]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blame_incremental - fs/ext4/ext4.h
jbd2: complain about descriptor block checksum errors
[mirror_ubuntu-zesty-kernel.git] / fs / ext4 / ext4.h
... / ...
CommitLineData
1/*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16#ifndef _EXT4_H
17#define _EXT4_H
18
19#include <linux/types.h>
20#include <linux/blkdev.h>
21#include <linux/magic.h>
22#include <linux/jbd2.h>
23#include <linux/quota.h>
24#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
28#include <linux/timer.h>
29#include <linux/wait.h>
30#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h>
32#include <linux/ratelimit.h>
33#include <crypto/hash.h>
34#include <linux/falloc.h>
35#ifdef __KERNEL__
36#include <linux/compat.h>
37#endif
38
39/*
40 * The fourth extended filesystem constants/structures
41 */
42
43/*
44 * Define EXT4FS_DEBUG to produce debug messages
45 */
46#undef EXT4FS_DEBUG
47
48/*
49 * Debug code
50 */
51#ifdef EXT4FS_DEBUG
52#define ext4_debug(f, a...) \
53 do { \
54 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
55 __FILE__, __LINE__, __func__); \
56 printk(KERN_DEBUG f, ## a); \
57 } while (0)
58#else
59#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
60#endif
61
62/*
63 * Turn on EXT_DEBUG to get lots of info about extents operations.
64 */
65#define EXT_DEBUG__
66#ifdef EXT_DEBUG
67#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
68#else
69#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
70#endif
71
72#define EXT4_ERROR_INODE(inode, fmt, a...) \
73 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
74
75#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
76 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
77
78#define EXT4_ERROR_FILE(file, block, fmt, a...) \
79 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
80
81/* data type for block offset of block group */
82typedef int ext4_grpblk_t;
83
84/* data type for filesystem-wide blocks number */
85typedef unsigned long long ext4_fsblk_t;
86
87/* data type for file logical block number */
88typedef __u32 ext4_lblk_t;
89
90/* data type for block group number */
91typedef unsigned int ext4_group_t;
92
93/*
94 * Flags used in mballoc's allocation_context flags field.
95 *
96 * Also used to show what's going on for debugging purposes when the
97 * flag field is exported via the traceport interface
98 */
99
100/* prefer goal again. length */
101#define EXT4_MB_HINT_MERGE 0x0001
102/* blocks already reserved */
103#define EXT4_MB_HINT_RESERVED 0x0002
104/* metadata is being allocated */
105#define EXT4_MB_HINT_METADATA 0x0004
106/* first blocks in the file */
107#define EXT4_MB_HINT_FIRST 0x0008
108/* search for the best chunk */
109#define EXT4_MB_HINT_BEST 0x0010
110/* data is being allocated */
111#define EXT4_MB_HINT_DATA 0x0020
112/* don't preallocate (for tails) */
113#define EXT4_MB_HINT_NOPREALLOC 0x0040
114/* allocate for locality group */
115#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
116/* allocate goal blocks or none */
117#define EXT4_MB_HINT_GOAL_ONLY 0x0100
118/* goal is meaningful */
119#define EXT4_MB_HINT_TRY_GOAL 0x0200
120/* blocks already pre-reserved by delayed allocation */
121#define EXT4_MB_DELALLOC_RESERVED 0x0400
122/* We are doing stream allocation */
123#define EXT4_MB_STREAM_ALLOC 0x0800
124/* Use reserved root blocks if needed */
125#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
126/* Use blocks from reserved pool */
127#define EXT4_MB_USE_RESERVED 0x2000
128
129struct ext4_allocation_request {
130 /* target inode for block we're allocating */
131 struct inode *inode;
132 /* how many blocks we want to allocate */
133 unsigned int len;
134 /* logical block in target inode */
135 ext4_lblk_t logical;
136 /* the closest logical allocated block to the left */
137 ext4_lblk_t lleft;
138 /* the closest logical allocated block to the right */
139 ext4_lblk_t lright;
140 /* phys. target (a hint) */
141 ext4_fsblk_t goal;
142 /* phys. block for the closest logical allocated block to the left */
143 ext4_fsblk_t pleft;
144 /* phys. block for the closest logical allocated block to the right */
145 ext4_fsblk_t pright;
146 /* flags. see above EXT4_MB_HINT_* */
147 unsigned int flags;
148};
149
150/*
151 * Logical to physical block mapping, used by ext4_map_blocks()
152 *
153 * This structure is used to pass requests into ext4_map_blocks() as
154 * well as to store the information returned by ext4_map_blocks(). It
155 * takes less room on the stack than a struct buffer_head.
156 */
157#define EXT4_MAP_NEW (1 << BH_New)
158#define EXT4_MAP_MAPPED (1 << BH_Mapped)
159#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
160#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
161#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
162 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
163
164struct ext4_map_blocks {
165 ext4_fsblk_t m_pblk;
166 ext4_lblk_t m_lblk;
167 unsigned int m_len;
168 unsigned int m_flags;
169};
170
171/*
172 * Flags for ext4_io_end->flags
173 */
174#define EXT4_IO_END_UNWRITTEN 0x0001
175
176/*
177 * For converting unwritten extents on a work queue. 'handle' is used for
178 * buffered writeback.
179 */
180typedef struct ext4_io_end {
181 struct list_head list; /* per-file finished IO list */
182 handle_t *handle; /* handle reserved for extent
183 * conversion */
184 struct inode *inode; /* file being written to */
185 struct bio *bio; /* Linked list of completed
186 * bios covering the extent */
187 unsigned int flag; /* unwritten or not */
188 loff_t offset; /* offset in the file */
189 ssize_t size; /* size of the extent */
190 atomic_t count; /* reference counter */
191} ext4_io_end_t;
192
193struct ext4_io_submit {
194 int io_op;
195 struct bio *io_bio;
196 ext4_io_end_t *io_end;
197 sector_t io_next_block;
198};
199
200/*
201 * Special inodes numbers
202 */
203#define EXT4_BAD_INO 1 /* Bad blocks inode */
204#define EXT4_ROOT_INO 2 /* Root inode */
205#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
206#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
207#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
208#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
209#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
210#define EXT4_JOURNAL_INO 8 /* Journal inode */
211
212/* First non-reserved inode for old ext4 filesystems */
213#define EXT4_GOOD_OLD_FIRST_INO 11
214
215/*
216 * Maximal count of links to a file
217 */
218#define EXT4_LINK_MAX 65000
219
220/*
221 * Macro-instructions used to manage several block sizes
222 */
223#define EXT4_MIN_BLOCK_SIZE 1024
224#define EXT4_MAX_BLOCK_SIZE 65536
225#define EXT4_MIN_BLOCK_LOG_SIZE 10
226#define EXT4_MAX_BLOCK_LOG_SIZE 16
227#ifdef __KERNEL__
228# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
229#else
230# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
231#endif
232#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
233#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
234 EXT4_SB(s)->s_cluster_bits)
235#ifdef __KERNEL__
236# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
237# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
238#else
239# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
240#endif
241#ifdef __KERNEL__
242#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
243#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
244#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
245#else
246#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
247 EXT4_GOOD_OLD_INODE_SIZE : \
248 (s)->s_inode_size)
249#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
250 EXT4_GOOD_OLD_FIRST_INO : \
251 (s)->s_first_ino)
252#endif
253#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
254
255/* Translate a block number to a cluster number */
256#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
257/* Translate a cluster number to a block number */
258#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
259/* Translate # of blks to # of clusters */
260#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
261 (sbi)->s_cluster_bits)
262/* Mask out the low bits to get the starting block of the cluster */
263#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
264 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
265#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
266 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
267/* Get the cluster offset */
268#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
269 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
270#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
271 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
272
273/*
274 * Structure of a blocks group descriptor
275 */
276struct ext4_group_desc
277{
278 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
279 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
280 __le32 bg_inode_table_lo; /* Inodes table block */
281 __le16 bg_free_blocks_count_lo;/* Free blocks count */
282 __le16 bg_free_inodes_count_lo;/* Free inodes count */
283 __le16 bg_used_dirs_count_lo; /* Directories count */
284 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
285 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
286 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
287 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
288 __le16 bg_itable_unused_lo; /* Unused inodes count */
289 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
290 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
291 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
292 __le32 bg_inode_table_hi; /* Inodes table block MSB */
293 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
294 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
295 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
296 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
297 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
298 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
299 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
300 __u32 bg_reserved;
301};
302
303#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
304 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
305 sizeof(__le16))
306#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
307 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
308 sizeof(__le16))
309
310/*
311 * Structure of a flex block group info
312 */
313
314struct flex_groups {
315 atomic64_t free_clusters;
316 atomic_t free_inodes;
317 atomic_t used_dirs;
318};
319
320#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
321#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
322#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
323
324/*
325 * Macro-instructions used to manage group descriptors
326 */
327#define EXT4_MIN_DESC_SIZE 32
328#define EXT4_MIN_DESC_SIZE_64BIT 64
329#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
330#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
331#ifdef __KERNEL__
332# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
333# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
334# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
335# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
336# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
337#else
338# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
339# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
340# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
341#endif
342
343/*
344 * Constants relative to the data blocks
345 */
346#define EXT4_NDIR_BLOCKS 12
347#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
348#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
349#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
350#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
351
352/*
353 * Inode flags
354 */
355#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
356#define EXT4_UNRM_FL 0x00000002 /* Undelete */
357#define EXT4_COMPR_FL 0x00000004 /* Compress file */
358#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
359#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
360#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
361#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
362#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
363/* Reserved for compression usage... */
364#define EXT4_DIRTY_FL 0x00000100
365#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
366#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
367#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
368/* End compression flags --- maybe not all used */
369#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
370#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
371#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
372#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
373#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
374#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
375#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
376#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
377#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
378#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
379#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
380#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
381
382#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
383#define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
384
385/* Flags that should be inherited by new inodes from their parent. */
386#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
387 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
388 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
389 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
390
391/* Flags that are appropriate for regular files (all but dir-specific ones). */
392#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
393
394/* Flags that are appropriate for non-directories/regular files. */
395#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
396
397/* Mask out flags that are inappropriate for the given type of inode. */
398static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
399{
400 if (S_ISDIR(mode))
401 return flags;
402 else if (S_ISREG(mode))
403 return flags & EXT4_REG_FLMASK;
404 else
405 return flags & EXT4_OTHER_FLMASK;
406}
407
408/*
409 * Inode flags used for atomic set/get
410 */
411enum {
412 EXT4_INODE_SECRM = 0, /* Secure deletion */
413 EXT4_INODE_UNRM = 1, /* Undelete */
414 EXT4_INODE_COMPR = 2, /* Compress file */
415 EXT4_INODE_SYNC = 3, /* Synchronous updates */
416 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
417 EXT4_INODE_APPEND = 5, /* writes to file may only append */
418 EXT4_INODE_NODUMP = 6, /* do not dump file */
419 EXT4_INODE_NOATIME = 7, /* do not update atime */
420/* Reserved for compression usage... */
421 EXT4_INODE_DIRTY = 8,
422 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
423 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
424 EXT4_INODE_ECOMPR = 11, /* Compression error */
425/* End compression flags --- maybe not all used */
426 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
427 EXT4_INODE_IMAGIC = 13, /* AFS directory */
428 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
429 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
430 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
431 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
432 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
433 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
434 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
435 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
436 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
437 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
438};
439
440/*
441 * Since it's pretty easy to mix up bit numbers and hex values, we use a
442 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
443 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
444 * any extra space in the compiled kernel image, otherwise, the build will fail.
445 * It's important that these values are the same, since we are using
446 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
447 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
448 * values found in ext2, ext3 and ext4 filesystems, and of course the values
449 * defined in e2fsprogs.
450 *
451 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
452 */
453#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
454#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
455
456static inline void ext4_check_flag_values(void)
457{
458 CHECK_FLAG_VALUE(SECRM);
459 CHECK_FLAG_VALUE(UNRM);
460 CHECK_FLAG_VALUE(COMPR);
461 CHECK_FLAG_VALUE(SYNC);
462 CHECK_FLAG_VALUE(IMMUTABLE);
463 CHECK_FLAG_VALUE(APPEND);
464 CHECK_FLAG_VALUE(NODUMP);
465 CHECK_FLAG_VALUE(NOATIME);
466 CHECK_FLAG_VALUE(DIRTY);
467 CHECK_FLAG_VALUE(COMPRBLK);
468 CHECK_FLAG_VALUE(NOCOMPR);
469 CHECK_FLAG_VALUE(ECOMPR);
470 CHECK_FLAG_VALUE(INDEX);
471 CHECK_FLAG_VALUE(IMAGIC);
472 CHECK_FLAG_VALUE(JOURNAL_DATA);
473 CHECK_FLAG_VALUE(NOTAIL);
474 CHECK_FLAG_VALUE(DIRSYNC);
475 CHECK_FLAG_VALUE(TOPDIR);
476 CHECK_FLAG_VALUE(HUGE_FILE);
477 CHECK_FLAG_VALUE(EXTENTS);
478 CHECK_FLAG_VALUE(EA_INODE);
479 CHECK_FLAG_VALUE(EOFBLOCKS);
480 CHECK_FLAG_VALUE(INLINE_DATA);
481 CHECK_FLAG_VALUE(RESERVED);
482}
483
484/* Used to pass group descriptor data when online resize is done */
485struct ext4_new_group_input {
486 __u32 group; /* Group number for this data */
487 __u64 block_bitmap; /* Absolute block number of block bitmap */
488 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
489 __u64 inode_table; /* Absolute block number of inode table start */
490 __u32 blocks_count; /* Total number of blocks in this group */
491 __u16 reserved_blocks; /* Number of reserved blocks in this group */
492 __u16 unused;
493};
494
495#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
496struct compat_ext4_new_group_input {
497 u32 group;
498 compat_u64 block_bitmap;
499 compat_u64 inode_bitmap;
500 compat_u64 inode_table;
501 u32 blocks_count;
502 u16 reserved_blocks;
503 u16 unused;
504};
505#endif
506
507/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
508struct ext4_new_group_data {
509 __u32 group;
510 __u64 block_bitmap;
511 __u64 inode_bitmap;
512 __u64 inode_table;
513 __u32 blocks_count;
514 __u16 reserved_blocks;
515 __u16 unused;
516 __u32 free_blocks_count;
517};
518
519/* Indexes used to index group tables in ext4_new_group_data */
520enum {
521 BLOCK_BITMAP = 0, /* block bitmap */
522 INODE_BITMAP, /* inode bitmap */
523 INODE_TABLE, /* inode tables */
524 GROUP_TABLE_COUNT,
525};
526
527/*
528 * Flags used by ext4_map_blocks()
529 */
530 /* Allocate any needed blocks and/or convert an unwritten
531 extent to be an initialized ext4 */
532#define EXT4_GET_BLOCKS_CREATE 0x0001
533 /* Request the creation of an unwritten extent */
534#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
535#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
536 EXT4_GET_BLOCKS_CREATE)
537 /* Caller is from the delayed allocation writeout path
538 * finally doing the actual allocation of delayed blocks */
539#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
540 /* caller is from the direct IO path, request to creation of an
541 unwritten extents if not allocated, split the unwritten
542 extent if blocks has been preallocated already*/
543#define EXT4_GET_BLOCKS_PRE_IO 0x0008
544#define EXT4_GET_BLOCKS_CONVERT 0x0010
545#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
546 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
547 /* Convert extent to initialized after IO complete */
548#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
549 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
550 /* Eventual metadata allocation (due to growing extent tree)
551 * should not fail, so try to use reserved blocks for that.*/
552#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
553 /* Don't normalize allocation size (used for fallocate) */
554#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
555 /* Request will not result in inode size update (user for fallocate) */
556#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
557 /* Do not take i_data_sem locking in ext4_map_blocks */
558#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
559 /* Convert written extents to unwritten */
560#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200
561
562/*
563 * The bit position of these flags must not overlap with any of the
564 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
565 * read_extent_tree_block(), ext4_split_extent_at(),
566 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
567 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
568 * caching the extents when reading from the extent tree while a
569 * truncate or punch hole operation is in progress.
570 */
571#define EXT4_EX_NOCACHE 0x40000000
572#define EXT4_EX_FORCE_CACHE 0x20000000
573
574/*
575 * Flags used by ext4_free_blocks
576 */
577#define EXT4_FREE_BLOCKS_METADATA 0x0001
578#define EXT4_FREE_BLOCKS_FORGET 0x0002
579#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
580#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
581#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
582#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
583
584/*
585 * ioctl commands
586 */
587#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
588#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
589#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
590#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
591#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
592#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
593#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
594#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
595#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
596#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
597#define EXT4_IOC_MIGRATE _IO('f', 9)
598 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
599 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
600#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
601#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
602#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
603#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
604#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
605
606#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
607/*
608 * ioctl commands in 32 bit emulation
609 */
610#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
611#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
612#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
613#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
614#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
615#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
616#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
617#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
618#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
619#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
620#endif
621
622/* Max physical block we can address w/o extents */
623#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
624
625/*
626 * Structure of an inode on the disk
627 */
628struct ext4_inode {
629 __le16 i_mode; /* File mode */
630 __le16 i_uid; /* Low 16 bits of Owner Uid */
631 __le32 i_size_lo; /* Size in bytes */
632 __le32 i_atime; /* Access time */
633 __le32 i_ctime; /* Inode Change time */
634 __le32 i_mtime; /* Modification time */
635 __le32 i_dtime; /* Deletion Time */
636 __le16 i_gid; /* Low 16 bits of Group Id */
637 __le16 i_links_count; /* Links count */
638 __le32 i_blocks_lo; /* Blocks count */
639 __le32 i_flags; /* File flags */
640 union {
641 struct {
642 __le32 l_i_version;
643 } linux1;
644 struct {
645 __u32 h_i_translator;
646 } hurd1;
647 struct {
648 __u32 m_i_reserved1;
649 } masix1;
650 } osd1; /* OS dependent 1 */
651 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
652 __le32 i_generation; /* File version (for NFS) */
653 __le32 i_file_acl_lo; /* File ACL */
654 __le32 i_size_high;
655 __le32 i_obso_faddr; /* Obsoleted fragment address */
656 union {
657 struct {
658 __le16 l_i_blocks_high; /* were l_i_reserved1 */
659 __le16 l_i_file_acl_high;
660 __le16 l_i_uid_high; /* these 2 fields */
661 __le16 l_i_gid_high; /* were reserved2[0] */
662 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
663 __le16 l_i_reserved;
664 } linux2;
665 struct {
666 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
667 __u16 h_i_mode_high;
668 __u16 h_i_uid_high;
669 __u16 h_i_gid_high;
670 __u32 h_i_author;
671 } hurd2;
672 struct {
673 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
674 __le16 m_i_file_acl_high;
675 __u32 m_i_reserved2[2];
676 } masix2;
677 } osd2; /* OS dependent 2 */
678 __le16 i_extra_isize;
679 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
680 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
681 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
682 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
683 __le32 i_crtime; /* File Creation time */
684 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
685 __le32 i_version_hi; /* high 32 bits for 64-bit version */
686};
687
688struct move_extent {
689 __u32 reserved; /* should be zero */
690 __u32 donor_fd; /* donor file descriptor */
691 __u64 orig_start; /* logical start offset in block for orig */
692 __u64 donor_start; /* logical start offset in block for donor */
693 __u64 len; /* block length to be moved */
694 __u64 moved_len; /* moved block length */
695};
696
697#define EXT4_EPOCH_BITS 2
698#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
699#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
700
701/*
702 * Extended fields will fit into an inode if the filesystem was formatted
703 * with large inodes (-I 256 or larger) and there are not currently any EAs
704 * consuming all of the available space. For new inodes we always reserve
705 * enough space for the kernel's known extended fields, but for inodes
706 * created with an old kernel this might not have been the case. None of
707 * the extended inode fields is critical for correct filesystem operation.
708 * This macro checks if a certain field fits in the inode. Note that
709 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
710 */
711#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
712 ((offsetof(typeof(*ext4_inode), field) + \
713 sizeof((ext4_inode)->field)) \
714 <= (EXT4_GOOD_OLD_INODE_SIZE + \
715 (einode)->i_extra_isize)) \
716
717static inline __le32 ext4_encode_extra_time(struct timespec *time)
718{
719 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
720 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
721 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
722}
723
724static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
725{
726 if (sizeof(time->tv_sec) > 4)
727 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
728 << 32;
729 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
730}
731
732#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
733do { \
734 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
735 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
736 (raw_inode)->xtime ## _extra = \
737 ext4_encode_extra_time(&(inode)->xtime); \
738} while (0)
739
740#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
741do { \
742 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
743 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
744 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
745 (raw_inode)->xtime ## _extra = \
746 ext4_encode_extra_time(&(einode)->xtime); \
747} while (0)
748
749#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
750do { \
751 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
752 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
753 ext4_decode_extra_time(&(inode)->xtime, \
754 raw_inode->xtime ## _extra); \
755 else \
756 (inode)->xtime.tv_nsec = 0; \
757} while (0)
758
759#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
760do { \
761 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
762 (einode)->xtime.tv_sec = \
763 (signed)le32_to_cpu((raw_inode)->xtime); \
764 else \
765 (einode)->xtime.tv_sec = 0; \
766 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
767 ext4_decode_extra_time(&(einode)->xtime, \
768 raw_inode->xtime ## _extra); \
769 else \
770 (einode)->xtime.tv_nsec = 0; \
771} while (0)
772
773#define i_disk_version osd1.linux1.l_i_version
774
775#if defined(__KERNEL__) || defined(__linux__)
776#define i_reserved1 osd1.linux1.l_i_reserved1
777#define i_file_acl_high osd2.linux2.l_i_file_acl_high
778#define i_blocks_high osd2.linux2.l_i_blocks_high
779#define i_uid_low i_uid
780#define i_gid_low i_gid
781#define i_uid_high osd2.linux2.l_i_uid_high
782#define i_gid_high osd2.linux2.l_i_gid_high
783#define i_checksum_lo osd2.linux2.l_i_checksum_lo
784
785#elif defined(__GNU__)
786
787#define i_translator osd1.hurd1.h_i_translator
788#define i_uid_high osd2.hurd2.h_i_uid_high
789#define i_gid_high osd2.hurd2.h_i_gid_high
790#define i_author osd2.hurd2.h_i_author
791
792#elif defined(__masix__)
793
794#define i_reserved1 osd1.masix1.m_i_reserved1
795#define i_file_acl_high osd2.masix2.m_i_file_acl_high
796#define i_reserved2 osd2.masix2.m_i_reserved2
797
798#endif /* defined(__KERNEL__) || defined(__linux__) */
799
800#include "extents_status.h"
801
802/*
803 * fourth extended file system inode data in memory
804 */
805struct ext4_inode_info {
806 __le32 i_data[15]; /* unconverted */
807 __u32 i_dtime;
808 ext4_fsblk_t i_file_acl;
809
810 /*
811 * i_block_group is the number of the block group which contains
812 * this file's inode. Constant across the lifetime of the inode,
813 * it is ued for making block allocation decisions - we try to
814 * place a file's data blocks near its inode block, and new inodes
815 * near to their parent directory's inode.
816 */
817 ext4_group_t i_block_group;
818 ext4_lblk_t i_dir_start_lookup;
819#if (BITS_PER_LONG < 64)
820 unsigned long i_state_flags; /* Dynamic state flags */
821#endif
822 unsigned long i_flags;
823
824 /*
825 * Extended attributes can be read independently of the main file
826 * data. Taking i_mutex even when reading would cause contention
827 * between readers of EAs and writers of regular file data, so
828 * instead we synchronize on xattr_sem when reading or changing
829 * EAs.
830 */
831 struct rw_semaphore xattr_sem;
832
833 struct list_head i_orphan; /* unlinked but open inodes */
834
835 /*
836 * i_disksize keeps track of what the inode size is ON DISK, not
837 * in memory. During truncate, i_size is set to the new size by
838 * the VFS prior to calling ext4_truncate(), but the filesystem won't
839 * set i_disksize to 0 until the truncate is actually under way.
840 *
841 * The intent is that i_disksize always represents the blocks which
842 * are used by this file. This allows recovery to restart truncate
843 * on orphans if we crash during truncate. We actually write i_disksize
844 * into the on-disk inode when writing inodes out, instead of i_size.
845 *
846 * The only time when i_disksize and i_size may be different is when
847 * a truncate is in progress. The only things which change i_disksize
848 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
849 */
850 loff_t i_disksize;
851
852 /*
853 * i_data_sem is for serialising ext4_truncate() against
854 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
855 * data tree are chopped off during truncate. We can't do that in
856 * ext4 because whenever we perform intermediate commits during
857 * truncate, the inode and all the metadata blocks *must* be in a
858 * consistent state which allows truncation of the orphans to restart
859 * during recovery. Hence we must fix the get_block-vs-truncate race
860 * by other means, so we have i_data_sem.
861 */
862 struct rw_semaphore i_data_sem;
863 struct inode vfs_inode;
864 struct jbd2_inode *jinode;
865
866 spinlock_t i_raw_lock; /* protects updates to the raw inode */
867
868 /*
869 * File creation time. Its function is same as that of
870 * struct timespec i_{a,c,m}time in the generic inode.
871 */
872 struct timespec i_crtime;
873
874 /* mballoc */
875 struct list_head i_prealloc_list;
876 spinlock_t i_prealloc_lock;
877
878 /* extents status tree */
879 struct ext4_es_tree i_es_tree;
880 rwlock_t i_es_lock;
881 struct list_head i_es_list;
882 unsigned int i_es_all_nr; /* protected by i_es_lock */
883 unsigned int i_es_shk_nr; /* protected by i_es_lock */
884 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
885 extents to shrink. Protected by
886 i_es_lock */
887
888 /* ialloc */
889 ext4_group_t i_last_alloc_group;
890
891 /* allocation reservation info for delalloc */
892 /* In case of bigalloc, these refer to clusters rather than blocks */
893 unsigned int i_reserved_data_blocks;
894 unsigned int i_reserved_meta_blocks;
895 unsigned int i_allocated_meta_blocks;
896 ext4_lblk_t i_da_metadata_calc_last_lblock;
897 int i_da_metadata_calc_len;
898
899 /* on-disk additional length */
900 __u16 i_extra_isize;
901
902 /* Indicate the inline data space. */
903 u16 i_inline_off;
904 u16 i_inline_size;
905
906#ifdef CONFIG_QUOTA
907 /* quota space reservation, managed internally by quota code */
908 qsize_t i_reserved_quota;
909#endif
910
911 /* Lock protecting lists below */
912 spinlock_t i_completed_io_lock;
913 /*
914 * Completed IOs that need unwritten extents handling and have
915 * transaction reserved
916 */
917 struct list_head i_rsv_conversion_list;
918 /*
919 * Completed IOs that need unwritten extents handling and don't have
920 * transaction reserved
921 */
922 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
923 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
924 struct work_struct i_rsv_conversion_work;
925
926 spinlock_t i_block_reservation_lock;
927
928 /*
929 * Transactions that contain inode's metadata needed to complete
930 * fsync and fdatasync, respectively.
931 */
932 tid_t i_sync_tid;
933 tid_t i_datasync_tid;
934
935#ifdef CONFIG_QUOTA
936 struct dquot *i_dquot[MAXQUOTAS];
937#endif
938
939 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
940 __u32 i_csum_seed;
941};
942
943/*
944 * File system states
945 */
946#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
947#define EXT4_ERROR_FS 0x0002 /* Errors detected */
948#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
949
950/*
951 * Misc. filesystem flags
952 */
953#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
954#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
955#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
956
957/*
958 * Mount flags set via mount options or defaults
959 */
960#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
961#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
962#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
963#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
964#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
965#define EXT4_MOUNT_ERRORS_MASK 0x00070
966#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
967#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
968#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
969#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
970#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
971#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
972#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
973#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
974#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
975#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
976#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
977#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
978#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
979#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
980#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
981#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
982#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
983#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
984#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
985#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
986#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
987#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
988#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
989
990/*
991 * Mount flags set either automatically (could not be set by mount option)
992 * based on per file system feature or property or in special cases such as
993 * distinguishing between explicit mount option definition and default.
994 */
995#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
996 specified delalloc */
997#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
998 size of blocksize * 8
999 blocks */
1000#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1001 file systems */
1002
1003#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1004 ~EXT4_MOUNT_##opt
1005#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1006 EXT4_MOUNT_##opt
1007#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1008 EXT4_MOUNT_##opt)
1009
1010#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1011 ~EXT4_MOUNT2_##opt
1012#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1013 EXT4_MOUNT2_##opt
1014#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1015 EXT4_MOUNT2_##opt)
1016
1017#define ext4_test_and_set_bit __test_and_set_bit_le
1018#define ext4_set_bit __set_bit_le
1019#define ext4_set_bit_atomic ext2_set_bit_atomic
1020#define ext4_test_and_clear_bit __test_and_clear_bit_le
1021#define ext4_clear_bit __clear_bit_le
1022#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1023#define ext4_test_bit test_bit_le
1024#define ext4_find_next_zero_bit find_next_zero_bit_le
1025#define ext4_find_next_bit find_next_bit_le
1026
1027extern void ext4_set_bits(void *bm, int cur, int len);
1028
1029/*
1030 * Maximal mount counts between two filesystem checks
1031 */
1032#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1033#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1034
1035/*
1036 * Behaviour when detecting errors
1037 */
1038#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1039#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1040#define EXT4_ERRORS_PANIC 3 /* Panic */
1041#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1042
1043/* Metadata checksum algorithm codes */
1044#define EXT4_CRC32C_CHKSUM 1
1045
1046/*
1047 * Structure of the super block
1048 */
1049struct ext4_super_block {
1050/*00*/ __le32 s_inodes_count; /* Inodes count */
1051 __le32 s_blocks_count_lo; /* Blocks count */
1052 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1053 __le32 s_free_blocks_count_lo; /* Free blocks count */
1054/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1055 __le32 s_first_data_block; /* First Data Block */
1056 __le32 s_log_block_size; /* Block size */
1057 __le32 s_log_cluster_size; /* Allocation cluster size */
1058/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1059 __le32 s_clusters_per_group; /* # Clusters per group */
1060 __le32 s_inodes_per_group; /* # Inodes per group */
1061 __le32 s_mtime; /* Mount time */
1062/*30*/ __le32 s_wtime; /* Write time */
1063 __le16 s_mnt_count; /* Mount count */
1064 __le16 s_max_mnt_count; /* Maximal mount count */
1065 __le16 s_magic; /* Magic signature */
1066 __le16 s_state; /* File system state */
1067 __le16 s_errors; /* Behaviour when detecting errors */
1068 __le16 s_minor_rev_level; /* minor revision level */
1069/*40*/ __le32 s_lastcheck; /* time of last check */
1070 __le32 s_checkinterval; /* max. time between checks */
1071 __le32 s_creator_os; /* OS */
1072 __le32 s_rev_level; /* Revision level */
1073/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1074 __le16 s_def_resgid; /* Default gid for reserved blocks */
1075 /*
1076 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1077 *
1078 * Note: the difference between the compatible feature set and
1079 * the incompatible feature set is that if there is a bit set
1080 * in the incompatible feature set that the kernel doesn't
1081 * know about, it should refuse to mount the filesystem.
1082 *
1083 * e2fsck's requirements are more strict; if it doesn't know
1084 * about a feature in either the compatible or incompatible
1085 * feature set, it must abort and not try to meddle with
1086 * things it doesn't understand...
1087 */
1088 __le32 s_first_ino; /* First non-reserved inode */
1089 __le16 s_inode_size; /* size of inode structure */
1090 __le16 s_block_group_nr; /* block group # of this superblock */
1091 __le32 s_feature_compat; /* compatible feature set */
1092/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1093 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1094/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1095/*78*/ char s_volume_name[16]; /* volume name */
1096/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1097/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1098 /*
1099 * Performance hints. Directory preallocation should only
1100 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1101 */
1102 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1103 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1104 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1105 /*
1106 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1107 */
1108/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1109/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1110 __le32 s_journal_dev; /* device number of journal file */
1111 __le32 s_last_orphan; /* start of list of inodes to delete */
1112 __le32 s_hash_seed[4]; /* HTREE hash seed */
1113 __u8 s_def_hash_version; /* Default hash version to use */
1114 __u8 s_jnl_backup_type;
1115 __le16 s_desc_size; /* size of group descriptor */
1116/*100*/ __le32 s_default_mount_opts;
1117 __le32 s_first_meta_bg; /* First metablock block group */
1118 __le32 s_mkfs_time; /* When the filesystem was created */
1119 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1120 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1121/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1122 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1123 __le32 s_free_blocks_count_hi; /* Free blocks count */
1124 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1125 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1126 __le32 s_flags; /* Miscellaneous flags */
1127 __le16 s_raid_stride; /* RAID stride */
1128 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1129 __le64 s_mmp_block; /* Block for multi-mount protection */
1130 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1131 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1132 __u8 s_checksum_type; /* metadata checksum algorithm used */
1133 __le16 s_reserved_pad;
1134 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1135 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1136 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1137 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1138 snapshot's future use */
1139 __le32 s_snapshot_list; /* inode number of the head of the
1140 on-disk snapshot list */
1141#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1142 __le32 s_error_count; /* number of fs errors */
1143 __le32 s_first_error_time; /* first time an error happened */
1144 __le32 s_first_error_ino; /* inode involved in first error */
1145 __le64 s_first_error_block; /* block involved of first error */
1146 __u8 s_first_error_func[32]; /* function where the error happened */
1147 __le32 s_first_error_line; /* line number where error happened */
1148 __le32 s_last_error_time; /* most recent time of an error */
1149 __le32 s_last_error_ino; /* inode involved in last error */
1150 __le32 s_last_error_line; /* line number where error happened */
1151 __le64 s_last_error_block; /* block involved of last error */
1152 __u8 s_last_error_func[32]; /* function where the error happened */
1153#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1154 __u8 s_mount_opts[64];
1155 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1156 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1157 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1158 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1159 __le32 s_reserved[106]; /* Padding to the end of the block */
1160 __le32 s_checksum; /* crc32c(superblock) */
1161};
1162
1163#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1164
1165#ifdef __KERNEL__
1166
1167/*
1168 * run-time mount flags
1169 */
1170#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1171#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1172
1173/* Number of quota types we support */
1174#define EXT4_MAXQUOTAS 2
1175
1176/*
1177 * fourth extended-fs super-block data in memory
1178 */
1179struct ext4_sb_info {
1180 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1181 unsigned long s_inodes_per_block;/* Number of inodes per block */
1182 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1183 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1184 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1185 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1186 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1187 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1188 ext4_group_t s_groups_count; /* Number of groups in the fs */
1189 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1190 unsigned long s_overhead; /* # of fs overhead clusters */
1191 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1192 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1193 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1194 struct buffer_head * s_sbh; /* Buffer containing the super block */
1195 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1196 struct buffer_head **s_group_desc;
1197 unsigned int s_mount_opt;
1198 unsigned int s_mount_opt2;
1199 unsigned int s_mount_flags;
1200 unsigned int s_def_mount_opt;
1201 ext4_fsblk_t s_sb_block;
1202 atomic64_t s_resv_clusters;
1203 kuid_t s_resuid;
1204 kgid_t s_resgid;
1205 unsigned short s_mount_state;
1206 unsigned short s_pad;
1207 int s_addr_per_block_bits;
1208 int s_desc_per_block_bits;
1209 int s_inode_size;
1210 int s_first_ino;
1211 unsigned int s_inode_readahead_blks;
1212 unsigned int s_inode_goal;
1213 spinlock_t s_next_gen_lock;
1214 u32 s_next_generation;
1215 u32 s_hash_seed[4];
1216 int s_def_hash_version;
1217 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1218 struct percpu_counter s_freeclusters_counter;
1219 struct percpu_counter s_freeinodes_counter;
1220 struct percpu_counter s_dirs_counter;
1221 struct percpu_counter s_dirtyclusters_counter;
1222 struct blockgroup_lock *s_blockgroup_lock;
1223 struct proc_dir_entry *s_proc;
1224 struct kobject s_kobj;
1225 struct completion s_kobj_unregister;
1226 struct super_block *s_sb;
1227
1228 /* Journaling */
1229 struct journal_s *s_journal;
1230 struct list_head s_orphan;
1231 struct mutex s_orphan_lock;
1232 unsigned long s_resize_flags; /* Flags indicating if there
1233 is a resizer */
1234 unsigned long s_commit_interval;
1235 u32 s_max_batch_time;
1236 u32 s_min_batch_time;
1237 struct block_device *journal_bdev;
1238#ifdef CONFIG_QUOTA
1239 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1240 int s_jquota_fmt; /* Format of quota to use */
1241#endif
1242 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1243 struct rb_root system_blks;
1244
1245#ifdef EXTENTS_STATS
1246 /* ext4 extents stats */
1247 unsigned long s_ext_min;
1248 unsigned long s_ext_max;
1249 unsigned long s_depth_max;
1250 spinlock_t s_ext_stats_lock;
1251 unsigned long s_ext_blocks;
1252 unsigned long s_ext_extents;
1253#endif
1254
1255 /* for buddy allocator */
1256 struct ext4_group_info ***s_group_info;
1257 struct inode *s_buddy_cache;
1258 spinlock_t s_md_lock;
1259 unsigned short *s_mb_offsets;
1260 unsigned int *s_mb_maxs;
1261 unsigned int s_group_info_size;
1262
1263 /* tunables */
1264 unsigned long s_stripe;
1265 unsigned int s_mb_stream_request;
1266 unsigned int s_mb_max_to_scan;
1267 unsigned int s_mb_min_to_scan;
1268 unsigned int s_mb_stats;
1269 unsigned int s_mb_order2_reqs;
1270 unsigned int s_mb_group_prealloc;
1271 unsigned int s_max_dir_size_kb;
1272 /* where last allocation was done - for stream allocation */
1273 unsigned long s_mb_last_group;
1274 unsigned long s_mb_last_start;
1275
1276 /* stats for buddy allocator */
1277 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1278 atomic_t s_bal_success; /* we found long enough chunks */
1279 atomic_t s_bal_allocated; /* in blocks */
1280 atomic_t s_bal_ex_scanned; /* total extents scanned */
1281 atomic_t s_bal_goals; /* goal hits */
1282 atomic_t s_bal_breaks; /* too long searches */
1283 atomic_t s_bal_2orders; /* 2^order hits */
1284 spinlock_t s_bal_lock;
1285 unsigned long s_mb_buddies_generated;
1286 unsigned long long s_mb_generation_time;
1287 atomic_t s_mb_lost_chunks;
1288 atomic_t s_mb_preallocated;
1289 atomic_t s_mb_discarded;
1290 atomic_t s_lock_busy;
1291
1292 /* locality groups */
1293 struct ext4_locality_group __percpu *s_locality_groups;
1294
1295 /* for write statistics */
1296 unsigned long s_sectors_written_start;
1297 u64 s_kbytes_written;
1298
1299 /* the size of zero-out chunk */
1300 unsigned int s_extent_max_zeroout_kb;
1301
1302 unsigned int s_log_groups_per_flex;
1303 struct flex_groups *s_flex_groups;
1304 ext4_group_t s_flex_groups_allocated;
1305
1306 /* workqueue for reserved extent conversions (buffered io) */
1307 struct workqueue_struct *rsv_conversion_wq;
1308
1309 /* timer for periodic error stats printing */
1310 struct timer_list s_err_report;
1311
1312 /* Lazy inode table initialization info */
1313 struct ext4_li_request *s_li_request;
1314 /* Wait multiplier for lazy initialization thread */
1315 unsigned int s_li_wait_mult;
1316
1317 /* Kernel thread for multiple mount protection */
1318 struct task_struct *s_mmp_tsk;
1319
1320 /* record the last minlen when FITRIM is called. */
1321 atomic_t s_last_trim_minblks;
1322
1323 /* Reference to checksum algorithm driver via cryptoapi */
1324 struct crypto_shash *s_chksum_driver;
1325
1326 /* Precomputed FS UUID checksum for seeding other checksums */
1327 __u32 s_csum_seed;
1328
1329 /* Reclaim extents from extent status tree */
1330 struct shrinker s_es_shrinker;
1331 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1332 long s_es_nr_inode;
1333 struct ext4_es_stats s_es_stats;
1334 struct mb_cache *s_mb_cache;
1335 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1336
1337 /* Ratelimit ext4 messages. */
1338 struct ratelimit_state s_err_ratelimit_state;
1339 struct ratelimit_state s_warning_ratelimit_state;
1340 struct ratelimit_state s_msg_ratelimit_state;
1341};
1342
1343static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1344{
1345 return sb->s_fs_info;
1346}
1347static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1348{
1349 return container_of(inode, struct ext4_inode_info, vfs_inode);
1350}
1351
1352static inline struct timespec ext4_current_time(struct inode *inode)
1353{
1354 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1355 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1356}
1357
1358static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1359{
1360 return ino == EXT4_ROOT_INO ||
1361 ino == EXT4_USR_QUOTA_INO ||
1362 ino == EXT4_GRP_QUOTA_INO ||
1363 ino == EXT4_BOOT_LOADER_INO ||
1364 ino == EXT4_JOURNAL_INO ||
1365 ino == EXT4_RESIZE_INO ||
1366 (ino >= EXT4_FIRST_INO(sb) &&
1367 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1368}
1369
1370static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1371 struct ext4_io_end *io_end)
1372{
1373 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1374 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1375 atomic_inc(&EXT4_I(inode)->i_unwritten);
1376 }
1377}
1378
1379static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1380{
1381 return inode->i_private;
1382}
1383
1384static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1385{
1386 inode->i_private = io;
1387}
1388
1389/*
1390 * Inode dynamic state flags
1391 */
1392enum {
1393 EXT4_STATE_JDATA, /* journaled data exists */
1394 EXT4_STATE_NEW, /* inode is newly created */
1395 EXT4_STATE_XATTR, /* has in-inode xattrs */
1396 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1397 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1398 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1399 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1400 EXT4_STATE_NEWENTRY, /* File just added to dir */
1401 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1402 nolocking */
1403 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1404 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1405 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1406};
1407
1408#define EXT4_INODE_BIT_FNS(name, field, offset) \
1409static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1410{ \
1411 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1412} \
1413static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1414{ \
1415 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1416} \
1417static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1418{ \
1419 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1420}
1421
1422/* Add these declarations here only so that these functions can be
1423 * found by name. Otherwise, they are very hard to locate. */
1424static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1425static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1426static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1427EXT4_INODE_BIT_FNS(flag, flags, 0)
1428
1429/* Add these declarations here only so that these functions can be
1430 * found by name. Otherwise, they are very hard to locate. */
1431static inline int ext4_test_inode_state(struct inode *inode, int bit);
1432static inline void ext4_set_inode_state(struct inode *inode, int bit);
1433static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1434#if (BITS_PER_LONG < 64)
1435EXT4_INODE_BIT_FNS(state, state_flags, 0)
1436
1437static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1438{
1439 (ei)->i_state_flags = 0;
1440}
1441#else
1442EXT4_INODE_BIT_FNS(state, flags, 32)
1443
1444static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1445{
1446 /* We depend on the fact that callers will set i_flags */
1447}
1448#endif
1449#else
1450/* Assume that user mode programs are passing in an ext4fs superblock, not
1451 * a kernel struct super_block. This will allow us to call the feature-test
1452 * macros from user land. */
1453#define EXT4_SB(sb) (sb)
1454#endif
1455
1456#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1457
1458/*
1459 * Codes for operating systems
1460 */
1461#define EXT4_OS_LINUX 0
1462#define EXT4_OS_HURD 1
1463#define EXT4_OS_MASIX 2
1464#define EXT4_OS_FREEBSD 3
1465#define EXT4_OS_LITES 4
1466
1467/*
1468 * Revision levels
1469 */
1470#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1471#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1472
1473#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1474#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1475
1476#define EXT4_GOOD_OLD_INODE_SIZE 128
1477
1478/*
1479 * Feature set definitions
1480 */
1481
1482#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1483 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1484#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1485 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1486#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1487 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1488#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1489 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1490#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1491 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1492#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1493 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1494#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1495 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1496#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1497 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1498#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1499 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1500
1501#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1502#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1503#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1504#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1505#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1506#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1507#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1508
1509#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1510#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1511#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1512#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1513#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1514#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1515#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1516#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1517#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1518/*
1519 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1520 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1521 * all other data structures' checksums. However, the METADATA_CSUM and
1522 * GDT_CSUM bits are mutually exclusive.
1523 */
1524#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1525
1526#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1527#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1528#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1529#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1530#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1531#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1532#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1533#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1534#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1535#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1536#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1537#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1538#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1539#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1540
1541#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1542#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1543 EXT4_FEATURE_INCOMPAT_META_BG)
1544#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1545 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1546 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1547
1548#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1549#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1550 EXT4_FEATURE_INCOMPAT_RECOVER| \
1551 EXT4_FEATURE_INCOMPAT_META_BG)
1552#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1553 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1554 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1555
1556#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1557#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1558 EXT4_FEATURE_INCOMPAT_RECOVER| \
1559 EXT4_FEATURE_INCOMPAT_META_BG| \
1560 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1561 EXT4_FEATURE_INCOMPAT_64BIT| \
1562 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1563 EXT4_FEATURE_INCOMPAT_MMP | \
1564 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1565#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1566 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1567 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1568 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1569 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1570 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1571 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1572 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1573 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1574 EXT4_FEATURE_RO_COMPAT_QUOTA)
1575
1576/*
1577 * Default values for user and/or group using reserved blocks
1578 */
1579#define EXT4_DEF_RESUID 0
1580#define EXT4_DEF_RESGID 0
1581
1582#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1583
1584/*
1585 * Default mount options
1586 */
1587#define EXT4_DEFM_DEBUG 0x0001
1588#define EXT4_DEFM_BSDGROUPS 0x0002
1589#define EXT4_DEFM_XATTR_USER 0x0004
1590#define EXT4_DEFM_ACL 0x0008
1591#define EXT4_DEFM_UID16 0x0010
1592#define EXT4_DEFM_JMODE 0x0060
1593#define EXT4_DEFM_JMODE_DATA 0x0020
1594#define EXT4_DEFM_JMODE_ORDERED 0x0040
1595#define EXT4_DEFM_JMODE_WBACK 0x0060
1596#define EXT4_DEFM_NOBARRIER 0x0100
1597#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1598#define EXT4_DEFM_DISCARD 0x0400
1599#define EXT4_DEFM_NODELALLOC 0x0800
1600
1601/*
1602 * Default journal batch times
1603 */
1604#define EXT4_DEF_MIN_BATCH_TIME 0
1605#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1606
1607/*
1608 * Minimum number of groups in a flexgroup before we separate out
1609 * directories into the first block group of a flexgroup
1610 */
1611#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1612
1613/*
1614 * Structure of a directory entry
1615 */
1616#define EXT4_NAME_LEN 255
1617
1618struct ext4_dir_entry {
1619 __le32 inode; /* Inode number */
1620 __le16 rec_len; /* Directory entry length */
1621 __le16 name_len; /* Name length */
1622 char name[EXT4_NAME_LEN]; /* File name */
1623};
1624
1625/*
1626 * The new version of the directory entry. Since EXT4 structures are
1627 * stored in intel byte order, and the name_len field could never be
1628 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1629 * file_type field.
1630 */
1631struct ext4_dir_entry_2 {
1632 __le32 inode; /* Inode number */
1633 __le16 rec_len; /* Directory entry length */
1634 __u8 name_len; /* Name length */
1635 __u8 file_type;
1636 char name[EXT4_NAME_LEN]; /* File name */
1637};
1638
1639/*
1640 * This is a bogus directory entry at the end of each leaf block that
1641 * records checksums.
1642 */
1643struct ext4_dir_entry_tail {
1644 __le32 det_reserved_zero1; /* Pretend to be unused */
1645 __le16 det_rec_len; /* 12 */
1646 __u8 det_reserved_zero2; /* Zero name length */
1647 __u8 det_reserved_ft; /* 0xDE, fake file type */
1648 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1649};
1650
1651#define EXT4_DIRENT_TAIL(block, blocksize) \
1652 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1653 ((blocksize) - \
1654 sizeof(struct ext4_dir_entry_tail))))
1655
1656/*
1657 * Ext4 directory file types. Only the low 3 bits are used. The
1658 * other bits are reserved for now.
1659 */
1660#define EXT4_FT_UNKNOWN 0
1661#define EXT4_FT_REG_FILE 1
1662#define EXT4_FT_DIR 2
1663#define EXT4_FT_CHRDEV 3
1664#define EXT4_FT_BLKDEV 4
1665#define EXT4_FT_FIFO 5
1666#define EXT4_FT_SOCK 6
1667#define EXT4_FT_SYMLINK 7
1668
1669#define EXT4_FT_MAX 8
1670
1671#define EXT4_FT_DIR_CSUM 0xDE
1672
1673/*
1674 * EXT4_DIR_PAD defines the directory entries boundaries
1675 *
1676 * NOTE: It must be a multiple of 4
1677 */
1678#define EXT4_DIR_PAD 4
1679#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1680#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1681 ~EXT4_DIR_ROUND)
1682#define EXT4_MAX_REC_LEN ((1<<16)-1)
1683
1684/*
1685 * If we ever get support for fs block sizes > page_size, we'll need
1686 * to remove the #if statements in the next two functions...
1687 */
1688static inline unsigned int
1689ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1690{
1691 unsigned len = le16_to_cpu(dlen);
1692
1693#if (PAGE_CACHE_SIZE >= 65536)
1694 if (len == EXT4_MAX_REC_LEN || len == 0)
1695 return blocksize;
1696 return (len & 65532) | ((len & 3) << 16);
1697#else
1698 return len;
1699#endif
1700}
1701
1702static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1703{
1704 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1705 BUG();
1706#if (PAGE_CACHE_SIZE >= 65536)
1707 if (len < 65536)
1708 return cpu_to_le16(len);
1709 if (len == blocksize) {
1710 if (blocksize == 65536)
1711 return cpu_to_le16(EXT4_MAX_REC_LEN);
1712 else
1713 return cpu_to_le16(0);
1714 }
1715 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1716#else
1717 return cpu_to_le16(len);
1718#endif
1719}
1720
1721/*
1722 * Hash Tree Directory indexing
1723 * (c) Daniel Phillips, 2001
1724 */
1725
1726#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1727 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1728 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1729#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1730#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1731
1732/* Legal values for the dx_root hash_version field: */
1733
1734#define DX_HASH_LEGACY 0
1735#define DX_HASH_HALF_MD4 1
1736#define DX_HASH_TEA 2
1737#define DX_HASH_LEGACY_UNSIGNED 3
1738#define DX_HASH_HALF_MD4_UNSIGNED 4
1739#define DX_HASH_TEA_UNSIGNED 5
1740
1741static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1742 const void *address, unsigned int length)
1743{
1744 struct {
1745 struct shash_desc shash;
1746 char ctx[4];
1747 } desc;
1748 int err;
1749
1750 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1751
1752 desc.shash.tfm = sbi->s_chksum_driver;
1753 desc.shash.flags = 0;
1754 *(u32 *)desc.ctx = crc;
1755
1756 err = crypto_shash_update(&desc.shash, address, length);
1757 BUG_ON(err);
1758
1759 return *(u32 *)desc.ctx;
1760}
1761
1762#ifdef __KERNEL__
1763
1764/* hash info structure used by the directory hash */
1765struct dx_hash_info
1766{
1767 u32 hash;
1768 u32 minor_hash;
1769 int hash_version;
1770 u32 *seed;
1771};
1772
1773
1774/* 32 and 64 bit signed EOF for dx directories */
1775#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1776#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1777
1778
1779/*
1780 * Control parameters used by ext4_htree_next_block
1781 */
1782#define HASH_NB_ALWAYS 1
1783
1784
1785/*
1786 * Describe an inode's exact location on disk and in memory
1787 */
1788struct ext4_iloc
1789{
1790 struct buffer_head *bh;
1791 unsigned long offset;
1792 ext4_group_t block_group;
1793};
1794
1795static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1796{
1797 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1798}
1799
1800/*
1801 * This structure is stuffed into the struct file's private_data field
1802 * for directories. It is where we put information so that we can do
1803 * readdir operations in hash tree order.
1804 */
1805struct dir_private_info {
1806 struct rb_root root;
1807 struct rb_node *curr_node;
1808 struct fname *extra_fname;
1809 loff_t last_pos;
1810 __u32 curr_hash;
1811 __u32 curr_minor_hash;
1812 __u32 next_hash;
1813};
1814
1815/* calculate the first block number of the group */
1816static inline ext4_fsblk_t
1817ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1818{
1819 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1820 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1821}
1822
1823/*
1824 * Special error return code only used by dx_probe() and its callers.
1825 */
1826#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
1827
1828/*
1829 * Timeout and state flag for lazy initialization inode thread.
1830 */
1831#define EXT4_DEF_LI_WAIT_MULT 10
1832#define EXT4_DEF_LI_MAX_START_DELAY 5
1833#define EXT4_LAZYINIT_QUIT 0x0001
1834#define EXT4_LAZYINIT_RUNNING 0x0002
1835
1836/*
1837 * Lazy inode table initialization info
1838 */
1839struct ext4_lazy_init {
1840 unsigned long li_state;
1841 struct list_head li_request_list;
1842 struct mutex li_list_mtx;
1843};
1844
1845struct ext4_li_request {
1846 struct super_block *lr_super;
1847 struct ext4_sb_info *lr_sbi;
1848 ext4_group_t lr_next_group;
1849 struct list_head lr_request;
1850 unsigned long lr_next_sched;
1851 unsigned long lr_timeout;
1852};
1853
1854struct ext4_features {
1855 struct kobject f_kobj;
1856 struct completion f_kobj_unregister;
1857};
1858
1859/*
1860 * This structure will be used for multiple mount protection. It will be
1861 * written into the block number saved in the s_mmp_block field in the
1862 * superblock. Programs that check MMP should assume that if
1863 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1864 * to use the filesystem, regardless of how old the timestamp is.
1865 */
1866#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1867#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1868#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1869#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1870
1871struct mmp_struct {
1872 __le32 mmp_magic; /* Magic number for MMP */
1873 __le32 mmp_seq; /* Sequence no. updated periodically */
1874
1875 /*
1876 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1877 * purposes and do not affect the correctness of the algorithm
1878 */
1879 __le64 mmp_time; /* Time last updated */
1880 char mmp_nodename[64]; /* Node which last updated MMP block */
1881 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1882
1883 /*
1884 * mmp_check_interval is used to verify if the MMP block has been
1885 * updated on the block device. The value is updated based on the
1886 * maximum time to write the MMP block during an update cycle.
1887 */
1888 __le16 mmp_check_interval;
1889
1890 __le16 mmp_pad1;
1891 __le32 mmp_pad2[226];
1892 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1893};
1894
1895/* arguments passed to the mmp thread */
1896struct mmpd_data {
1897 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1898 struct super_block *sb; /* super block of the fs */
1899};
1900
1901/*
1902 * Check interval multiplier
1903 * The MMP block is written every update interval and initially checked every
1904 * update interval x the multiplier (the value is then adapted based on the
1905 * write latency). The reason is that writes can be delayed under load and we
1906 * don't want readers to incorrectly assume that the filesystem is no longer
1907 * in use.
1908 */
1909#define EXT4_MMP_CHECK_MULT 2UL
1910
1911/*
1912 * Minimum interval for MMP checking in seconds.
1913 */
1914#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1915
1916/*
1917 * Maximum interval for MMP checking in seconds.
1918 */
1919#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1920
1921/*
1922 * Function prototypes
1923 */
1924
1925/*
1926 * Ok, these declarations are also in <linux/kernel.h> but none of the
1927 * ext4 source programs needs to include it so they are duplicated here.
1928 */
1929# define NORET_TYPE /**/
1930# define ATTRIB_NORET __attribute__((noreturn))
1931# define NORET_AND noreturn,
1932
1933/* bitmap.c */
1934extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1935void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1936 struct ext4_group_desc *gdp,
1937 struct buffer_head *bh, int sz);
1938int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1939 struct ext4_group_desc *gdp,
1940 struct buffer_head *bh, int sz);
1941void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1942 struct ext4_group_desc *gdp,
1943 struct buffer_head *bh);
1944int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1945 struct ext4_group_desc *gdp,
1946 struct buffer_head *bh);
1947
1948/* balloc.c */
1949extern void ext4_get_group_no_and_offset(struct super_block *sb,
1950 ext4_fsblk_t blocknr,
1951 ext4_group_t *blockgrpp,
1952 ext4_grpblk_t *offsetp);
1953extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1954 ext4_fsblk_t block);
1955
1956extern unsigned int ext4_block_group(struct super_block *sb,
1957 ext4_fsblk_t blocknr);
1958extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1959 ext4_fsblk_t blocknr);
1960extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1961extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1962 ext4_group_t group);
1963extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1964 ext4_fsblk_t goal,
1965 unsigned int flags,
1966 unsigned long *count,
1967 int *errp);
1968extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1969 s64 nclusters, unsigned int flags);
1970extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1971extern void ext4_check_blocks_bitmap(struct super_block *);
1972extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1973 ext4_group_t block_group,
1974 struct buffer_head ** bh);
1975extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1976
1977extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1978 ext4_group_t block_group);
1979extern int ext4_wait_block_bitmap(struct super_block *sb,
1980 ext4_group_t block_group,
1981 struct buffer_head *bh);
1982extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1983 ext4_group_t block_group);
1984extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1985 ext4_group_t block_group,
1986 struct ext4_group_desc *gdp);
1987ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1988
1989/* dir.c */
1990extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1991 struct file *,
1992 struct ext4_dir_entry_2 *,
1993 struct buffer_head *, char *, int,
1994 unsigned int);
1995#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1996 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1997 (de), (bh), (buf), (size), (offset)))
1998extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1999 __u32 minor_hash,
2000 struct ext4_dir_entry_2 *dirent);
2001extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2002extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2003 struct buffer_head *bh,
2004 void *buf, int buf_size,
2005 const char *name, int namelen,
2006 struct ext4_dir_entry_2 **dest_de);
2007void ext4_insert_dentry(struct inode *inode,
2008 struct ext4_dir_entry_2 *de,
2009 int buf_size,
2010 const char *name, int namelen);
2011static inline void ext4_update_dx_flag(struct inode *inode)
2012{
2013 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2014 EXT4_FEATURE_COMPAT_DIR_INDEX))
2015 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2016}
2017static unsigned char ext4_filetype_table[] = {
2018 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2019};
2020
2021static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2022{
2023 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2024 (filetype >= EXT4_FT_MAX))
2025 return DT_UNKNOWN;
2026
2027 return ext4_filetype_table[filetype];
2028}
2029extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2030 void *buf, int buf_size);
2031
2032/* fsync.c */
2033extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2034
2035/* hash.c */
2036extern int ext4fs_dirhash(const char *name, int len, struct
2037 dx_hash_info *hinfo);
2038
2039/* ialloc.c */
2040extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2041 const struct qstr *qstr, __u32 goal,
2042 uid_t *owner, int handle_type,
2043 unsigned int line_no, int nblocks);
2044
2045#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2046 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2047 0, 0, 0)
2048#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2049 type, nblocks) \
2050 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2051 (type), __LINE__, (nblocks))
2052
2053
2054extern void ext4_free_inode(handle_t *, struct inode *);
2055extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2056extern unsigned long ext4_count_free_inodes(struct super_block *);
2057extern unsigned long ext4_count_dirs(struct super_block *);
2058extern void ext4_check_inodes_bitmap(struct super_block *);
2059extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2060extern int ext4_init_inode_table(struct super_block *sb,
2061 ext4_group_t group, int barrier);
2062extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2063
2064/* mballoc.c */
2065extern long ext4_mb_stats;
2066extern long ext4_mb_max_to_scan;
2067extern int ext4_mb_init(struct super_block *);
2068extern int ext4_mb_release(struct super_block *);
2069extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2070 struct ext4_allocation_request *, int *);
2071extern int ext4_mb_reserve_blocks(struct super_block *, int);
2072extern void ext4_discard_preallocations(struct inode *);
2073extern int __init ext4_init_mballoc(void);
2074extern void ext4_exit_mballoc(void);
2075extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2076 struct buffer_head *bh, ext4_fsblk_t block,
2077 unsigned long count, int flags);
2078extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2079 ext4_group_t ngroups);
2080extern int ext4_mb_add_groupinfo(struct super_block *sb,
2081 ext4_group_t i, struct ext4_group_desc *desc);
2082extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2083 ext4_fsblk_t block, unsigned long count);
2084extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2085
2086/* inode.c */
2087struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2088struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2089int ext4_get_block_write(struct inode *inode, sector_t iblock,
2090 struct buffer_head *bh_result, int create);
2091int ext4_get_block(struct inode *inode, sector_t iblock,
2092 struct buffer_head *bh_result, int create);
2093int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2094 struct buffer_head *bh, int create);
2095int ext4_walk_page_buffers(handle_t *handle,
2096 struct buffer_head *head,
2097 unsigned from,
2098 unsigned to,
2099 int *partial,
2100 int (*fn)(handle_t *handle,
2101 struct buffer_head *bh));
2102int do_journal_get_write_access(handle_t *handle,
2103 struct buffer_head *bh);
2104#define FALL_BACK_TO_NONDELALLOC 1
2105#define CONVERT_INLINE_DATA 2
2106
2107extern struct inode *ext4_iget(struct super_block *, unsigned long);
2108extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2109extern int ext4_write_inode(struct inode *, struct writeback_control *);
2110extern int ext4_setattr(struct dentry *, struct iattr *);
2111extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2112 struct kstat *stat);
2113extern void ext4_evict_inode(struct inode *);
2114extern void ext4_clear_inode(struct inode *);
2115extern int ext4_sync_inode(handle_t *, struct inode *);
2116extern void ext4_dirty_inode(struct inode *, int);
2117extern int ext4_change_inode_journal_flag(struct inode *, int);
2118extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2119extern int ext4_inode_attach_jinode(struct inode *inode);
2120extern int ext4_can_truncate(struct inode *inode);
2121extern void ext4_truncate(struct inode *);
2122extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2123extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2124extern void ext4_set_inode_flags(struct inode *);
2125extern void ext4_get_inode_flags(struct ext4_inode_info *);
2126extern int ext4_alloc_da_blocks(struct inode *inode);
2127extern void ext4_set_aops(struct inode *inode);
2128extern int ext4_writepage_trans_blocks(struct inode *);
2129extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2130extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2131 loff_t lstart, loff_t lend);
2132extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2133extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2134extern void ext4_da_update_reserve_space(struct inode *inode,
2135 int used, int quota_claim);
2136
2137/* indirect.c */
2138extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2139 struct ext4_map_blocks *map, int flags);
2140extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2141 struct iov_iter *iter, loff_t offset);
2142extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2143extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2144extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2145extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2146 ext4_lblk_t start, ext4_lblk_t end);
2147
2148/* ioctl.c */
2149extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2150extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2151
2152/* migrate.c */
2153extern int ext4_ext_migrate(struct inode *);
2154extern int ext4_ind_migrate(struct inode *inode);
2155
2156/* namei.c */
2157extern int ext4_dirent_csum_verify(struct inode *inode,
2158 struct ext4_dir_entry *dirent);
2159extern int ext4_orphan_add(handle_t *, struct inode *);
2160extern int ext4_orphan_del(handle_t *, struct inode *);
2161extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2162 __u32 start_minor_hash, __u32 *next_hash);
2163extern int search_dir(struct buffer_head *bh,
2164 char *search_buf,
2165 int buf_size,
2166 struct inode *dir,
2167 const struct qstr *d_name,
2168 unsigned int offset,
2169 struct ext4_dir_entry_2 **res_dir);
2170extern int ext4_generic_delete_entry(handle_t *handle,
2171 struct inode *dir,
2172 struct ext4_dir_entry_2 *de_del,
2173 struct buffer_head *bh,
2174 void *entry_buf,
2175 int buf_size,
2176 int csum_size);
2177
2178/* resize.c */
2179extern int ext4_group_add(struct super_block *sb,
2180 struct ext4_new_group_data *input);
2181extern int ext4_group_extend(struct super_block *sb,
2182 struct ext4_super_block *es,
2183 ext4_fsblk_t n_blocks_count);
2184extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2185
2186/* super.c */
2187extern int ext4_calculate_overhead(struct super_block *sb);
2188extern void ext4_superblock_csum_set(struct super_block *sb);
2189extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2190extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2191extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2192 ext4_group_t ngroup);
2193extern const char *ext4_decode_error(struct super_block *sb, int errno,
2194 char nbuf[16]);
2195
2196extern __printf(4, 5)
2197void __ext4_error(struct super_block *, const char *, unsigned int,
2198 const char *, ...);
2199extern __printf(5, 6)
2200void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2201 const char *, ...);
2202extern __printf(5, 6)
2203void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2204 const char *, ...);
2205extern void __ext4_std_error(struct super_block *, const char *,
2206 unsigned int, int);
2207extern __printf(4, 5)
2208void __ext4_abort(struct super_block *, const char *, unsigned int,
2209 const char *, ...);
2210extern __printf(4, 5)
2211void __ext4_warning(struct super_block *, const char *, unsigned int,
2212 const char *, ...);
2213extern __printf(3, 4)
2214void __ext4_msg(struct super_block *, const char *, const char *, ...);
2215extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2216 const char *, unsigned int, const char *);
2217extern __printf(7, 8)
2218void __ext4_grp_locked_error(const char *, unsigned int,
2219 struct super_block *, ext4_group_t,
2220 unsigned long, ext4_fsblk_t,
2221 const char *, ...);
2222
2223#ifdef CONFIG_PRINTK
2224
2225#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2226 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2227#define ext4_error_file(file, func, line, block, fmt, ...) \
2228 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2229#define ext4_error(sb, fmt, ...) \
2230 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2231#define ext4_abort(sb, fmt, ...) \
2232 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2233#define ext4_warning(sb, fmt, ...) \
2234 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2235#define ext4_msg(sb, level, fmt, ...) \
2236 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2237#define dump_mmp_msg(sb, mmp, msg) \
2238 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2239#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2240 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2241 fmt, ##__VA_ARGS__)
2242
2243#else
2244
2245#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2246do { \
2247 no_printk(fmt, ##__VA_ARGS__); \
2248 __ext4_error_inode(inode, "", 0, block, " "); \
2249} while (0)
2250#define ext4_error_file(file, func, line, block, fmt, ...) \
2251do { \
2252 no_printk(fmt, ##__VA_ARGS__); \
2253 __ext4_error_file(file, "", 0, block, " "); \
2254} while (0)
2255#define ext4_error(sb, fmt, ...) \
2256do { \
2257 no_printk(fmt, ##__VA_ARGS__); \
2258 __ext4_error(sb, "", 0, " "); \
2259} while (0)
2260#define ext4_abort(sb, fmt, ...) \
2261do { \
2262 no_printk(fmt, ##__VA_ARGS__); \
2263 __ext4_abort(sb, "", 0, " "); \
2264} while (0)
2265#define ext4_warning(sb, fmt, ...) \
2266do { \
2267 no_printk(fmt, ##__VA_ARGS__); \
2268 __ext4_warning(sb, "", 0, " "); \
2269} while (0)
2270#define ext4_msg(sb, level, fmt, ...) \
2271do { \
2272 no_printk(fmt, ##__VA_ARGS__); \
2273 __ext4_msg(sb, "", " "); \
2274} while (0)
2275#define dump_mmp_msg(sb, mmp, msg) \
2276 __dump_mmp_msg(sb, mmp, "", 0, "")
2277#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2278do { \
2279 no_printk(fmt, ##__VA_ARGS__); \
2280 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2281} while (0)
2282
2283#endif
2284
2285extern void ext4_update_dynamic_rev(struct super_block *sb);
2286extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2287 __u32 compat);
2288extern int ext4_update_rocompat_feature(handle_t *handle,
2289 struct super_block *sb, __u32 rocompat);
2290extern int ext4_update_incompat_feature(handle_t *handle,
2291 struct super_block *sb, __u32 incompat);
2292extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2293 struct ext4_group_desc *bg);
2294extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2295 struct ext4_group_desc *bg);
2296extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2297 struct ext4_group_desc *bg);
2298extern __u32 ext4_free_group_clusters(struct super_block *sb,
2299 struct ext4_group_desc *bg);
2300extern __u32 ext4_free_inodes_count(struct super_block *sb,
2301 struct ext4_group_desc *bg);
2302extern __u32 ext4_used_dirs_count(struct super_block *sb,
2303 struct ext4_group_desc *bg);
2304extern __u32 ext4_itable_unused_count(struct super_block *sb,
2305 struct ext4_group_desc *bg);
2306extern void ext4_block_bitmap_set(struct super_block *sb,
2307 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2308extern void ext4_inode_bitmap_set(struct super_block *sb,
2309 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2310extern void ext4_inode_table_set(struct super_block *sb,
2311 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2312extern void ext4_free_group_clusters_set(struct super_block *sb,
2313 struct ext4_group_desc *bg,
2314 __u32 count);
2315extern void ext4_free_inodes_set(struct super_block *sb,
2316 struct ext4_group_desc *bg, __u32 count);
2317extern void ext4_used_dirs_set(struct super_block *sb,
2318 struct ext4_group_desc *bg, __u32 count);
2319extern void ext4_itable_unused_set(struct super_block *sb,
2320 struct ext4_group_desc *bg, __u32 count);
2321extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2322 struct ext4_group_desc *gdp);
2323extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2324 struct ext4_group_desc *gdp);
2325extern int ext4_register_li_request(struct super_block *sb,
2326 ext4_group_t first_not_zeroed);
2327
2328static inline int ext4_has_group_desc_csum(struct super_block *sb)
2329{
2330 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2331 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2332 (EXT4_SB(sb)->s_chksum_driver != NULL);
2333}
2334
2335static inline int ext4_has_metadata_csum(struct super_block *sb)
2336{
2337 WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb,
2338 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) &&
2339 !EXT4_SB(sb)->s_chksum_driver);
2340
2341 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2342}
2343static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2344{
2345 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2346 le32_to_cpu(es->s_blocks_count_lo);
2347}
2348
2349static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2350{
2351 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2352 le32_to_cpu(es->s_r_blocks_count_lo);
2353}
2354
2355static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2356{
2357 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2358 le32_to_cpu(es->s_free_blocks_count_lo);
2359}
2360
2361static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2362 ext4_fsblk_t blk)
2363{
2364 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2365 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2366}
2367
2368static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2369 ext4_fsblk_t blk)
2370{
2371 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2372 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2373}
2374
2375static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2376 ext4_fsblk_t blk)
2377{
2378 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2379 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2380}
2381
2382static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2383{
2384 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2385 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2386 le32_to_cpu(raw_inode->i_size_lo);
2387 else
2388 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2389}
2390
2391static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2392{
2393 raw_inode->i_size_lo = cpu_to_le32(i_size);
2394 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2395}
2396
2397static inline
2398struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2399 ext4_group_t group)
2400{
2401 struct ext4_group_info ***grp_info;
2402 long indexv, indexh;
2403 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2404 grp_info = EXT4_SB(sb)->s_group_info;
2405 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2406 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2407 return grp_info[indexv][indexh];
2408}
2409
2410/*
2411 * Reading s_groups_count requires using smp_rmb() afterwards. See
2412 * the locking protocol documented in the comments of ext4_group_add()
2413 * in resize.c
2414 */
2415static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2416{
2417 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2418
2419 smp_rmb();
2420 return ngroups;
2421}
2422
2423static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2424 ext4_group_t block_group)
2425{
2426 return block_group >> sbi->s_log_groups_per_flex;
2427}
2428
2429static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2430{
2431 return 1 << sbi->s_log_groups_per_flex;
2432}
2433
2434#define ext4_std_error(sb, errno) \
2435do { \
2436 if ((errno)) \
2437 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2438} while (0)
2439
2440#ifdef CONFIG_SMP
2441/* Each CPU can accumulate percpu_counter_batch clusters in their local
2442 * counters. So we need to make sure we have free clusters more
2443 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2444 */
2445#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2446#else
2447#define EXT4_FREECLUSTERS_WATERMARK 0
2448#endif
2449
2450/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2451static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2452{
2453 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2454 !mutex_is_locked(&inode->i_mutex));
2455 down_write(&EXT4_I(inode)->i_data_sem);
2456 if (newsize > EXT4_I(inode)->i_disksize)
2457 EXT4_I(inode)->i_disksize = newsize;
2458 up_write(&EXT4_I(inode)->i_data_sem);
2459}
2460
2461/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2462static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2463{
2464 int changed = 0;
2465
2466 if (newsize > inode->i_size) {
2467 i_size_write(inode, newsize);
2468 changed = 1;
2469 }
2470 if (newsize > EXT4_I(inode)->i_disksize) {
2471 ext4_update_i_disksize(inode, newsize);
2472 changed |= 2;
2473 }
2474 return changed;
2475}
2476
2477struct ext4_group_info {
2478 unsigned long bb_state;
2479 struct rb_root bb_free_root;
2480 ext4_grpblk_t bb_first_free; /* first free block */
2481 ext4_grpblk_t bb_free; /* total free blocks */
2482 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2483 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2484 struct list_head bb_prealloc_list;
2485#ifdef DOUBLE_CHECK
2486 void *bb_bitmap;
2487#endif
2488 struct rw_semaphore alloc_sem;
2489 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2490 * regions, index is order.
2491 * bb_counters[3] = 5 means
2492 * 5 free 8-block regions. */
2493};
2494
2495#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2496#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2497#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2498#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2499
2500#define EXT4_MB_GRP_NEED_INIT(grp) \
2501 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2502#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2503 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2504#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2505 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2506
2507#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2508 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2509#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2510 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2511#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2512 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2513
2514#define EXT4_MAX_CONTENTION 8
2515#define EXT4_CONTENTION_THRESHOLD 2
2516
2517static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2518 ext4_group_t group)
2519{
2520 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2521}
2522
2523/*
2524 * Returns true if the filesystem is busy enough that attempts to
2525 * access the block group locks has run into contention.
2526 */
2527static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2528{
2529 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2530}
2531
2532static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2533{
2534 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2535 if (spin_trylock(lock))
2536 /*
2537 * We're able to grab the lock right away, so drop the
2538 * lock contention counter.
2539 */
2540 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2541 else {
2542 /*
2543 * The lock is busy, so bump the contention counter,
2544 * and then wait on the spin lock.
2545 */
2546 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2547 EXT4_MAX_CONTENTION);
2548 spin_lock(lock);
2549 }
2550}
2551
2552static inline void ext4_unlock_group(struct super_block *sb,
2553 ext4_group_t group)
2554{
2555 spin_unlock(ext4_group_lock_ptr(sb, group));
2556}
2557
2558/*
2559 * Block validity checking
2560 */
2561#define ext4_check_indirect_blockref(inode, bh) \
2562 ext4_check_blockref(__func__, __LINE__, inode, \
2563 (__le32 *)(bh)->b_data, \
2564 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2565
2566#define ext4_ind_check_inode(inode) \
2567 ext4_check_blockref(__func__, __LINE__, inode, \
2568 EXT4_I(inode)->i_data, \
2569 EXT4_NDIR_BLOCKS)
2570
2571/*
2572 * Inodes and files operations
2573 */
2574
2575/* dir.c */
2576extern const struct file_operations ext4_dir_operations;
2577
2578/* file.c */
2579extern const struct inode_operations ext4_file_inode_operations;
2580extern const struct file_operations ext4_file_operations;
2581extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2582
2583/* inline.c */
2584extern int ext4_get_max_inline_size(struct inode *inode);
2585extern int ext4_find_inline_data_nolock(struct inode *inode);
2586extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2587 unsigned int len);
2588extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2589
2590extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2591extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2592 struct inode *inode,
2593 loff_t pos, unsigned len,
2594 unsigned flags,
2595 struct page **pagep);
2596extern int ext4_write_inline_data_end(struct inode *inode,
2597 loff_t pos, unsigned len,
2598 unsigned copied,
2599 struct page *page);
2600extern struct buffer_head *
2601ext4_journalled_write_inline_data(struct inode *inode,
2602 unsigned len,
2603 struct page *page);
2604extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2605 struct inode *inode,
2606 loff_t pos, unsigned len,
2607 unsigned flags,
2608 struct page **pagep,
2609 void **fsdata);
2610extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2611 unsigned len, unsigned copied,
2612 struct page *page);
2613extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2614 struct inode *inode);
2615extern int ext4_try_create_inline_dir(handle_t *handle,
2616 struct inode *parent,
2617 struct inode *inode);
2618extern int ext4_read_inline_dir(struct file *filp,
2619 struct dir_context *ctx,
2620 int *has_inline_data);
2621extern int htree_inlinedir_to_tree(struct file *dir_file,
2622 struct inode *dir, ext4_lblk_t block,
2623 struct dx_hash_info *hinfo,
2624 __u32 start_hash, __u32 start_minor_hash,
2625 int *has_inline_data);
2626extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2627 const struct qstr *d_name,
2628 struct ext4_dir_entry_2 **res_dir,
2629 int *has_inline_data);
2630extern int ext4_delete_inline_entry(handle_t *handle,
2631 struct inode *dir,
2632 struct ext4_dir_entry_2 *de_del,
2633 struct buffer_head *bh,
2634 int *has_inline_data);
2635extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2636extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2637 struct ext4_dir_entry_2 **parent_de,
2638 int *retval);
2639extern int ext4_inline_data_fiemap(struct inode *inode,
2640 struct fiemap_extent_info *fieinfo,
2641 int *has_inline, __u64 start, __u64 len);
2642extern int ext4_try_to_evict_inline_data(handle_t *handle,
2643 struct inode *inode,
2644 int needed);
2645extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2646
2647extern int ext4_convert_inline_data(struct inode *inode);
2648
2649static inline int ext4_has_inline_data(struct inode *inode)
2650{
2651 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2652 EXT4_I(inode)->i_inline_off;
2653}
2654
2655/* namei.c */
2656extern const struct inode_operations ext4_dir_inode_operations;
2657extern const struct inode_operations ext4_special_inode_operations;
2658extern struct dentry *ext4_get_parent(struct dentry *child);
2659extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2660 struct ext4_dir_entry_2 *de,
2661 int blocksize, int csum_size,
2662 unsigned int parent_ino, int dotdot_real_len);
2663extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2664 unsigned int blocksize);
2665extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2666 struct inode *inode,
2667 struct buffer_head *bh);
2668#define S_SHIFT 12
2669static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2670 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2671 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2672 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2673 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2674 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2675 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2676 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2677};
2678
2679static inline void ext4_set_de_type(struct super_block *sb,
2680 struct ext4_dir_entry_2 *de,
2681 umode_t mode) {
2682 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2683 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2684}
2685
2686
2687/* symlink.c */
2688extern const struct inode_operations ext4_symlink_inode_operations;
2689extern const struct inode_operations ext4_fast_symlink_inode_operations;
2690
2691/* block_validity */
2692extern void ext4_release_system_zone(struct super_block *sb);
2693extern int ext4_setup_system_zone(struct super_block *sb);
2694extern int __init ext4_init_system_zone(void);
2695extern void ext4_exit_system_zone(void);
2696extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2697 ext4_fsblk_t start_blk,
2698 unsigned int count);
2699extern int ext4_check_blockref(const char *, unsigned int,
2700 struct inode *, __le32 *, unsigned int);
2701
2702/* extents.c */
2703struct ext4_ext_path;
2704struct ext4_extent;
2705
2706/*
2707 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2708 * __le32.
2709 */
2710#define EXT_MAX_BLOCKS 0xffffffff
2711
2712extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2713extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2714extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2715extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2716 struct ext4_map_blocks *map, int flags);
2717extern void ext4_ext_truncate(handle_t *, struct inode *);
2718extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2719 ext4_lblk_t end);
2720extern void ext4_ext_init(struct super_block *);
2721extern void ext4_ext_release(struct super_block *);
2722extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2723 loff_t len);
2724extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2725 loff_t offset, ssize_t len);
2726extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2727 struct ext4_map_blocks *map, int flags);
2728extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2729 ext4_lblk_t lblocks);
2730extern int ext4_extent_tree_init(handle_t *, struct inode *);
2731extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2732 int num,
2733 struct ext4_ext_path *path);
2734extern int ext4_can_extents_be_merged(struct inode *inode,
2735 struct ext4_extent *ex1,
2736 struct ext4_extent *ex2);
2737extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2738 struct ext4_ext_path **,
2739 struct ext4_extent *, int);
2740extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
2741 struct ext4_ext_path **,
2742 int flags);
2743extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2744extern int ext4_ext_check_inode(struct inode *inode);
2745extern int ext4_find_delalloc_range(struct inode *inode,
2746 ext4_lblk_t lblk_start,
2747 ext4_lblk_t lblk_end);
2748extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2749extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
2750extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2751 __u64 start, __u64 len);
2752extern int ext4_ext_precache(struct inode *inode);
2753extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
2754extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
2755 struct inode *inode2, ext4_lblk_t lblk1,
2756 ext4_lblk_t lblk2, ext4_lblk_t count,
2757 int mark_unwritten,int *err);
2758
2759/* move_extent.c */
2760extern void ext4_double_down_write_data_sem(struct inode *first,
2761 struct inode *second);
2762extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2763 struct inode *donor_inode);
2764extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2765 __u64 start_orig, __u64 start_donor,
2766 __u64 len, __u64 *moved_len);
2767
2768/* page-io.c */
2769extern int __init ext4_init_pageio(void);
2770extern void ext4_exit_pageio(void);
2771extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2772extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2773extern int ext4_put_io_end(ext4_io_end_t *io_end);
2774extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2775extern void ext4_io_submit_init(struct ext4_io_submit *io,
2776 struct writeback_control *wbc);
2777extern void ext4_end_io_rsv_work(struct work_struct *work);
2778extern void ext4_io_submit(struct ext4_io_submit *io);
2779extern int ext4_bio_write_page(struct ext4_io_submit *io,
2780 struct page *page,
2781 int len,
2782 struct writeback_control *wbc,
2783 bool keep_towrite);
2784
2785/* mmp.c */
2786extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2787
2788/*
2789 * Add new method to test whether block and inode bitmaps are properly
2790 * initialized. With uninit_bg reading the block from disk is not enough
2791 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2792 */
2793#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2794
2795static inline int bitmap_uptodate(struct buffer_head *bh)
2796{
2797 return (buffer_uptodate(bh) &&
2798 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2799}
2800static inline void set_bitmap_uptodate(struct buffer_head *bh)
2801{
2802 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2803}
2804
2805/*
2806 * Disable DIO read nolock optimization, so new dioreaders will be forced
2807 * to grab i_mutex
2808 */
2809static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2810{
2811 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2812 smp_mb();
2813}
2814static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2815{
2816 smp_mb();
2817 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2818}
2819
2820#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2821
2822/* For ioend & aio unwritten conversion wait queues */
2823#define EXT4_WQ_HASH_SZ 37
2824#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2825 EXT4_WQ_HASH_SZ])
2826#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2827 EXT4_WQ_HASH_SZ])
2828extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2829extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2830
2831#define EXT4_RESIZING 0
2832extern int ext4_resize_begin(struct super_block *sb);
2833extern void ext4_resize_end(struct super_block *sb);
2834
2835#endif /* __KERNEL__ */
2836
2837#endif /* _EXT4_H */