osd2.linux2.l_i_uid_high
577#define i_gid_high osd2.linux2.l_i_gid_high
578#define i_reserved2 osd2.linux2.l_i_reserved2
579
580#elif defined(__GNU__)
581
582#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
583#define i_uid_high osd2.hurd2.h_i_uid_high
584#define i_gid_high osd2.hurd2.h_i_gid_high
585#define i_author osd2.hurd2.h_i_author
586
587#elif defined(__masix__)
588
589#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 590#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
591#define i_reserved2 osd2.masix2.m_i_reserved2
592
593#endif /* defined(__KERNEL__) || defined(__linux__) */
594
d444c3c3
TT
595/*
596 * storage for cached extent
597 */
598struct ext4_ext_cache {
599 ext4_fsblk_t ec_start;
600 ext4_lblk_t ec_block;
601 __u32 ec_len; /* must be 32bit to return holes */
602 __u32 ec_type;
603};
604
605/*
606 * fourth extended file system inode data in memory
607 */
608struct ext4_inode_info {
609 __le32 i_data[15]; /* unconverted */
610 __u32 i_flags;
611 ext4_fsblk_t i_file_acl;
612 __u32 i_dtime;
613
614 /*
615 * i_block_group is the number of the block group which contains
616 * this file's inode. Constant across the lifetime of the inode,
617 * it is ued for making block allocation decisions - we try to
618 * place a file's data blocks near its inode block, and new inodes
619 * near to their parent directory's inode.
620 */
621 ext4_group_t i_block_group;
622 __u32 i_state; /* Dynamic state flags for ext4 */
623
624 ext4_lblk_t i_dir_start_lookup;
625#ifdef CONFIG_EXT4_FS_XATTR
626 /*
627 * Extended attributes can be read independently of the main file
628 * data. Taking i_mutex even when reading would cause contention
629 * between readers of EAs and writers of regular file data, so
630 * instead we synchronize on xattr_sem when reading or changing
631 * EAs.
632 */
633 struct rw_semaphore xattr_sem;
634#endif
d444c3c3
TT
635
636 struct list_head i_orphan; /* unlinked but open inodes */
637
638 /*
639 * i_disksize keeps track of what the inode size is ON DISK, not
640 * in memory. During truncate, i_size is set to the new size by
641 * the VFS prior to calling ext4_truncate(), but the filesystem won't
642 * set i_disksize to 0 until the truncate is actually under way.
643 *
644 * The intent is that i_disksize always represents the blocks which
645 * are used by this file. This allows recovery to restart truncate
646 * on orphans if we crash during truncate. We actually write i_disksize
647 * into the on-disk inode when writing inodes out, instead of i_size.
648 *
649 * The only time when i_disksize and i_size may be different is when
650 * a truncate is in progress. The only things which change i_disksize
651 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
652 */
653 loff_t i_disksize;
654
655 /*
656 * i_data_sem is for serialising ext4_truncate() against
657 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
658 * data tree are chopped off during truncate. We can't do that in
659 * ext4 because whenever we perform intermediate commits during
660 * truncate, the inode and all the metadata blocks *must* be in a
661 * consistent state which allows truncation of the orphans to restart
662 * during recovery. Hence we must fix the get_block-vs-truncate race
663 * by other means, so we have i_data_sem.
664 */
665 struct rw_semaphore i_data_sem;
666 struct inode vfs_inode;
667 struct jbd2_inode jinode;
668
669 struct ext4_ext_cache i_cached_extent;
670 /*
671 * File creation time. Its function is same as that of
672 * struct timespec i_{a,c,m}time in the generic inode.
673 */
674 struct timespec i_crtime;
675
676 /* mballoc */
677 struct list_head i_prealloc_list;
678 spinlock_t i_prealloc_lock;
679
680 /* ialloc */
681 ext4_group_t i_last_alloc_group;
682
683 /* allocation reservation info for delalloc */
684 unsigned int i_reserved_data_blocks;
685 unsigned int i_reserved_meta_blocks;
686 unsigned int i_allocated_meta_blocks;
687 unsigned short i_delalloc_reserved_flag;
688
689 /* on-disk additional length */
690 __u16 i_extra_isize;
691
692 spinlock_t i_block_reservation_lock;
693};
694
ac27a0ec
DK
695/*
696 * File system states
697 */
617ba13b
MC
698#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
699#define EXT4_ERROR_FS 0x0002 /* Errors detected */
700#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 701
469108ff
TT
702/*
703 * Misc. filesystem flags
704 */
705#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
706#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
707#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
708
ac27a0ec
DK
709/*
710 * Mount flags
711 */
617ba13b
MC
712#define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
713#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
714#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
715#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
716#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
717#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
718#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
719#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
720#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
721#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
722#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
723#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
724#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
725#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
726#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
727#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 728#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b
MC
729#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
730#define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
731#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
732#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
733#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
818d276c 734#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
25ec56b5 735#define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
64769240 736#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 737#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 738#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5bf5683a 739
617ba13b
MC
740#define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
741#define set_opt(o, opt) o |= EXT4_MOUNT_##opt
742#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
743 EXT4_MOUNT_##opt)
ac27a0ec 744
617ba13b
MC
745#define ext4_set_bit ext2_set_bit
746#define ext4_set_bit_atomic ext2_set_bit_atomic
747#define ext4_clear_bit ext2_clear_bit
748#define ext4_clear_bit_atomic ext2_clear_bit_atomic
749#define ext4_test_bit ext2_test_bit
750#define ext4_find_first_zero_bit ext2_find_first_zero_bit
751#define ext4_find_next_zero_bit ext2_find_next_zero_bit
aa02ad67 752#define ext4_find_next_bit ext2_find_next_bit
ac27a0ec
DK
753
754/*
755 * Maximal mount counts between two filesystem checks
756 */
617ba13b
MC
757#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
758#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
759
760/*
761 * Behaviour when detecting errors
762 */
617ba13b
MC
763#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
764#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
765#define EXT4_ERRORS_PANIC 3 /* Panic */
766#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec
DK
767
768/*
769 * Structure of the super block
770 */
617ba13b 771struct ext4_super_block {
ac27a0ec 772/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 773 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
774 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
775 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
776/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
777 __le32 s_first_data_block; /* First Data Block */
778 __le32 s_log_block_size; /* Block size */
f077d0d7 779 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
ac27a0ec 780/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
f077d0d7 781 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
ac27a0ec
DK
782 __le32 s_inodes_per_group; /* # Inodes per group */
783 __le32 s_mtime; /* Mount time */
784/*30*/ __le32 s_wtime; /* Write time */
785 __le16 s_mnt_count; /* Mount count */
786 __le16 s_max_mnt_count; /* Maximal mount count */
787 __le16 s_magic; /* Magic signature */
788 __le16 s_state; /* File system state */
789 __le16 s_errors; /* Behaviour when detecting errors */
790 __le16 s_minor_rev_level; /* minor revision level */
791/*40*/ __le32 s_lastcheck; /* time of last check */
792 __le32 s_checkinterval; /* max. time between checks */
793 __le32 s_creator_os; /* OS */
794 __le32 s_rev_level; /* Revision level */
795/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
796 __le16 s_def_resgid; /* Default gid for reserved blocks */
797 /*
617ba13b 798 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
799 *
800 * Note: the difference between the compatible feature set and
801 * the incompatible feature set is that if there is a bit set
802 * in the incompatible feature set that the kernel doesn't
803 * know about, it should refuse to mount the filesystem.
804 *
805 * e2fsck's requirements are more strict; if it doesn't know
806 * about a feature in either the compatible or incompatible
807 * feature set, it must abort and not try to meddle with
808 * things it doesn't understand...
809 */
810 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 811 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
812 __le16 s_block_group_nr; /* block group # of this superblock */
813 __le32 s_feature_compat; /* compatible feature set */
814/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
815 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
816/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
817/*78*/ char s_volume_name[16]; /* volume name */
818/*88*/ char s_last_mounted[64]; /* directory where last mounted */
819/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
820 /*
821 * Performance hints. Directory preallocation should only
617ba13b 822 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
823 */
824 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
825 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
826 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
827 /*
617ba13b 828 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
829 */
830/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
831/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
832 __le32 s_journal_dev; /* device number of journal file */
833 __le32 s_last_orphan; /* start of list of inodes to delete */
834 __le32 s_hash_seed[4]; /* HTREE hash seed */
835 __u8 s_def_hash_version; /* Default hash version to use */
836 __u8 s_reserved_char_pad;
0d1ee42f 837 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 838/*100*/ __le32 s_default_mount_opts;
ac27a0ec 839 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
840 __le32 s_mkfs_time; /* When the filesystem was created */
841 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
842 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
843/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
844 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
845 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
846 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
847 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
848 __le32 s_flags; /* Miscellaneous flags */
849 __le16 s_raid_stride; /* RAID stride */
850 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */
851 __le64 s_mmp_block; /* Block for multi-mount protection */
852 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8
JS
853 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
854 __u8 s_reserved_char_pad2;
855 __le16 s_reserved_pad;
afc32f7e
TT
856 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
857 __u32 s_reserved[160]; /* Padding to the end of the block */
ac27a0ec
DK
858};
859
860#ifdef __KERNEL__
bc0b0d6d
TT
861
862/*
4ab2f15b 863 * run-time mount flags
bc0b0d6d
TT
864 */
865#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 866#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 867
ca0faba0
TT
868/*
869 * fourth extended-fs super-block data in memory
870 */
871struct ext4_sb_info {
872 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
873 unsigned long s_inodes_per_block;/* Number of inodes per block */
874 unsigned long s_blocks_per_group;/* Number of blocks in a group */
875 unsigned long s_inodes_per_group;/* Number of inodes in a group */
876 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
877 unsigned long s_gdb_count; /* Number of group descriptor blocks */
878 unsigned long s_desc_per_block; /* Number of group descriptors per block */
879 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 880 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
ca0faba0
TT
881 unsigned long s_overhead_last; /* Last calculated overhead */
882 unsigned long s_blocks_last; /* Last seen block count */
883 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
884 struct buffer_head * s_sbh; /* Buffer containing the super block */
885 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
886 struct buffer_head **s_group_desc;
7f4520cc 887 unsigned int s_mount_opt;
bc0b0d6d 888 unsigned int s_mount_flags;
ca0faba0
TT
889 ext4_fsblk_t s_sb_block;
890 uid_t s_resuid;
891 gid_t s_resgid;
892 unsigned short s_mount_state;
893 unsigned short s_pad;
894 int s_addr_per_block_bits;
895 int s_desc_per_block_bits;
896 int s_inode_size;
897 int s_first_ino;
898 unsigned int s_inode_readahead_blks;
11013911 899 unsigned int s_inode_goal;
ca0faba0
TT
900 spinlock_t s_next_gen_lock;
901 u32 s_next_generation;
902 u32 s_hash_seed[4];
903 int s_def_hash_version;
904 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
905 struct percpu_counter s_freeblocks_counter;
906 struct percpu_counter s_freeinodes_counter;
907 struct percpu_counter s_dirs_counter;
908 struct percpu_counter s_dirtyblocks_counter;
909 struct blockgroup_lock *s_blockgroup_lock;
910 struct proc_dir_entry *s_proc;
911 struct kobject s_kobj;
912 struct completion s_kobj_unregister;
913
914 /* Journaling */
915 struct inode *s_journal_inode;
916 struct journal_s *s_journal;
917 struct list_head s_orphan;
918 struct mutex s_orphan_lock;
919 struct mutex s_resize_lock;
920 unsigned long s_commit_interval;
921 u32 s_max_batch_time;
922 u32 s_min_batch_time;
923 struct block_device *journal_bdev;
924#ifdef CONFIG_JBD2_DEBUG
925 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
926 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
927#endif
928#ifdef CONFIG_QUOTA
929 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
930 int s_jquota_fmt; /* Format of quota to use */
931#endif
932 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 933 struct rb_root system_blks;
ca0faba0
TT
934
935#ifdef EXTENTS_STATS
936 /* ext4 extents stats */
937 unsigned long s_ext_min;
938 unsigned long s_ext_max;
939 unsigned long s_depth_max;
940 spinlock_t s_ext_stats_lock;
941 unsigned long s_ext_blocks;
942 unsigned long s_ext_extents;
943#endif
944
945 /* for buddy allocator */
946 struct ext4_group_info ***s_group_info;
947 struct inode *s_buddy_cache;
948 long s_blocks_reserved;
949 spinlock_t s_reserve_lock;
950 spinlock_t s_md_lock;
951 tid_t s_last_transaction;
952 unsigned short *s_mb_offsets;
953 unsigned int *s_mb_maxs;
954
955 /* tunables */
956 unsigned long s_stripe;
957 unsigned int s_mb_stream_request;
958 unsigned int s_mb_max_to_scan;
959 unsigned int s_mb_min_to_scan;
960 unsigned int s_mb_stats;
961 unsigned int s_mb_order2_reqs;
962 unsigned int s_mb_group_prealloc;
55138e0b 963 unsigned int s_max_writeback_mb_bump;
ca0faba0
TT
964 /* where last allocation was done - for stream allocation */
965 unsigned long s_mb_last_group;
966 unsigned long s_mb_last_start;
967
968 /* history to debug policy */
969 struct ext4_mb_history *s_mb_history;
970 int s_mb_history_cur;
971 int s_mb_history_max;
972 int s_mb_history_num;
973 spinlock_t s_mb_history_lock;
974 int s_mb_history_filter;
975
976 /* stats for buddy allocator */
977 spinlock_t s_mb_pa_lock;
978 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
979 atomic_t s_bal_success; /* we found long enough chunks */
980 atomic_t s_bal_allocated; /* in blocks */
981 atomic_t s_bal_ex_scanned; /* total extents scanned */
982 atomic_t s_bal_goals; /* goal hits */
983 atomic_t s_bal_breaks; /* too long searches */
984 atomic_t s_bal_2orders; /* 2^order hits */
985 spinlock_t s_bal_lock;
986 unsigned long s_mb_buddies_generated;
987 unsigned long long s_mb_generation_time;
988 atomic_t s_mb_lost_chunks;
989 atomic_t s_mb_preallocated;
990 atomic_t s_mb_discarded;
50797481 991 atomic_t s_lock_busy;
ca0faba0
TT
992
993 /* locality groups */
994 struct ext4_locality_group *s_locality_groups;
995
996 /* for write statistics */
997 unsigned long s_sectors_written_start;
998 u64 s_kbytes_written;
999
1000 unsigned int s_log_groups_per_flex;
1001 struct flex_groups *s_flex_groups;
4c0425ff
MC
1002
1003 /* workqueue for dio unwritten */
1004 struct workqueue_struct *dio_unwritten_wq;
ca0faba0
TT
1005};
1006
af5bc92d 1007static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1008{
1009 return sb->s_fs_info;
1010}
617ba13b 1011static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1012{
617ba13b 1013 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1014}
1015
ef7f3835
KS
1016static inline struct timespec ext4_current_time(struct inode *inode)
1017{
1018 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1019 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1020}
1021
617ba13b 1022static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1023{
617ba13b
MC
1024 return ino == EXT4_ROOT_INO ||
1025 ino == EXT4_JOURNAL_INO ||
1026 ino == EXT4_RESIZE_INO ||
1027 (ino >= EXT4_FIRST_INO(sb) &&
1028 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec
DK
1029}
1030#else
617ba13b 1031/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1032 * a kernel struct super_block. This will allow us to call the feature-test
1033 * macros from user land. */
617ba13b 1034#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1035#endif
1036
617ba13b 1037#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1038
1039/*
1040 * Codes for operating systems
1041 */
617ba13b
MC
1042#define EXT4_OS_LINUX 0
1043#define EXT4_OS_HURD 1
1044#define EXT4_OS_MASIX 2
1045#define EXT4_OS_FREEBSD 3
1046#define EXT4_OS_LITES 4
ac27a0ec
DK
1047
1048/*
1049 * Revision levels
1050 */
617ba13b
MC
1051#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1052#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1053
617ba13b
MC
1054#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1055#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1056
617ba13b 1057#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1058
1059/*
1060 * Feature set definitions
1061 */
1062
617ba13b 1063#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1064 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1065#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1066 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1067#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1068 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1069#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1070 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1071#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1072 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1073#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1074 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1075#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1076 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1077#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1078 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1079#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1080 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1081
1082#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1083#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1084#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1085#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1086#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1087#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1088
1089#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1090#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1091#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1092#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1093#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1094#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1095#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
617ba13b
MC
1096
1097#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1098#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1099#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1100#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1101#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1102#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1103#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1104#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1105#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
617ba13b
MC
1106
1107#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1108#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1109 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1110 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1111 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581
JS
1112 EXT4_FEATURE_INCOMPAT_64BIT| \
1113 EXT4_FEATURE_INCOMPAT_FLEX_BG)
617ba13b
MC
1114#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1115 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1116 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1117 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1118 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451
AK
1119 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1120 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
ac27a0ec
DK
1121
1122/*
1123 * Default values for user and/or group using reserved blocks
1124 */
617ba13b
MC
1125#define EXT4_DEF_RESUID 0
1126#define EXT4_DEF_RESGID 0
ac27a0ec 1127
240799cd
TT
1128#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1129
ac27a0ec
DK
1130/*
1131 * Default mount options
1132 */
617ba13b
MC
1133#define EXT4_DEFM_DEBUG 0x0001
1134#define EXT4_DEFM_BSDGROUPS 0x0002
1135#define EXT4_DEFM_XATTR_USER 0x0004
1136#define EXT4_DEFM_ACL 0x0008
1137#define EXT4_DEFM_UID16 0x0010
1138#define EXT4_DEFM_JMODE 0x0060
1139#define EXT4_DEFM_JMODE_DATA 0x0020
1140#define EXT4_DEFM_JMODE_ORDERED 0x0040
1141#define EXT4_DEFM_JMODE_WBACK 0x0060
ac27a0ec 1142
30773840
TT
1143/*
1144 * Default journal batch times
1145 */
1146#define EXT4_DEF_MIN_BATCH_TIME 0
1147#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1148
a4912123
TT
1149/*
1150 * Minimum number of groups in a flexgroup before we separate out
1151 * directories into the first block group of a flexgroup
1152 */
1153#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1154
ac27a0ec
DK
1155/*
1156 * Structure of a directory entry
1157 */
617ba13b 1158#define EXT4_NAME_LEN 255
ac27a0ec 1159
617ba13b 1160struct ext4_dir_entry {
ac27a0ec
DK
1161 __le32 inode; /* Inode number */
1162 __le16 rec_len; /* Directory entry length */
1163 __le16 name_len; /* Name length */
617ba13b 1164 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1165};
1166
1167/*
617ba13b 1168 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1169 * stored in intel byte order, and the name_len field could never be
1170 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1171 * file_type field.
1172 */
617ba13b 1173struct ext4_dir_entry_2 {
ac27a0ec
DK
1174 __le32 inode; /* Inode number */
1175 __le16 rec_len; /* Directory entry length */
1176 __u8 name_len; /* Name length */
1177 __u8 file_type;
617ba13b 1178 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1179};
1180
1181/*
617ba13b 1182 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1183 * other bits are reserved for now.
1184 */
617ba13b
MC
1185#define EXT4_FT_UNKNOWN 0
1186#define EXT4_FT_REG_FILE 1
1187#define EXT4_FT_DIR 2
1188#define EXT4_FT_CHRDEV 3
1189#define EXT4_FT_BLKDEV 4
1190#define EXT4_FT_FIFO 5
1191#define EXT4_FT_SOCK 6
1192#define EXT4_FT_SYMLINK 7
ac27a0ec 1193
617ba13b 1194#define EXT4_FT_MAX 8
ac27a0ec
DK
1195
1196/*
617ba13b 1197 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1198 *
1199 * NOTE: It must be a multiple of 4
1200 */
617ba13b
MC
1201#define EXT4_DIR_PAD 4
1202#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1203#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1204 ~EXT4_DIR_ROUND)
a72d7f83
JK
1205#define EXT4_MAX_REC_LEN ((1<<16)-1)
1206
ac27a0ec
DK
1207/*
1208 * Hash Tree Directory indexing
1209 * (c) Daniel Phillips, 2001
1210 */
1211
4074fe37
ES
1212#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1213 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
617ba13b
MC
1214 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1215#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1216#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1217
1218/* Legal values for the dx_root hash_version field: */
1219
1220#define DX_HASH_LEGACY 0
1221#define DX_HASH_HALF_MD4 1
1222#define DX_HASH_TEA 2
f99b2589
TT
1223#define DX_HASH_LEGACY_UNSIGNED 3
1224#define DX_HASH_HALF_MD4_UNSIGNED 4
1225#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec
DK
1226
1227#ifdef __KERNEL__
1228
1229/* hash info structure used by the directory hash */
1230struct dx_hash_info
1231{
1232 u32 hash;
1233 u32 minor_hash;
1234 int hash_version;
1235 u32 *seed;
1236};
1237
617ba13b 1238#define EXT4_HTREE_EOF 0x7fffffff
ac27a0ec
DK
1239
1240/*
617ba13b 1241 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1242 */
1243#define HASH_NB_ALWAYS 1
1244
1245
1246/*
1247 * Describe an inode's exact location on disk and in memory
1248 */
617ba13b 1249struct ext4_iloc
ac27a0ec
DK
1250{
1251 struct buffer_head *bh;
1252 unsigned long offset;
fd2d4291 1253 ext4_group_t block_group;
ac27a0ec
DK
1254};
1255
617ba13b 1256static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1257{
617ba13b 1258 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1259}
1260
1261/*
1262 * This structure is stuffed into the struct file's private_data field
1263 * for directories. It is where we put information so that we can do
1264 * readdir operations in hash tree order.
1265 */
1266struct dir_private_info {
1267 struct rb_root root;
1268 struct rb_node *curr_node;
1269 struct fname *extra_fname;
1270 loff_t last_pos;
1271 __u32 curr_hash;
1272 __u32 curr_minor_hash;
1273 __u32 next_hash;
1274};
1275
1276/* calculate the first block number of the group */
617ba13b 1277static inline ext4_fsblk_t
fd2d4291 1278ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1279{
617ba13b
MC
1280 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1281 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1282}
1283
1284/*
1285 * Special error return code only used by dx_probe() and its callers.
1286 */
1287#define ERR_BAD_DX_DIR -75000
1288
72b64b59 1289void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
a9df9a49 1290 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
3a5b2ecd 1291
9f6200bb
TT
1292extern struct proc_dir_entry *ext4_proc_root;
1293
ac27a0ec
DK
1294/*
1295 * Function prototypes
1296 */
1297
1298/*
1299 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1300 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1301 */
8c55e204
DK
1302# define NORET_TYPE /**/
1303# define ATTRIB_NORET __attribute__((noreturn))
1304# define NORET_AND noreturn,
ac27a0ec 1305
498e5f24
TT
1306/* bitmap.c */
1307extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1308
ac27a0ec 1309/* balloc.c */
3a5b2ecd
MC
1310extern unsigned int ext4_block_group(struct super_block *sb,
1311 ext4_fsblk_t blocknr);
1312extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1313 ext4_fsblk_t blocknr);
fd2d4291
AM
1314extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1315extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1316 ext4_group_t group);
7061eba7 1317extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
617ba13b 1318 ext4_fsblk_t goal, unsigned long *count, int *errp);
5c791616 1319extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
8c3bf8a0 1320extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
af5bc92d 1321extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
c9de560d 1322 ext4_fsblk_t block, unsigned long count, int metadata);
e21675d4
AK
1323extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1324 ext4_fsblk_t block, unsigned long count);
af5bc92d
TT
1325extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1326extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1327extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1328 ext4_group_t block_group,
ac27a0ec 1329 struct buffer_head ** bh);
617ba13b 1330extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
bb23c20a
TT
1331struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1332 ext4_group_t block_group);
1333extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1334 struct buffer_head *bh,
1335 ext4_group_t group,
1336 struct ext4_group_desc *desc);
1337#define ext4_free_blocks_after_init(sb, group, desc) \
1338 ext4_init_block_bitmap(sb, NULL, group, desc)
ac27a0ec
DK
1339
1340/* dir.c */
617ba13b
MC
1341extern int ext4_check_dir_entry(const char *, struct inode *,
1342 struct ext4_dir_entry_2 *,
498e5f24 1343 struct buffer_head *, unsigned int);
617ba13b 1344extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 1345 __u32 minor_hash,
617ba13b
MC
1346 struct ext4_dir_entry_2 *dirent);
1347extern void ext4_htree_free_dir_info(struct dir_private_info *p);
ac27a0ec
DK
1348
1349/* fsync.c */
af5bc92d 1350extern int ext4_sync_file(struct file *, struct dentry *, int);
ac27a0ec
DK
1351
1352/* hash.c */
617ba13b 1353extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
1354 dx_hash_info *hinfo);
1355
1356/* ialloc.c */
f157a4aa 1357extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
11013911 1358 const struct qstr *qstr, __u32 goal);
af5bc92d
TT
1359extern void ext4_free_inode(handle_t *, struct inode *);
1360extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1361extern unsigned long ext4_count_free_inodes(struct super_block *);
1362extern unsigned long ext4_count_dirs(struct super_block *);
1363extern void ext4_check_inodes_bitmap(struct super_block *);
bb23c20a
TT
1364extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1365 struct buffer_head *bh,
1366 ext4_group_t group,
1367 struct ext4_group_desc *desc);
1368extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
ac27a0ec 1369
c9de560d
AT
1370/* mballoc.c */
1371extern long ext4_mb_stats;
1372extern long ext4_mb_max_to_scan;
1373extern int ext4_mb_init(struct super_block *, int);
1374extern int ext4_mb_release(struct super_block *);
1375extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1376 struct ext4_allocation_request *, int *);
1377extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 1378extern void ext4_discard_preallocations(struct inode *);
c9de560d
AT
1379extern int __init init_ext4_mballoc(void);
1380extern void exit_ext4_mballoc(void);
1381extern void ext4_mb_free_blocks(handle_t *, struct inode *,
0610b6e9 1382 ext4_fsblk_t, unsigned long, int, unsigned long *);
920313a7 1383extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 1384 ext4_group_t i, struct ext4_group_desc *desc);
920313a7
AK
1385extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1386extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1387 ext4_group_t, int);
ac27a0ec 1388/* inode.c */
617ba13b
MC
1389int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
1390 struct buffer_head *bh, ext4_fsblk_t blocknr);
725d26d3
AK
1391struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1392 ext4_lblk_t, int, int *);
1393struct buffer_head *ext4_bread(handle_t *, struct inode *,
1394 ext4_lblk_t, int, int *);
6873fa0d
ES
1395int ext4_get_block(struct inode *inode, sector_t iblock,
1396 struct buffer_head *bh_result, int create);
ac27a0ec 1397
1d1fe1ee 1398extern struct inode *ext4_iget(struct super_block *, unsigned long);
af5bc92d
TT
1399extern int ext4_write_inode(struct inode *, int);
1400extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
1401extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1402 struct kstat *stat);
af5bc92d
TT
1403extern void ext4_delete_inode(struct inode *);
1404extern int ext4_sync_inode(handle_t *, struct inode *);
617ba13b
MC
1405extern void ext4_dirty_inode(struct inode *);
1406extern int ext4_change_inode_journal_flag(struct inode *, int);
1407extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
91ef4caf 1408extern int ext4_can_truncate(struct inode *inode);
af5bc92d 1409extern void ext4_truncate(struct inode *);
487caeef 1410extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 1411extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 1412extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 1413extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 1414extern void ext4_set_aops(struct inode *inode);
a86c6181 1415extern int ext4_writepage_trans_blocks(struct inode *);
a02908f1 1416extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
f3bd1f3f 1417extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
cf108bca 1418extern int ext4_block_truncate_page(handle_t *handle,
a86c6181 1419 struct address_space *mapping, loff_t from);
c2ec175c 1420extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
60e58e0f 1421extern qsize_t ext4_get_reserved_space(struct inode *inode);
ac27a0ec
DK
1422
1423/* ioctl.c */
5cdd7b2d 1424extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 1425extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 1426
c14c6fd5 1427/* migrate.c */
2a43a878 1428extern int ext4_ext_migrate(struct inode *);
3d0518f4 1429
ac27a0ec 1430/* namei.c */
3d0518f4
WY
1431extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1432extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
617ba13b
MC
1433extern int ext4_orphan_add(handle_t *, struct inode *);
1434extern int ext4_orphan_del(handle_t *, struct inode *);
1435extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec
DK
1436 __u32 start_minor_hash, __u32 *next_hash);
1437
1438/* resize.c */
617ba13b
MC
1439extern int ext4_group_add(struct super_block *sb,
1440 struct ext4_new_group_data *input);
1441extern int ext4_group_extend(struct super_block *sb,
1442 struct ext4_super_block *es,
1443 ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
1444
1445/* super.c */
af5bc92d 1446extern void ext4_error(struct super_block *, const char *, const char *, ...)
ac27a0ec 1447 __attribute__ ((format (printf, 3, 4)));
af5bc92d
TT
1448extern void __ext4_std_error(struct super_block *, const char *, int);
1449extern void ext4_abort(struct super_block *, const char *, const char *, ...)
ac27a0ec 1450 __attribute__ ((format (printf, 3, 4)));
af5bc92d 1451extern void ext4_warning(struct super_block *, const char *, const char *, ...)
ac27a0ec 1452 __attribute__ ((format (printf, 3, 4)));
b31e1552
ES
1453extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1454 __attribute__ ((format (printf, 3, 4)));
5d1b1b3f
AK
1455extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1456 const char *, const char *, ...)
1457 __attribute__ ((format (printf, 4, 5)));
af5bc92d 1458extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
1459extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1460 __u32 compat);
1461extern int ext4_update_rocompat_feature(handle_t *handle,
1462 struct super_block *sb, __u32 rocompat);
1463extern int ext4_update_incompat_feature(handle_t *handle,
1464 struct super_block *sb, __u32 incompat);
8fadc143
AR
1465extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1466 struct ext4_group_desc *bg);
1467extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1468 struct ext4_group_desc *bg);
1469extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1470 struct ext4_group_desc *bg);
560671a0
AK
1471extern __u32 ext4_free_blks_count(struct super_block *sb,
1472 struct ext4_group_desc *bg);
1473extern __u32 ext4_free_inodes_count(struct super_block *sb,
1474 struct ext4_group_desc *bg);
1475extern __u32 ext4_used_dirs_count(struct super_block *sb,
1476 struct ext4_group_desc *bg);
1477extern __u32 ext4_itable_unused_count(struct super_block *sb,
1478 struct ext4_group_desc *bg);
8fadc143
AR
1479extern void ext4_block_bitmap_set(struct super_block *sb,
1480 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1481extern void ext4_inode_bitmap_set(struct super_block *sb,
1482 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1483extern void ext4_inode_table_set(struct super_block *sb,
1484 struct ext4_group_desc *bg, ext4_fsblk_t blk);
560671a0
AK
1485extern void ext4_free_blks_set(struct super_block *sb,
1486 struct ext4_group_desc *bg, __u32 count);
1487extern void ext4_free_inodes_set(struct super_block *sb,
1488 struct ext4_group_desc *bg, __u32 count);
1489extern void ext4_used_dirs_set(struct super_block *sb,
1490 struct ext4_group_desc *bg, __u32 count);
1491extern void ext4_itable_unused_set(struct super_block *sb,
1492 struct ext4_group_desc *bg, __u32 count);
bb23c20a
TT
1493extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1494 struct ext4_group_desc *gdp);
1495extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1496 struct ext4_group_desc *gdp);
bd81d8ee
LV
1497
1498static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1499{
1500 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 1501 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
1502}
1503
1504static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1505{
1506 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 1507 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
1508}
1509
1510static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1511{
1512 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 1513 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
1514}
1515
1516static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1517 ext4_fsblk_t blk)
1518{
6bc9feff 1519 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1520 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1521}
1522
1523static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1524 ext4_fsblk_t blk)
1525{
308ba3ec 1526 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1527 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1528}
1529
1530static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1531 ext4_fsblk_t blk)
1532{
308ba3ec 1533 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1534 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1535}
1536
a48380f7
AK
1537static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1538{
06a279d6
TT
1539 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1540 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1541 le32_to_cpu(raw_inode->i_size_lo);
1542 else
1543 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 1544}
bd81d8ee 1545
a48380f7
AK
1546static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1547{
1548 raw_inode->i_size_lo = cpu_to_le32(i_size);
1549 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1550}
ac27a0ec 1551
c9de560d
AT
1552static inline
1553struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 1554 ext4_group_t group)
c9de560d
AT
1555{
1556 struct ext4_group_info ***grp_info;
1557 long indexv, indexh;
1558 grp_info = EXT4_SB(sb)->s_group_info;
1559 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1560 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1561 return grp_info[indexv][indexh];
1562}
1563
8df9675f
TT
1564/*
1565 * Reading s_groups_count requires using smp_rmb() afterwards. See
1566 * the locking protocol documented in the comments of ext4_group_add()
1567 * in resize.c
1568 */
1569static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1570{
1571 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
1572
1573 smp_rmb();
1574 return ngroups;
1575}
c9de560d 1576
772cb7c8
JS
1577static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1578 ext4_group_t block_group)
1579{
1580 return block_group >> sbi->s_log_groups_per_flex;
1581}
1582
1583static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1584{
1585 return 1 << sbi->s_log_groups_per_flex;
1586}
1587
617ba13b 1588#define ext4_std_error(sb, errno) \
ac27a0ec
DK
1589do { \
1590 if ((errno)) \
4db9c54a 1591 __ext4_std_error((sb), __func__, (errno)); \
ac27a0ec
DK
1592} while (0)
1593
a30d542a 1594#ifdef CONFIG_SMP
179f7ebf 1595/* Each CPU can accumulate percpu_counter_batch blocks in their local
a30d542a 1596 * counters. So we need to make sure we have free blocks more
179f7ebf 1597 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 1598 */
179f7ebf 1599#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a
AK
1600#else
1601#define EXT4_FREEBLOCKS_WATERMARK 0
1602#endif
1603
cf17fea6
AK
1604static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1605{
1606 /*
1607 * XXX: replace with spinlock if seen contended -bzzz
1608 */
1609 down_write(&EXT4_I(inode)->i_data_sem);
1610 if (newsize > EXT4_I(inode)->i_disksize)
1611 EXT4_I(inode)->i_disksize = newsize;
1612 up_write(&EXT4_I(inode)->i_data_sem);
1613 return ;
1614}
a30d542a 1615
5d1b1b3f
AK
1616struct ext4_group_info {
1617 unsigned long bb_state;
1618 struct rb_root bb_free_root;
a36b4498
ES
1619 ext4_grpblk_t bb_first_free; /* first free block */
1620 ext4_grpblk_t bb_free; /* total free blocks */
1621 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
5d1b1b3f
AK
1622 struct list_head bb_prealloc_list;
1623#ifdef DOUBLE_CHECK
1624 void *bb_bitmap;
1625#endif
1626 struct rw_semaphore alloc_sem;
a36b4498
ES
1627 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
1628 * regions, index is order.
1629 * bb_counters[3] = 5 means
1630 * 5 free 8-block regions. */
5d1b1b3f
AK
1631};
1632
1633#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
5d1b1b3f
AK
1634
1635#define EXT4_MB_GRP_NEED_INIT(grp) \
1636 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1637
50797481
TT
1638#define EXT4_MAX_CONTENTION 8
1639#define EXT4_CONTENTION_THRESHOLD 2
1640
955ce5f5
AK
1641static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1642 ext4_group_t group)
5d1b1b3f 1643{
955ce5f5
AK
1644 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1645}
5d1b1b3f 1646
50797481
TT
1647/*
1648 * Returns true if the filesystem is busy enough that attempts to
1649 * access the block group locks has run into contention.
1650 */
1651static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1652{
1653 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1654}
1655
955ce5f5
AK
1656static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1657{
50797481
TT
1658 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1659 if (spin_trylock(lock))
1660 /*
1661 * We're able to grab the lock right away, so drop the
1662 * lock contention counter.
1663 */
1664 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1665 else {
1666 /*
1667 * The lock is busy, so bump the contention counter,
1668 * and then wait on the spin lock.
1669 */
1670 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1671 EXT4_MAX_CONTENTION);
1672 spin_lock(lock);
1673 }
5d1b1b3f
AK
1674}
1675
1676static inline void ext4_unlock_group(struct super_block *sb,
1677 ext4_group_t group)
1678{
955ce5f5 1679 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
1680}
1681
ac27a0ec
DK
1682/*
1683 * Inodes and files operations
1684 */
1685
1686/* dir.c */
617ba13b 1687extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
1688
1689/* file.c */
c5ef1c42 1690extern const struct inode_operations ext4_file_inode_operations;
617ba13b 1691extern const struct file_operations ext4_file_operations;
ac27a0ec
DK
1692
1693/* namei.c */
c5ef1c42
AV
1694extern const struct inode_operations ext4_dir_inode_operations;
1695extern const struct inode_operations ext4_special_inode_operations;
596397b7 1696extern struct dentry *ext4_get_parent(struct dentry *child);
ac27a0ec
DK
1697
1698/* symlink.c */
c5ef1c42
AV
1699extern const struct inode_operations ext4_symlink_inode_operations;
1700extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 1701
6fd058f7
TT
1702/* block_validity */
1703extern void ext4_release_system_zone(struct super_block *sb);
1704extern int ext4_setup_system_zone(struct super_block *sb);
1705extern int __init init_ext4_system_zone(void);
1706extern void exit_ext4_system_zone(void);
1707extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1708 ext4_fsblk_t start_blk,
1709 unsigned int count);
1710
a86c6181
AT
1711/* extents.c */
1712extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1713extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
a02908f1
MC
1714extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1715 int chunk);
a86c6181 1716extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
498e5f24 1717 ext4_lblk_t iblock, unsigned int max_blocks,
c2177057 1718 struct buffer_head *bh_result, int flags);
cf108bca 1719extern void ext4_ext_truncate(struct inode *);
a86c6181
AT
1720extern void ext4_ext_init(struct super_block *);
1721extern void ext4_ext_release(struct super_block *);
a2df2a63
AA
1722extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1723 loff_t len);
0031462b
MC
1724extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1725 loff_t len);
12b7ac17
TT
1726extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1727 sector_t block, unsigned int max_blocks,
c2177057 1728 struct buffer_head *bh, int flags);
3a06d778
AK
1729extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1730 __u64 start, __u64 len);
748de673
AF
1731/* move_extent.c */
1732extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1733 __u64 start_orig, __u64 start_donor,
1734 __u64 len, __u64 *moved_len);
1735
3a06d778 1736
2ccb5fb9
AK
1737/*
1738 * Add new method to test wether block and inode bitmaps are properly
1739 * initialized. With uninit_bg reading the block from disk is not enough
1740 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1741 */
1742#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1743
1744static inline int bitmap_uptodate(struct buffer_head *bh)
1745{
1746 return (buffer_uptodate(bh) &&
1747 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1748}
1749static inline void set_bitmap_uptodate(struct buffer_head *bh)
1750{
1751 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1752}
1753
ac27a0ec
DK
1754#endif /* __KERNEL__ */
1755
3dcf5451 1756#endif /* _EXT4_H */