]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - fs/ext4/ext4.h
tracing: Convert more ext4 events to DEFINE_EVENT
[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
273df556 56#define EXT4_ERROR_INODE(inode, fmt, a...) \
24676da4 57 ext4_error_inode(__func__, (inode), (fmt), ## a)
273df556
FM
58
59#define EXT4_ERROR_FILE(file, fmt, a...) \
24676da4 60 ext4_error_file(__func__, (file), (fmt), ## a)
273df556 61
d444c3c3
TT
62/* data type for block offset of block group */
63typedef int ext4_grpblk_t;
64
65/* data type for filesystem-wide blocks number */
66typedef unsigned long long ext4_fsblk_t;
67
68/* data type for file logical block number */
69typedef __u32 ext4_lblk_t;
70
71/* data type for block group number */
72typedef unsigned int ext4_group_t;
73
296c355c
TT
74/*
75 * Flags used in mballoc's allocation_context flags field.
76 *
77 * Also used to show what's going on for debugging purposes when the
78 * flag field is exported via the traceport interface
79 */
d444c3c3 80
c9de560d 81/* prefer goal again. length */
0ef90db9 82#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 83/* blocks already reserved */
0ef90db9 84#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 85/* metadata is being allocated */
0ef90db9 86#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 87/* first blocks in the file */
0ef90db9 88#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 89/* search for the best chunk */
0ef90db9 90#define EXT4_MB_HINT_BEST 0x0010
c9de560d 91/* data is being allocated */
0ef90db9 92#define EXT4_MB_HINT_DATA 0x0020
c9de560d 93/* don't preallocate (for tails) */
0ef90db9 94#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 95/* allocate for locality group */
0ef90db9 96#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 97/* allocate goal blocks or none */
0ef90db9 98#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 99/* goal is meaningful */
0ef90db9 100#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 101/* blocks already pre-reserved by delayed allocation */
0ef90db9 102#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
103/* We are doing stream allocation */
104#define EXT4_MB_STREAM_ALLOC 0x0800
d2a17637 105
c9de560d
AT
106
107struct ext4_allocation_request {
108 /* target inode for block we're allocating */
109 struct inode *inode;
726447d8
ES
110 /* how many blocks we want to allocate */
111 unsigned int len;
c9de560d
AT
112 /* logical block in target inode */
113 ext4_lblk_t logical;
c9de560d
AT
114 /* the closest logical allocated block to the left */
115 ext4_lblk_t lleft;
c9de560d
AT
116 /* the closest logical allocated block to the right */
117 ext4_lblk_t lright;
726447d8
ES
118 /* phys. target (a hint) */
119 ext4_fsblk_t goal;
120 /* phys. block for the closest logical allocated block to the left */
121 ext4_fsblk_t pleft;
122 /* phys. block for the closest logical allocated block to the right */
c9de560d 123 ext4_fsblk_t pright;
c9de560d 124 /* flags. see above EXT4_MB_HINT_* */
498e5f24 125 unsigned int flags;
c9de560d
AT
126};
127
e35fd660
TT
128/*
129 * Logical to physical block mapping, used by ext4_map_blocks()
130 *
131 * This structure is used to pass requests into ext4_map_blocks() as
132 * well as to store the information returned by ext4_map_blocks(). It
133 * takes less room on the stack than a struct buffer_head.
134 */
135#define EXT4_MAP_NEW (1 << BH_New)
136#define EXT4_MAP_MAPPED (1 << BH_Mapped)
137#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
138#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
139#define EXT4_MAP_UNINIT (1 << BH_Uninit)
140#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
141 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
142 EXT4_MAP_UNINIT)
143
144struct ext4_map_blocks {
145 ext4_fsblk_t m_pblk;
146 ext4_lblk_t m_lblk;
147 unsigned int m_len;
148 unsigned int m_flags;
149};
150
b3a3ca8c
TT
151/*
152 * For delayed allocation tracking
153 */
154struct mpage_da_data {
155 struct inode *inode;
156 sector_t b_blocknr; /* start block number of extent */
157 size_t b_size; /* size of extent */
158 unsigned long b_state; /* state of the extent */
159 unsigned long first_page, next_page; /* extent of pages */
160 struct writeback_control *wbc;
161 int io_done;
162 int pages_written;
163 int retval;
164};
c7064ef1 165#define EXT4_IO_UNWRITTEN 0x1
0031462b 166typedef struct ext4_io_end {
8d5d02e6 167 struct list_head list; /* per-file finished AIO list */
0031462b 168 struct inode *inode; /* file being written to */
8d5d02e6 169 unsigned int flag; /* unwritten or not */
744692dc 170 struct page *page; /* page struct for buffer write */
a1de02dc
ES
171 loff_t offset; /* offset in the file */
172 ssize_t size; /* size of the extent */
0031462b
MC
173 struct work_struct work; /* data work queue */
174} ext4_io_end_t;
175
ac27a0ec
DK
176/*
177 * Special inodes numbers
178 */
617ba13b
MC
179#define EXT4_BAD_INO 1 /* Bad blocks inode */
180#define EXT4_ROOT_INO 2 /* Root inode */
181#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
182#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
183#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
184#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 185
617ba13b
MC
186/* First non-reserved inode for old ext4 filesystems */
187#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
188
189/*
190 * Maximal count of links to a file
191 */
f8628a14 192#define EXT4_LINK_MAX 65000
ac27a0ec
DK
193
194/*
195 * Macro-instructions used to manage several block sizes
196 */
617ba13b 197#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
198#define EXT4_MAX_BLOCK_SIZE 65536
199#define EXT4_MIN_BLOCK_LOG_SIZE 10
ac27a0ec 200#ifdef __KERNEL__
617ba13b 201# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 202#else
617ba13b 203# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 204#endif
af5bc92d 205#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
ac27a0ec 206#ifdef __KERNEL__
617ba13b 207# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
ac27a0ec 208#else
617ba13b 209# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
210#endif
211#ifdef __KERNEL__
617ba13b
MC
212#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
213#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
214#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 215#else
617ba13b
MC
216#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
217 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 218 (s)->s_inode_size)
617ba13b
MC
219#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
220 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
221 (s)->s_first_ino)
222#endif
a2df2a63 223#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 224
ac27a0ec
DK
225/*
226 * Structure of a blocks group descriptor
227 */
617ba13b 228struct ext4_group_desc
ac27a0ec 229{
3a14589c 230 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
231 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
232 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
233 __le16 bg_free_blocks_count_lo;/* Free blocks count */
234 __le16 bg_free_inodes_count_lo;/* Free inodes count */
235 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4
AD
236 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
237 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
560671a0 238 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 239 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
240 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
241 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
242 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
243 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
244 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
245 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 246 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
91b51a01 247 __u32 bg_reserved2[3];
ac27a0ec
DK
248};
249
772cb7c8
JS
250/*
251 * Structure of a flex block group info
252 */
253
254struct flex_groups {
9f24e420
TT
255 atomic_t free_inodes;
256 atomic_t free_blocks;
7d39db14 257 atomic_t used_dirs;
772cb7c8
JS
258};
259
717d50e4
AD
260#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
261#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
262#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
263
ac27a0ec
DK
264/*
265 * Macro-instructions used to manage group descriptors
266 */
0d1ee42f 267#define EXT4_MIN_DESC_SIZE 32
8fadc143 268#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
269#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
270#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 271#ifdef __KERNEL__
617ba13b
MC
272# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
273# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
274# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
275# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 276#else
617ba13b 277# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 278# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 279# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
280#endif
281
282/*
283 * Constants relative to the data blocks
284 */
617ba13b
MC
285#define EXT4_NDIR_BLOCKS 12
286#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
287#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
288#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
289#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
290
291/*
292 * Inode flags
293 */
617ba13b
MC
294#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
295#define EXT4_UNRM_FL 0x00000002 /* Undelete */
296#define EXT4_COMPR_FL 0x00000004 /* Compress file */
297#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
298#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
299#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
300#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
301#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 302/* Reserved for compression usage... */
617ba13b
MC
303#define EXT4_DIRTY_FL 0x00000100
304#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
305#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
306#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
ac27a0ec 307/* End compression flags --- maybe not all used */
617ba13b
MC
308#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
309#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
310#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
311#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
312#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
313#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 314#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 315#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
f710b4b9 316#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
c8d46e41 317#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
8180a562 318#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 319
c8d46e41
JZ
320#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
321#define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
ac27a0ec 322
8fa43a81
DG
323/* Flags that should be inherited by new inodes from their parent. */
324#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
325 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
326 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
327 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
328 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
329
2dc6b0d4
DG
330/* Flags that are appropriate for regular files (all but dir-specific ones). */
331#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
332
333/* Flags that are appropriate for non-directories/regular files. */
334#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
335
336/* Mask out flags that are inappropriate for the given type of inode. */
337static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
338{
339 if (S_ISDIR(mode))
340 return flags;
341 else if (S_ISREG(mode))
342 return flags & EXT4_REG_FLMASK;
343 else
344 return flags & EXT4_OTHER_FLMASK;
345}
346
12e9b892
DM
347/*
348 * Inode flags used for atomic set/get
349 */
350enum {
351 EXT4_INODE_SECRM = 0, /* Secure deletion */
352 EXT4_INODE_UNRM = 1, /* Undelete */
353 EXT4_INODE_COMPR = 2, /* Compress file */
354 EXT4_INODE_SYNC = 3, /* Synchronous updates */
355 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
356 EXT4_INODE_APPEND = 5, /* writes to file may only append */
357 EXT4_INODE_NODUMP = 6, /* do not dump file */
358 EXT4_INODE_NOATIME = 7, /* do not update atime */
359/* Reserved for compression usage... */
360 EXT4_INODE_DIRTY = 8,
361 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
362 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
363 EXT4_INODE_ECOMPR = 11, /* Compression error */
364/* End compression flags --- maybe not all used */
365 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
366 EXT4_INODE_IMAGIC = 13, /* AFS directory */
367 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
368 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
369 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
370 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
371 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
372 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
373 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
374 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
375 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
376};
377
378#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
379#define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
380 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
381 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
382
383/*
384 * Since it's pretty easy to mix up bit numbers and hex values, and we
385 * can't do a compile-time test for ENUM values, we use a run-time
386 * test to make sure that EXT4_XXX_FL is consistent with respect to
387 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
388 * out so it won't cost any extra space in the compiled kernel image.
389 * But it's important that these values are the same, since we are
390 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
391 * must be consistent with the values of FS_XXX_FL defined in
392 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
393 * ext4 filesystems, and of course the values defined in e2fsprogs.
394 *
395 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
396 */
397static inline void ext4_check_flag_values(void)
398{
399 CHECK_FLAG_VALUE(SECRM);
400 CHECK_FLAG_VALUE(UNRM);
401 CHECK_FLAG_VALUE(COMPR);
402 CHECK_FLAG_VALUE(SYNC);
403 CHECK_FLAG_VALUE(IMMUTABLE);
404 CHECK_FLAG_VALUE(APPEND);
405 CHECK_FLAG_VALUE(NODUMP);
406 CHECK_FLAG_VALUE(NOATIME);
407 CHECK_FLAG_VALUE(DIRTY);
408 CHECK_FLAG_VALUE(COMPRBLK);
409 CHECK_FLAG_VALUE(NOCOMPR);
410 CHECK_FLAG_VALUE(ECOMPR);
411 CHECK_FLAG_VALUE(INDEX);
412 CHECK_FLAG_VALUE(IMAGIC);
413 CHECK_FLAG_VALUE(JOURNAL_DATA);
414 CHECK_FLAG_VALUE(NOTAIL);
415 CHECK_FLAG_VALUE(DIRSYNC);
416 CHECK_FLAG_VALUE(TOPDIR);
417 CHECK_FLAG_VALUE(HUGE_FILE);
418 CHECK_FLAG_VALUE(EXTENTS);
419 CHECK_FLAG_VALUE(EA_INODE);
420 CHECK_FLAG_VALUE(EOFBLOCKS);
421 CHECK_FLAG_VALUE(RESERVED);
422}
423
ac27a0ec 424/* Used to pass group descriptor data when online resize is done */
617ba13b 425struct ext4_new_group_input {
8c55e204
DK
426 __u32 group; /* Group number for this data */
427 __u64 block_bitmap; /* Absolute block number of block bitmap */
428 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
429 __u64 inode_table; /* Absolute block number of inode table start */
430 __u32 blocks_count; /* Total number of blocks in this group */
431 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
432 __u16 unused;
433};
434
617ba13b
MC
435/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
436struct ext4_new_group_data {
ac27a0ec 437 __u32 group;
bd81d8ee
LV
438 __u64 block_bitmap;
439 __u64 inode_bitmap;
440 __u64 inode_table;
ac27a0ec
DK
441 __u32 blocks_count;
442 __u16 reserved_blocks;
443 __u16 unused;
444 __u32 free_blocks_count;
445};
446
a2df2a63 447/*
c2177057 448 * Flags used by ext4_get_blocks()
a2df2a63 449 */
c2177057
TT
450 /* Allocate any needed blocks and/or convert an unitialized
451 extent to be an initialized ext4 */
2ac3b6e0 452#define EXT4_GET_BLOCKS_CREATE 0x0001
c2177057 453 /* Request the creation of an unitialized extent */
2ac3b6e0 454#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
c2177057
TT
455#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
456 EXT4_GET_BLOCKS_CREATE)
c2177057 457 /* Caller is from the delayed allocation writeout path,
2ac3b6e0
TT
458 so set the magic i_delalloc_reserve_flag after taking the
459 inode allocation semaphore for */
03f5d8bc 460#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b
MC
461 /* caller is from the direct IO path, request to creation of an
462 unitialized extents if not allocated, split the uninitialized
463 extent if blocks has been preallocated already*/
c7064ef1 464#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 465#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 466#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
0031462b 467 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
c7064ef1
JZ
468 /* Convert extent to initialized after IO complete */
469#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
0031462b 470 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
ac27a0ec 471
e6362609
TT
472/*
473 * Flags used by ext4_free_blocks
474 */
475#define EXT4_FREE_BLOCKS_METADATA 0x0001
476#define EXT4_FREE_BLOCKS_FORGET 0x0002
1f2acb60 477#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
e6362609 478
ac27a0ec
DK
479/*
480 * ioctl commands
481 */
617ba13b
MC
482#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
483#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
484#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
485#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
486#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
487#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
e23291b9 488#ifdef CONFIG_JBD2_DEBUG
617ba13b 489#define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
ac27a0ec 490#endif
617ba13b
MC
491#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
492#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
493#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
494#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
495#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 496 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 497 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 498#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 499#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
ac27a0ec
DK
500
501/*
502 * ioctl commands in 32 bit emulation
503 */
617ba13b
MC
504#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
505#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
506#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
507#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
508#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
509#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
510#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
e23291b9 511#ifdef CONFIG_JBD2_DEBUG
617ba13b 512#define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
ac27a0ec 513#endif
617ba13b
MC
514#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
515#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
ac27a0ec
DK
516
517
518/*
519 * Mount options
520 */
617ba13b 521struct ext4_mount_options {
ac27a0ec
DK
522 unsigned long s_mount_opt;
523 uid_t s_resuid;
524 gid_t s_resgid;
525 unsigned long s_commit_interval;
30773840 526 u32 s_min_batch_time, s_max_batch_time;
ac27a0ec
DK
527#ifdef CONFIG_QUOTA
528 int s_jquota_fmt;
529 char *s_qf_names[MAXQUOTAS];
530#endif
531};
532
fb0a387d
ES
533/* Max physical block we can addres w/o extents */
534#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
535
ac27a0ec
DK
536/*
537 * Structure of an inode on the disk
538 */
617ba13b 539struct ext4_inode {
ac27a0ec
DK
540 __le16 i_mode; /* File mode */
541 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 542 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 543 __le32 i_atime; /* Access time */
ef7f3835 544 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
545 __le32 i_mtime; /* Modification time */
546 __le32 i_dtime; /* Deletion Time */
547 __le16 i_gid; /* Low 16 bits of Group Id */
548 __le16 i_links_count; /* Links count */
0fc1b451 549 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
550 __le32 i_flags; /* File flags */
551 union {
552 struct {
25ec56b5 553 __le32 l_i_version;
ac27a0ec
DK
554 } linux1;
555 struct {
556 __u32 h_i_translator;
557 } hurd1;
558 struct {
559 __u32 m_i_reserved1;
560 } masix1;
561 } osd1; /* OS dependent 1 */
617ba13b 562 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 563 __le32 i_generation; /* File version (for NFS) */
7973c0c1 564 __le32 i_file_acl_lo; /* File ACL */
a48380f7 565 __le32 i_size_high;
f077d0d7 566 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
567 union {
568 struct {
0fc1b451 569 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 570 __le16 l_i_file_acl_high;
8c55e204 571 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec
DK
572 __le16 l_i_gid_high; /* were reserved2[0] */
573 __u32 l_i_reserved2;
574 } linux2;
575 struct {
f077d0d7 576 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
577 __u16 h_i_mode_high;
578 __u16 h_i_uid_high;
579 __u16 h_i_gid_high;
580 __u32 h_i_author;
581 } hurd2;
582 struct {
f077d0d7 583 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 584 __le16 m_i_file_acl_high;
ac27a0ec
DK
585 __u32 m_i_reserved2[2];
586 } masix2;
587 } osd2; /* OS dependent 2 */
588 __le16 i_extra_isize;
589 __le16 i_pad1;
ef7f3835
KS
590 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
591 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
592 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
593 __le32 i_crtime; /* File Creation time */
594 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 595 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
596};
597
748de673
AF
598struct move_extent {
599 __u32 reserved; /* should be zero */
600 __u32 donor_fd; /* donor file descriptor */
601 __u64 orig_start; /* logical start offset in block for orig */
602 __u64 donor_start; /* logical start offset in block for donor */
603 __u64 len; /* block length to be moved */
604 __u64 moved_len; /* moved block length */
605};
ac27a0ec 606
ef7f3835
KS
607#define EXT4_EPOCH_BITS 2
608#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
609#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
610
611/*
612 * Extended fields will fit into an inode if the filesystem was formatted
613 * with large inodes (-I 256 or larger) and there are not currently any EAs
614 * consuming all of the available space. For new inodes we always reserve
615 * enough space for the kernel's known extended fields, but for inodes
616 * created with an old kernel this might not have been the case. None of
617 * the extended inode fields is critical for correct filesystem operation.
618 * This macro checks if a certain field fits in the inode. Note that
619 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
620 */
621#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
622 ((offsetof(typeof(*ext4_inode), field) + \
623 sizeof((ext4_inode)->field)) \
624 <= (EXT4_GOOD_OLD_INODE_SIZE + \
625 (einode)->i_extra_isize)) \
626
627static inline __le32 ext4_encode_extra_time(struct timespec *time)
628{
629 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
630 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
631 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
632}
633
634static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
635{
636 if (sizeof(time->tv_sec) > 4)
637 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
638 << 32;
c1fccc06 639 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
640}
641
642#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
643do { \
644 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
645 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
646 (raw_inode)->xtime ## _extra = \
647 ext4_encode_extra_time(&(inode)->xtime); \
648} while (0)
649
650#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
651do { \
652 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
653 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
654 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
655 (raw_inode)->xtime ## _extra = \
656 ext4_encode_extra_time(&(einode)->xtime); \
657} while (0)
658
659#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
660do { \
661 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
662 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
663 ext4_decode_extra_time(&(inode)->xtime, \
664 raw_inode->xtime ## _extra); \
665} while (0)
666
667#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
668do { \
669 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
670 (einode)->xtime.tv_sec = \
671 (signed)le32_to_cpu((raw_inode)->xtime); \
672 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
673 ext4_decode_extra_time(&(einode)->xtime, \
674 raw_inode->xtime ## _extra); \
675} while (0)
676
25ec56b5
JNC
677#define i_disk_version osd1.linux1.l_i_version
678
ac27a0ec
DK
679#if defined(__KERNEL__) || defined(__linux__)
680#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 681#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 682#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
683#define i_uid_low i_uid
684#define i_gid_low i_gid
685#define i_uid_high osd2.linux2.l_i_uid_high
686#define i_gid_high osd2.linux2.l_i_gid_high
687#define i_reserved2 osd2.linux2.l_i_reserved2
688
689#elif defined(__GNU__)
690
691#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
692#define i_uid_high osd2.hurd2.h_i_uid_high
693#define i_gid_high osd2.hurd2.h_i_gid_high
694#define i_author osd2.hurd2.h_i_author
695
696#elif defined(__masix__)
697
698#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 699#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
700#define i_reserved2 osd2.masix2.m_i_reserved2
701
702#endif /* defined(__KERNEL__) || defined(__linux__) */
703
d444c3c3
TT
704/*
705 * storage for cached extent
706 */
707struct ext4_ext_cache {
708 ext4_fsblk_t ec_start;
709 ext4_lblk_t ec_block;
710 __u32 ec_len; /* must be 32bit to return holes */
711 __u32 ec_type;
712};
713
714/*
715 * fourth extended file system inode data in memory
716 */
717struct ext4_inode_info {
718 __le32 i_data[15]; /* unconverted */
d444c3c3 719 __u32 i_dtime;
12e9b892 720 ext4_fsblk_t i_file_acl;
d444c3c3
TT
721
722 /*
723 * i_block_group is the number of the block group which contains
724 * this file's inode. Constant across the lifetime of the inode,
725 * it is ued for making block allocation decisions - we try to
726 * place a file's data blocks near its inode block, and new inodes
727 * near to their parent directory's inode.
728 */
729 ext4_group_t i_block_group;
19f5fb7a 730 unsigned long i_state_flags; /* Dynamic state flags */
12e9b892 731 unsigned long i_flags;
d444c3c3
TT
732
733 ext4_lblk_t i_dir_start_lookup;
734#ifdef CONFIG_EXT4_FS_XATTR
735 /*
736 * Extended attributes can be read independently of the main file
737 * data. Taking i_mutex even when reading would cause contention
738 * between readers of EAs and writers of regular file data, so
739 * instead we synchronize on xattr_sem when reading or changing
740 * EAs.
741 */
742 struct rw_semaphore xattr_sem;
743#endif
d444c3c3
TT
744
745 struct list_head i_orphan; /* unlinked but open inodes */
746
747 /*
748 * i_disksize keeps track of what the inode size is ON DISK, not
749 * in memory. During truncate, i_size is set to the new size by
750 * the VFS prior to calling ext4_truncate(), but the filesystem won't
751 * set i_disksize to 0 until the truncate is actually under way.
752 *
753 * The intent is that i_disksize always represents the blocks which
754 * are used by this file. This allows recovery to restart truncate
755 * on orphans if we crash during truncate. We actually write i_disksize
756 * into the on-disk inode when writing inodes out, instead of i_size.
757 *
758 * The only time when i_disksize and i_size may be different is when
759 * a truncate is in progress. The only things which change i_disksize
760 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
761 */
762 loff_t i_disksize;
763
764 /*
765 * i_data_sem is for serialising ext4_truncate() against
766 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
767 * data tree are chopped off during truncate. We can't do that in
768 * ext4 because whenever we perform intermediate commits during
769 * truncate, the inode and all the metadata blocks *must* be in a
770 * consistent state which allows truncation of the orphans to restart
771 * during recovery. Hence we must fix the get_block-vs-truncate race
772 * by other means, so we have i_data_sem.
773 */
774 struct rw_semaphore i_data_sem;
775 struct inode vfs_inode;
776 struct jbd2_inode jinode;
777
778 struct ext4_ext_cache i_cached_extent;
779 /*
780 * File creation time. Its function is same as that of
781 * struct timespec i_{a,c,m}time in the generic inode.
782 */
783 struct timespec i_crtime;
784
785 /* mballoc */
786 struct list_head i_prealloc_list;
787 spinlock_t i_prealloc_lock;
788
789 /* ialloc */
790 ext4_group_t i_last_alloc_group;
791
792 /* allocation reservation info for delalloc */
793 unsigned int i_reserved_data_blocks;
794 unsigned int i_reserved_meta_blocks;
795 unsigned int i_allocated_meta_blocks;
796 unsigned short i_delalloc_reserved_flag;
9d0be502
TT
797 sector_t i_da_metadata_calc_last_lblock;
798 int i_da_metadata_calc_len;
d444c3c3
TT
799
800 /* on-disk additional length */
801 __u16 i_extra_isize;
802
803 spinlock_t i_block_reservation_lock;
a9e7f447
DM
804#ifdef CONFIG_QUOTA
805 /* quota space reservation, managed internally by quota code */
806 qsize_t i_reserved_quota;
807#endif
8d5d02e6 808
c7064ef1
JZ
809 /* completed IOs that might need unwritten extents handling */
810 struct list_head i_completed_io_list;
744692dc 811 spinlock_t i_completed_io_lock;
8d5d02e6
MC
812 /* current io_end structure for async DIO write*/
813 ext4_io_end_t *cur_aio_dio;
b436b9be
JK
814
815 /*
816 * Transactions that contain inode's metadata needed to complete
817 * fsync and fdatasync, respectively.
818 */
819 tid_t i_sync_tid;
820 tid_t i_datasync_tid;
d444c3c3
TT
821};
822
ac27a0ec
DK
823/*
824 * File system states
825 */
617ba13b
MC
826#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
827#define EXT4_ERROR_FS 0x0002 /* Errors detected */
828#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 829
469108ff
TT
830/*
831 * Misc. filesystem flags
832 */
833#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
834#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
835#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
836
ac27a0ec
DK
837/*
838 * Mount flags
839 */
617ba13b
MC
840#define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
841#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
842#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
843#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
844#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
845#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
846#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
847#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
848#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
849#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
850#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
851#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
852#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
853#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
854#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
855#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 856#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b
MC
857#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
858#define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
859#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
860#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
861#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
744692dc 862#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 863#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 864#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
25ec56b5 865#define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
64769240 866#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 867#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 868#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 869#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
5bf5683a 870
617ba13b
MC
871#define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
872#define set_opt(o, opt) o |= EXT4_MOUNT_##opt
873#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
874 EXT4_MOUNT_##opt)
ac27a0ec 875
617ba13b
MC
876#define ext4_set_bit ext2_set_bit
877#define ext4_set_bit_atomic ext2_set_bit_atomic
878#define ext4_clear_bit ext2_clear_bit
879#define ext4_clear_bit_atomic ext2_clear_bit_atomic
880#define ext4_test_bit ext2_test_bit
881#define ext4_find_first_zero_bit ext2_find_first_zero_bit
882#define ext4_find_next_zero_bit ext2_find_next_zero_bit
aa02ad67 883#define ext4_find_next_bit ext2_find_next_bit
ac27a0ec
DK
884
885/*
886 * Maximal mount counts between two filesystem checks
887 */
617ba13b
MC
888#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
889#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
890
891/*
892 * Behaviour when detecting errors
893 */
617ba13b
MC
894#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
895#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
896#define EXT4_ERRORS_PANIC 3 /* Panic */
897#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec
DK
898
899/*
900 * Structure of the super block
901 */
617ba13b 902struct ext4_super_block {
ac27a0ec 903/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 904 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
905 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
906 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
907/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
908 __le32 s_first_data_block; /* First Data Block */
909 __le32 s_log_block_size; /* Block size */
f077d0d7 910 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
ac27a0ec 911/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
f077d0d7 912 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
ac27a0ec
DK
913 __le32 s_inodes_per_group; /* # Inodes per group */
914 __le32 s_mtime; /* Mount time */
915/*30*/ __le32 s_wtime; /* Write time */
916 __le16 s_mnt_count; /* Mount count */
917 __le16 s_max_mnt_count; /* Maximal mount count */
918 __le16 s_magic; /* Magic signature */
919 __le16 s_state; /* File system state */
920 __le16 s_errors; /* Behaviour when detecting errors */
921 __le16 s_minor_rev_level; /* minor revision level */
922/*40*/ __le32 s_lastcheck; /* time of last check */
923 __le32 s_checkinterval; /* max. time between checks */
924 __le32 s_creator_os; /* OS */
925 __le32 s_rev_level; /* Revision level */
926/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
927 __le16 s_def_resgid; /* Default gid for reserved blocks */
928 /*
617ba13b 929 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
930 *
931 * Note: the difference between the compatible feature set and
932 * the incompatible feature set is that if there is a bit set
933 * in the incompatible feature set that the kernel doesn't
934 * know about, it should refuse to mount the filesystem.
935 *
936 * e2fsck's requirements are more strict; if it doesn't know
937 * about a feature in either the compatible or incompatible
938 * feature set, it must abort and not try to meddle with
939 * things it doesn't understand...
940 */
941 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 942 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
943 __le16 s_block_group_nr; /* block group # of this superblock */
944 __le32 s_feature_compat; /* compatible feature set */
945/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
946 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
947/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
948/*78*/ char s_volume_name[16]; /* volume name */
949/*88*/ char s_last_mounted[64]; /* directory where last mounted */
950/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
951 /*
952 * Performance hints. Directory preallocation should only
617ba13b 953 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
954 */
955 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
956 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
957 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
958 /*
617ba13b 959 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
960 */
961/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
962/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
963 __le32 s_journal_dev; /* device number of journal file */
964 __le32 s_last_orphan; /* start of list of inodes to delete */
965 __le32 s_hash_seed[4]; /* HTREE hash seed */
966 __u8 s_def_hash_version; /* Default hash version to use */
967 __u8 s_reserved_char_pad;
0d1ee42f 968 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 969/*100*/ __le32 s_default_mount_opts;
ac27a0ec 970 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
971 __le32 s_mkfs_time; /* When the filesystem was created */
972 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
973 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
974/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
975 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
976 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
977 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
978 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
979 __le32 s_flags; /* Miscellaneous flags */
980 __le16 s_raid_stride; /* RAID stride */
981 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */
982 __le64 s_mmp_block; /* Block for multi-mount protection */
983 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8
JS
984 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
985 __u8 s_reserved_char_pad2;
986 __le16 s_reserved_pad;
afc32f7e
TT
987 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
988 __u32 s_reserved[160]; /* Padding to the end of the block */
ac27a0ec
DK
989};
990
991#ifdef __KERNEL__
bc0b0d6d
TT
992
993/*
4ab2f15b 994 * run-time mount flags
bc0b0d6d
TT
995 */
996#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 997#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 998
ca0faba0
TT
999/*
1000 * fourth extended-fs super-block data in memory
1001 */
1002struct ext4_sb_info {
1003 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1004 unsigned long s_inodes_per_block;/* Number of inodes per block */
1005 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1006 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1007 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1008 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1009 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1010 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1011 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
ca0faba0
TT
1012 unsigned long s_overhead_last; /* Last calculated overhead */
1013 unsigned long s_blocks_last; /* Last seen block count */
1014 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1015 struct buffer_head * s_sbh; /* Buffer containing the super block */
1016 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1017 struct buffer_head **s_group_desc;
7f4520cc 1018 unsigned int s_mount_opt;
bc0b0d6d 1019 unsigned int s_mount_flags;
ca0faba0
TT
1020 ext4_fsblk_t s_sb_block;
1021 uid_t s_resuid;
1022 gid_t s_resgid;
1023 unsigned short s_mount_state;
1024 unsigned short s_pad;
1025 int s_addr_per_block_bits;
1026 int s_desc_per_block_bits;
1027 int s_inode_size;
1028 int s_first_ino;
1029 unsigned int s_inode_readahead_blks;
11013911 1030 unsigned int s_inode_goal;
ca0faba0
TT
1031 spinlock_t s_next_gen_lock;
1032 u32 s_next_generation;
1033 u32 s_hash_seed[4];
1034 int s_def_hash_version;
1035 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1036 struct percpu_counter s_freeblocks_counter;
1037 struct percpu_counter s_freeinodes_counter;
1038 struct percpu_counter s_dirs_counter;
1039 struct percpu_counter s_dirtyblocks_counter;
1040 struct blockgroup_lock *s_blockgroup_lock;
1041 struct proc_dir_entry *s_proc;
1042 struct kobject s_kobj;
1043 struct completion s_kobj_unregister;
1044
1045 /* Journaling */
1046 struct inode *s_journal_inode;
1047 struct journal_s *s_journal;
1048 struct list_head s_orphan;
1049 struct mutex s_orphan_lock;
1050 struct mutex s_resize_lock;
1051 unsigned long s_commit_interval;
1052 u32 s_max_batch_time;
1053 u32 s_min_batch_time;
1054 struct block_device *journal_bdev;
1055#ifdef CONFIG_JBD2_DEBUG
1056 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
1057 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
1058#endif
1059#ifdef CONFIG_QUOTA
1060 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1061 int s_jquota_fmt; /* Format of quota to use */
1062#endif
1063 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 1064 struct rb_root system_blks;
ca0faba0
TT
1065
1066#ifdef EXTENTS_STATS
1067 /* ext4 extents stats */
1068 unsigned long s_ext_min;
1069 unsigned long s_ext_max;
1070 unsigned long s_depth_max;
1071 spinlock_t s_ext_stats_lock;
1072 unsigned long s_ext_blocks;
1073 unsigned long s_ext_extents;
1074#endif
1075
1076 /* for buddy allocator */
1077 struct ext4_group_info ***s_group_info;
1078 struct inode *s_buddy_cache;
1079 long s_blocks_reserved;
1080 spinlock_t s_reserve_lock;
1081 spinlock_t s_md_lock;
1082 tid_t s_last_transaction;
1083 unsigned short *s_mb_offsets;
1084 unsigned int *s_mb_maxs;
1085
1086 /* tunables */
1087 unsigned long s_stripe;
1088 unsigned int s_mb_stream_request;
1089 unsigned int s_mb_max_to_scan;
1090 unsigned int s_mb_min_to_scan;
1091 unsigned int s_mb_stats;
1092 unsigned int s_mb_order2_reqs;
1093 unsigned int s_mb_group_prealloc;
55138e0b 1094 unsigned int s_max_writeback_mb_bump;
ca0faba0
TT
1095 /* where last allocation was done - for stream allocation */
1096 unsigned long s_mb_last_group;
1097 unsigned long s_mb_last_start;
1098
ca0faba0
TT
1099 /* stats for buddy allocator */
1100 spinlock_t s_mb_pa_lock;
1101 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1102 atomic_t s_bal_success; /* we found long enough chunks */
1103 atomic_t s_bal_allocated; /* in blocks */
1104 atomic_t s_bal_ex_scanned; /* total extents scanned */
1105 atomic_t s_bal_goals; /* goal hits */
1106 atomic_t s_bal_breaks; /* too long searches */
1107 atomic_t s_bal_2orders; /* 2^order hits */
1108 spinlock_t s_bal_lock;
1109 unsigned long s_mb_buddies_generated;
1110 unsigned long long s_mb_generation_time;
1111 atomic_t s_mb_lost_chunks;
1112 atomic_t s_mb_preallocated;
1113 atomic_t s_mb_discarded;
50797481 1114 atomic_t s_lock_busy;
ca0faba0
TT
1115
1116 /* locality groups */
003cb608 1117 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1118
1119 /* for write statistics */
1120 unsigned long s_sectors_written_start;
1121 u64 s_kbytes_written;
1122
1123 unsigned int s_log_groups_per_flex;
1124 struct flex_groups *s_flex_groups;
4c0425ff
MC
1125
1126 /* workqueue for dio unwritten */
1127 struct workqueue_struct *dio_unwritten_wq;
ca0faba0
TT
1128};
1129
af5bc92d 1130static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1131{
1132 return sb->s_fs_info;
1133}
617ba13b 1134static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1135{
617ba13b 1136 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1137}
1138
ef7f3835
KS
1139static inline struct timespec ext4_current_time(struct inode *inode)
1140{
1141 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1142 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1143}
1144
617ba13b 1145static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1146{
617ba13b
MC
1147 return ino == EXT4_ROOT_INO ||
1148 ino == EXT4_JOURNAL_INO ||
1149 ino == EXT4_RESIZE_INO ||
1150 (ino >= EXT4_FIRST_INO(sb) &&
1151 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1152}
19f5fb7a
TT
1153
1154/*
1155 * Inode dynamic state flags
1156 */
1157enum {
1158 EXT4_STATE_JDATA, /* journaled data exists */
1159 EXT4_STATE_NEW, /* inode is newly created */
1160 EXT4_STATE_XATTR, /* has in-inode xattrs */
1161 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1162 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1163 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1164 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1165};
1166
12e9b892
DM
1167#define EXT4_INODE_BIT_FNS(name, field) \
1168static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1169{ \
1170 return test_bit(bit, &EXT4_I(inode)->i_##field); \
1171} \
1172static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1173{ \
1174 set_bit(bit, &EXT4_I(inode)->i_##field); \
1175} \
1176static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1177{ \
1178 clear_bit(bit, &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1179}
1180
12e9b892
DM
1181EXT4_INODE_BIT_FNS(flag, flags)
1182EXT4_INODE_BIT_FNS(state, state_flags)
ac27a0ec 1183#else
617ba13b 1184/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1185 * a kernel struct super_block. This will allow us to call the feature-test
1186 * macros from user land. */
617ba13b 1187#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1188#endif
1189
617ba13b 1190#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1191
1192/*
1193 * Codes for operating systems
1194 */
617ba13b
MC
1195#define EXT4_OS_LINUX 0
1196#define EXT4_OS_HURD 1
1197#define EXT4_OS_MASIX 2
1198#define EXT4_OS_FREEBSD 3
1199#define EXT4_OS_LITES 4
ac27a0ec
DK
1200
1201/*
1202 * Revision levels
1203 */
617ba13b
MC
1204#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1205#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1206
617ba13b
MC
1207#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1208#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1209
617ba13b 1210#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1211
1212/*
1213 * Feature set definitions
1214 */
1215
617ba13b 1216#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1217 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1218#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1219 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1220#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1221 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1222#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1223 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1224#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1225 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1226#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1227 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1228#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1229 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1230#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1231 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1232#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1233 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1234
1235#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1236#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1237#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1238#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1239#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1240#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1241
1242#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1243#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1244#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1245#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1246#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1247#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1248#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
617ba13b
MC
1249
1250#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1251#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1252#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1253#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1254#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1255#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1256#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1257#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1258#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
1259#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1260#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
617ba13b
MC
1261
1262#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1263#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1264 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1265 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1266 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581
JS
1267 EXT4_FEATURE_INCOMPAT_64BIT| \
1268 EXT4_FEATURE_INCOMPAT_FLEX_BG)
617ba13b
MC
1269#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1270 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1271 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1272 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1273 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451
AK
1274 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1275 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
ac27a0ec
DK
1276
1277/*
1278 * Default values for user and/or group using reserved blocks
1279 */
617ba13b
MC
1280#define EXT4_DEF_RESUID 0
1281#define EXT4_DEF_RESGID 0
ac27a0ec 1282
240799cd
TT
1283#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1284
ac27a0ec
DK
1285/*
1286 * Default mount options
1287 */
617ba13b
MC
1288#define EXT4_DEFM_DEBUG 0x0001
1289#define EXT4_DEFM_BSDGROUPS 0x0002
1290#define EXT4_DEFM_XATTR_USER 0x0004
1291#define EXT4_DEFM_ACL 0x0008
1292#define EXT4_DEFM_UID16 0x0010
1293#define EXT4_DEFM_JMODE 0x0060
1294#define EXT4_DEFM_JMODE_DATA 0x0020
1295#define EXT4_DEFM_JMODE_ORDERED 0x0040
1296#define EXT4_DEFM_JMODE_WBACK 0x0060
ac27a0ec 1297
30773840
TT
1298/*
1299 * Default journal batch times
1300 */
1301#define EXT4_DEF_MIN_BATCH_TIME 0
1302#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1303
a4912123
TT
1304/*
1305 * Minimum number of groups in a flexgroup before we separate out
1306 * directories into the first block group of a flexgroup
1307 */
1308#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1309
ac27a0ec
DK
1310/*
1311 * Structure of a directory entry
1312 */
617ba13b 1313#define EXT4_NAME_LEN 255
ac27a0ec 1314
617ba13b 1315struct ext4_dir_entry {
ac27a0ec
DK
1316 __le32 inode; /* Inode number */
1317 __le16 rec_len; /* Directory entry length */
1318 __le16 name_len; /* Name length */
617ba13b 1319 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1320};
1321
1322/*
617ba13b 1323 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1324 * stored in intel byte order, and the name_len field could never be
1325 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1326 * file_type field.
1327 */
617ba13b 1328struct ext4_dir_entry_2 {
ac27a0ec
DK
1329 __le32 inode; /* Inode number */
1330 __le16 rec_len; /* Directory entry length */
1331 __u8 name_len; /* Name length */
1332 __u8 file_type;
617ba13b 1333 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1334};
1335
1336/*
617ba13b 1337 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1338 * other bits are reserved for now.
1339 */
617ba13b
MC
1340#define EXT4_FT_UNKNOWN 0
1341#define EXT4_FT_REG_FILE 1
1342#define EXT4_FT_DIR 2
1343#define EXT4_FT_CHRDEV 3
1344#define EXT4_FT_BLKDEV 4
1345#define EXT4_FT_FIFO 5
1346#define EXT4_FT_SOCK 6
1347#define EXT4_FT_SYMLINK 7
ac27a0ec 1348
617ba13b 1349#define EXT4_FT_MAX 8
ac27a0ec
DK
1350
1351/*
617ba13b 1352 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1353 *
1354 * NOTE: It must be a multiple of 4
1355 */
617ba13b
MC
1356#define EXT4_DIR_PAD 4
1357#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1358#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1359 ~EXT4_DIR_ROUND)
a72d7f83
JK
1360#define EXT4_MAX_REC_LEN ((1<<16)-1)
1361
ac27a0ec
DK
1362/*
1363 * Hash Tree Directory indexing
1364 * (c) Daniel Phillips, 2001
1365 */
1366
4074fe37
ES
1367#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1368 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
12e9b892 1369 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
617ba13b
MC
1370#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1371#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1372
1373/* Legal values for the dx_root hash_version field: */
1374
1375#define DX_HASH_LEGACY 0
1376#define DX_HASH_HALF_MD4 1
1377#define DX_HASH_TEA 2
f99b2589
TT
1378#define DX_HASH_LEGACY_UNSIGNED 3
1379#define DX_HASH_HALF_MD4_UNSIGNED 4
1380#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec
DK
1381
1382#ifdef __KERNEL__
1383
1384/* hash info structure used by the directory hash */
1385struct dx_hash_info
1386{
1387 u32 hash;
1388 u32 minor_hash;
1389 int hash_version;
1390 u32 *seed;
1391};
1392
617ba13b 1393#define EXT4_HTREE_EOF 0x7fffffff
ac27a0ec
DK
1394
1395/*
617ba13b 1396 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1397 */
1398#define HASH_NB_ALWAYS 1
1399
1400
1401/*
1402 * Describe an inode's exact location on disk and in memory
1403 */
617ba13b 1404struct ext4_iloc
ac27a0ec
DK
1405{
1406 struct buffer_head *bh;
1407 unsigned long offset;
fd2d4291 1408 ext4_group_t block_group;
ac27a0ec
DK
1409};
1410
617ba13b 1411static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1412{
617ba13b 1413 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1414}
1415
1416/*
1417 * This structure is stuffed into the struct file's private_data field
1418 * for directories. It is where we put information so that we can do
1419 * readdir operations in hash tree order.
1420 */
1421struct dir_private_info {
1422 struct rb_root root;
1423 struct rb_node *curr_node;
1424 struct fname *extra_fname;
1425 loff_t last_pos;
1426 __u32 curr_hash;
1427 __u32 curr_minor_hash;
1428 __u32 next_hash;
1429};
1430
1431/* calculate the first block number of the group */
617ba13b 1432static inline ext4_fsblk_t
fd2d4291 1433ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1434{
617ba13b
MC
1435 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1436 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1437}
1438
1439/*
1440 * Special error return code only used by dx_probe() and its callers.
1441 */
1442#define ERR_BAD_DX_DIR -75000
1443
72b64b59 1444void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
a9df9a49 1445 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
3a5b2ecd 1446
9f6200bb
TT
1447extern struct proc_dir_entry *ext4_proc_root;
1448
ac27a0ec
DK
1449/*
1450 * Function prototypes
1451 */
1452
1453/*
1454 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1455 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1456 */
8c55e204
DK
1457# define NORET_TYPE /**/
1458# define ATTRIB_NORET __attribute__((noreturn))
1459# define NORET_AND noreturn,
ac27a0ec 1460
498e5f24
TT
1461/* bitmap.c */
1462extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1463
ac27a0ec 1464/* balloc.c */
3a5b2ecd
MC
1465extern unsigned int ext4_block_group(struct super_block *sb,
1466 ext4_fsblk_t blocknr);
1467extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1468 ext4_fsblk_t blocknr);
fd2d4291
AM
1469extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1470extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1471 ext4_group_t group);
7061eba7 1472extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
617ba13b 1473 ext4_fsblk_t goal, unsigned long *count, int *errp);
5c791616 1474extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
8c3bf8a0 1475extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
e21675d4
AK
1476extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1477 ext4_fsblk_t block, unsigned long count);
af5bc92d
TT
1478extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1479extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1480extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1481 ext4_group_t block_group,
ac27a0ec 1482 struct buffer_head ** bh);
617ba13b 1483extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
bb23c20a
TT
1484struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1485 ext4_group_t block_group);
1486extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1487 struct buffer_head *bh,
1488 ext4_group_t group,
1489 struct ext4_group_desc *desc);
1490#define ext4_free_blocks_after_init(sb, group, desc) \
1491 ext4_init_block_bitmap(sb, NULL, group, desc)
ac27a0ec
DK
1492
1493/* dir.c */
617ba13b
MC
1494extern int ext4_check_dir_entry(const char *, struct inode *,
1495 struct ext4_dir_entry_2 *,
498e5f24 1496 struct buffer_head *, unsigned int);
617ba13b 1497extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 1498 __u32 minor_hash,
617ba13b
MC
1499 struct ext4_dir_entry_2 *dirent);
1500extern void ext4_htree_free_dir_info(struct dir_private_info *p);
ac27a0ec
DK
1501
1502/* fsync.c */
af5bc92d 1503extern int ext4_sync_file(struct file *, struct dentry *, int);
ac27a0ec
DK
1504
1505/* hash.c */
617ba13b 1506extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
1507 dx_hash_info *hinfo);
1508
1509/* ialloc.c */
f157a4aa 1510extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
11013911 1511 const struct qstr *qstr, __u32 goal);
af5bc92d
TT
1512extern void ext4_free_inode(handle_t *, struct inode *);
1513extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1514extern unsigned long ext4_count_free_inodes(struct super_block *);
1515extern unsigned long ext4_count_dirs(struct super_block *);
1516extern void ext4_check_inodes_bitmap(struct super_block *);
bb23c20a
TT
1517extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1518 struct buffer_head *bh,
1519 ext4_group_t group,
1520 struct ext4_group_desc *desc);
1521extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
ac27a0ec 1522
c9de560d
AT
1523/* mballoc.c */
1524extern long ext4_mb_stats;
1525extern long ext4_mb_max_to_scan;
1526extern int ext4_mb_init(struct super_block *, int);
1527extern int ext4_mb_release(struct super_block *);
1528extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1529 struct ext4_allocation_request *, int *);
1530extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 1531extern void ext4_discard_preallocations(struct inode *);
c9de560d
AT
1532extern int __init init_ext4_mballoc(void);
1533extern void exit_ext4_mballoc(void);
44338711 1534extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
1535 struct buffer_head *bh, ext4_fsblk_t block,
1536 unsigned long count, int flags);
920313a7 1537extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 1538 ext4_group_t i, struct ext4_group_desc *desc);
920313a7
AK
1539extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1540extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1541 ext4_group_t, int);
ac27a0ec 1542/* inode.c */
725d26d3
AK
1543struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1544 ext4_lblk_t, int, int *);
1545struct buffer_head *ext4_bread(handle_t *, struct inode *,
1546 ext4_lblk_t, int, int *);
6873fa0d
ES
1547int ext4_get_block(struct inode *inode, sector_t iblock,
1548 struct buffer_head *bh_result, int create);
ac27a0ec 1549
1d1fe1ee 1550extern struct inode *ext4_iget(struct super_block *, unsigned long);
a9185b41 1551extern int ext4_write_inode(struct inode *, struct writeback_control *);
af5bc92d 1552extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
1553extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1554 struct kstat *stat);
af5bc92d
TT
1555extern void ext4_delete_inode(struct inode *);
1556extern int ext4_sync_inode(handle_t *, struct inode *);
617ba13b
MC
1557extern void ext4_dirty_inode(struct inode *);
1558extern int ext4_change_inode_journal_flag(struct inode *, int);
1559extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
91ef4caf 1560extern int ext4_can_truncate(struct inode *inode);
af5bc92d 1561extern void ext4_truncate(struct inode *);
487caeef 1562extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 1563extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 1564extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 1565extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 1566extern void ext4_set_aops(struct inode *inode);
a86c6181 1567extern int ext4_writepage_trans_blocks(struct inode *);
a02908f1 1568extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
f3bd1f3f 1569extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
cf108bca 1570extern int ext4_block_truncate_page(handle_t *handle,
a86c6181 1571 struct address_space *mapping, loff_t from);
c2ec175c 1572extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
a9e7f447 1573extern qsize_t *ext4_get_reserved_space(struct inode *inode);
c7064ef1 1574extern int flush_completed_IO(struct inode *inode);
5f634d06
AK
1575extern void ext4_da_update_reserve_space(struct inode *inode,
1576 int used, int quota_claim);
ac27a0ec 1577/* ioctl.c */
5cdd7b2d 1578extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 1579extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 1580
c14c6fd5 1581/* migrate.c */
2a43a878 1582extern int ext4_ext_migrate(struct inode *);
3d0518f4 1583
ac27a0ec 1584/* namei.c */
3d0518f4
WY
1585extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1586extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
617ba13b
MC
1587extern int ext4_orphan_add(handle_t *, struct inode *);
1588extern int ext4_orphan_del(handle_t *, struct inode *);
1589extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec
DK
1590 __u32 start_minor_hash, __u32 *next_hash);
1591
1592/* resize.c */
617ba13b
MC
1593extern int ext4_group_add(struct super_block *sb,
1594 struct ext4_new_group_data *input);
1595extern int ext4_group_extend(struct super_block *sb,
1596 struct ext4_super_block *es,
1597 ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
1598
1599/* super.c */
12062ddd 1600extern void __ext4_error(struct super_block *, const char *, const char *, ...)
ac27a0ec 1601 __attribute__ ((format (printf, 3, 4)));
12062ddd 1602#define ext4_error(sb, message...) __ext4_error(sb, __func__, ## message)
273df556
FM
1603extern void ext4_error_inode(const char *, struct inode *, const char *, ...)
1604 __attribute__ ((format (printf, 3, 4)));
1605extern void ext4_error_file(const char *, struct file *, const char *, ...)
ac27a0ec 1606 __attribute__ ((format (printf, 3, 4)));
af5bc92d
TT
1607extern void __ext4_std_error(struct super_block *, const char *, int);
1608extern void ext4_abort(struct super_block *, const char *, const char *, ...)
ac27a0ec 1609 __attribute__ ((format (printf, 3, 4)));
12062ddd
ES
1610extern void __ext4_warning(struct super_block *, const char *,
1611 const char *, ...)
ac27a0ec 1612 __attribute__ ((format (printf, 3, 4)));
12062ddd 1613#define ext4_warning(sb, message...) __ext4_warning(sb, __func__, ## message)
b31e1552
ES
1614extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1615 __attribute__ ((format (printf, 3, 4)));
5d1b1b3f
AK
1616extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1617 const char *, const char *, ...)
1618 __attribute__ ((format (printf, 4, 5)));
af5bc92d 1619extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
1620extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1621 __u32 compat);
1622extern int ext4_update_rocompat_feature(handle_t *handle,
1623 struct super_block *sb, __u32 rocompat);
1624extern int ext4_update_incompat_feature(handle_t *handle,
1625 struct super_block *sb, __u32 incompat);
8fadc143
AR
1626extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1627 struct ext4_group_desc *bg);
1628extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1629 struct ext4_group_desc *bg);
1630extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1631 struct ext4_group_desc *bg);
560671a0
AK
1632extern __u32 ext4_free_blks_count(struct super_block *sb,
1633 struct ext4_group_desc *bg);
1634extern __u32 ext4_free_inodes_count(struct super_block *sb,
1635 struct ext4_group_desc *bg);
1636extern __u32 ext4_used_dirs_count(struct super_block *sb,
1637 struct ext4_group_desc *bg);
1638extern __u32 ext4_itable_unused_count(struct super_block *sb,
1639 struct ext4_group_desc *bg);
8fadc143
AR
1640extern void ext4_block_bitmap_set(struct super_block *sb,
1641 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1642extern void ext4_inode_bitmap_set(struct super_block *sb,
1643 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1644extern void ext4_inode_table_set(struct super_block *sb,
1645 struct ext4_group_desc *bg, ext4_fsblk_t blk);
560671a0
AK
1646extern void ext4_free_blks_set(struct super_block *sb,
1647 struct ext4_group_desc *bg, __u32 count);
1648extern void ext4_free_inodes_set(struct super_block *sb,
1649 struct ext4_group_desc *bg, __u32 count);
1650extern void ext4_used_dirs_set(struct super_block *sb,
1651 struct ext4_group_desc *bg, __u32 count);
1652extern void ext4_itable_unused_set(struct super_block *sb,
1653 struct ext4_group_desc *bg, __u32 count);
bb23c20a
TT
1654extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1655 struct ext4_group_desc *gdp);
1656extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1657 struct ext4_group_desc *gdp);
bd81d8ee
LV
1658
1659static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1660{
1661 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 1662 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
1663}
1664
1665static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1666{
1667 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 1668 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
1669}
1670
1671static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1672{
1673 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 1674 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
1675}
1676
1677static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1678 ext4_fsblk_t blk)
1679{
6bc9feff 1680 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1681 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1682}
1683
1684static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1685 ext4_fsblk_t blk)
1686{
308ba3ec 1687 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1688 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1689}
1690
1691static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1692 ext4_fsblk_t blk)
1693{
308ba3ec 1694 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
1695 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1696}
1697
a48380f7
AK
1698static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1699{
06a279d6
TT
1700 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1701 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1702 le32_to_cpu(raw_inode->i_size_lo);
1703 else
1704 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 1705}
bd81d8ee 1706
a48380f7
AK
1707static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1708{
1709 raw_inode->i_size_lo = cpu_to_le32(i_size);
1710 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1711}
ac27a0ec 1712
c9de560d
AT
1713static inline
1714struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 1715 ext4_group_t group)
c9de560d
AT
1716{
1717 struct ext4_group_info ***grp_info;
1718 long indexv, indexh;
1719 grp_info = EXT4_SB(sb)->s_group_info;
1720 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1721 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1722 return grp_info[indexv][indexh];
1723}
1724
8df9675f
TT
1725/*
1726 * Reading s_groups_count requires using smp_rmb() afterwards. See
1727 * the locking protocol documented in the comments of ext4_group_add()
1728 * in resize.c
1729 */
1730static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1731{
1732 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
1733
1734 smp_rmb();
1735 return ngroups;
1736}
c9de560d 1737
772cb7c8
JS
1738static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1739 ext4_group_t block_group)
1740{
1741 return block_group >> sbi->s_log_groups_per_flex;
1742}
1743
1744static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1745{
1746 return 1 << sbi->s_log_groups_per_flex;
1747}
1748
617ba13b 1749#define ext4_std_error(sb, errno) \
ac27a0ec
DK
1750do { \
1751 if ((errno)) \
4db9c54a 1752 __ext4_std_error((sb), __func__, (errno)); \
ac27a0ec
DK
1753} while (0)
1754
a30d542a 1755#ifdef CONFIG_SMP
179f7ebf 1756/* Each CPU can accumulate percpu_counter_batch blocks in their local
a30d542a 1757 * counters. So we need to make sure we have free blocks more
179f7ebf 1758 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 1759 */
179f7ebf 1760#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a
AK
1761#else
1762#define EXT4_FREEBLOCKS_WATERMARK 0
1763#endif
1764
cf17fea6
AK
1765static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1766{
1767 /*
1768 * XXX: replace with spinlock if seen contended -bzzz
1769 */
1770 down_write(&EXT4_I(inode)->i_data_sem);
1771 if (newsize > EXT4_I(inode)->i_disksize)
1772 EXT4_I(inode)->i_disksize = newsize;
1773 up_write(&EXT4_I(inode)->i_data_sem);
1774 return ;
1775}
a30d542a 1776
5d1b1b3f
AK
1777struct ext4_group_info {
1778 unsigned long bb_state;
1779 struct rb_root bb_free_root;
a36b4498
ES
1780 ext4_grpblk_t bb_first_free; /* first free block */
1781 ext4_grpblk_t bb_free; /* total free blocks */
1782 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
8a57d9d6 1783 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
5d1b1b3f
AK
1784 struct list_head bb_prealloc_list;
1785#ifdef DOUBLE_CHECK
1786 void *bb_bitmap;
1787#endif
1788 struct rw_semaphore alloc_sem;
a36b4498
ES
1789 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
1790 * regions, index is order.
1791 * bb_counters[3] = 5 means
1792 * 5 free 8-block regions. */
5d1b1b3f
AK
1793};
1794
1795#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
5d1b1b3f
AK
1796
1797#define EXT4_MB_GRP_NEED_INIT(grp) \
1798 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1799
50797481
TT
1800#define EXT4_MAX_CONTENTION 8
1801#define EXT4_CONTENTION_THRESHOLD 2
1802
955ce5f5
AK
1803static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1804 ext4_group_t group)
5d1b1b3f 1805{
955ce5f5
AK
1806 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1807}
5d1b1b3f 1808
50797481
TT
1809/*
1810 * Returns true if the filesystem is busy enough that attempts to
1811 * access the block group locks has run into contention.
1812 */
1813static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1814{
1815 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1816}
1817
955ce5f5
AK
1818static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1819{
50797481
TT
1820 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1821 if (spin_trylock(lock))
1822 /*
1823 * We're able to grab the lock right away, so drop the
1824 * lock contention counter.
1825 */
1826 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1827 else {
1828 /*
1829 * The lock is busy, so bump the contention counter,
1830 * and then wait on the spin lock.
1831 */
1832 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1833 EXT4_MAX_CONTENTION);
1834 spin_lock(lock);
1835 }
5d1b1b3f
AK
1836}
1837
1838static inline void ext4_unlock_group(struct super_block *sb,
1839 ext4_group_t group)
1840{
955ce5f5 1841 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
1842}
1843
ac27a0ec
DK
1844/*
1845 * Inodes and files operations
1846 */
1847
1848/* dir.c */
617ba13b 1849extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
1850
1851/* file.c */
c5ef1c42 1852extern const struct inode_operations ext4_file_inode_operations;
617ba13b 1853extern const struct file_operations ext4_file_operations;
ac27a0ec
DK
1854
1855/* namei.c */
c5ef1c42
AV
1856extern const struct inode_operations ext4_dir_inode_operations;
1857extern const struct inode_operations ext4_special_inode_operations;
596397b7 1858extern struct dentry *ext4_get_parent(struct dentry *child);
ac27a0ec
DK
1859
1860/* symlink.c */
c5ef1c42
AV
1861extern const struct inode_operations ext4_symlink_inode_operations;
1862extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 1863
6fd058f7
TT
1864/* block_validity */
1865extern void ext4_release_system_zone(struct super_block *sb);
1866extern int ext4_setup_system_zone(struct super_block *sb);
1867extern int __init init_ext4_system_zone(void);
1868extern void exit_ext4_system_zone(void);
1869extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1870 ext4_fsblk_t start_blk,
1871 unsigned int count);
1872
a86c6181
AT
1873/* extents.c */
1874extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1875extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
a02908f1
MC
1876extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1877 int chunk);
e35fd660
TT
1878extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
1879 struct ext4_map_blocks *map, int flags);
cf108bca 1880extern void ext4_ext_truncate(struct inode *);
a86c6181
AT
1881extern void ext4_ext_init(struct super_block *);
1882extern void ext4_ext_release(struct super_block *);
a2df2a63
AA
1883extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1884 loff_t len);
0031462b 1885extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
a1de02dc 1886 ssize_t len);
e35fd660
TT
1887extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
1888 struct ext4_map_blocks *map, int flags);
12b7ac17
TT
1889extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1890 sector_t block, unsigned int max_blocks,
c2177057 1891 struct buffer_head *bh, int flags);
3a06d778
AK
1892extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1893 __u64 start, __u64 len);
748de673
AF
1894/* move_extent.c */
1895extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1896 __u64 start_orig, __u64 start_donor,
1897 __u64 len, __u64 *moved_len);
1898
3a06d778 1899
744692dc
JZ
1900/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
1901enum ext4_state_bits {
1902 BH_Uninit /* blocks are allocated but uninitialized on disk */
1903 = BH_JBDPrivateStart,
1904};
1905
1906BUFFER_FNS(Uninit, uninit)
1907TAS_BUFFER_FNS(Uninit, uninit)
1908
2ccb5fb9
AK
1909/*
1910 * Add new method to test wether block and inode bitmaps are properly
1911 * initialized. With uninit_bg reading the block from disk is not enough
1912 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1913 */
1914#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1915
1916static inline int bitmap_uptodate(struct buffer_head *bh)
1917{
1918 return (buffer_uptodate(bh) &&
1919 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1920}
1921static inline void set_bitmap_uptodate(struct buffer_head *bh)
1922{
1923 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1924}
1925
731eb1a0
AM
1926#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
1927
ac27a0ec
DK
1928#endif /* __KERNEL__ */
1929
3dcf5451 1930#endif /* _EXT4_H */