1// SPDX-License-Identifier: GPL-2.0
2/*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17#ifndef _EXT4_H
18#define _EXT4_H
19
20#include <linux/types.h>
21#include <linux/blkdev.h>
22#include <linux/magic.h>
23#include <linux/jbd2.h>
24#include <linux/quota.h>
25#include <linux/rwsem.h>
26#include <linux/rbtree.h>
27#include <linux/seqlock.h>
28#include <linux/mutex.h>
29#include <linux/timer.h>
30#include <linux/version.h>
31#include <linux/wait.h>
32#include <linux/sched/signal.h>
33#include <linux/blockgroup_lock.h>
34#include <linux/percpu_counter.h>
35#include <linux/ratelimit.h>
36#include <crypto/hash.h>
37#include <linux/falloc.h>
38#include <linux/percpu-rwsem.h>
39#ifdef __KERNEL__
40#include <linux/compat.h>
41#endif
42
43#include <linux/fscrypt.h>
44
45#include <linux/compiler.h>
46
47/*
48 * The fourth extended filesystem constants/structures
49 */
50
51/*
52 * with AGGRESSIVE_CHECK allocator runs consistency checks over
53 * structures. these checks slow things down a lot
54 */
55#define AGGRESSIVE_CHECK__
56
57/*
58 * with DOUBLE_CHECK defined mballoc creates persistent in-core
59 * bitmaps, maintains and uses them to check for double allocations
60 */
61#define DOUBLE_CHECK__
62
63/*
64 * Define EXT4FS_DEBUG to produce debug messages
65 */
66#undef EXT4FS_DEBUG
67
68/*
69 * Debug code
70 */
71#ifdef EXT4FS_DEBUG
72#define ext4_debug(f, a...) \
73 do { \
74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
75 __FILE__, __LINE__, __func__); \
76 printk(KERN_DEBUG f, ## a); \
77 } while (0)
78#else
79#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
80#endif
81
82/*
83 * Turn on EXT_DEBUG to get lots of info about extents operations.
84 */
85#define EXT_DEBUG__
86#ifdef EXT_DEBUG
87#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
88#else
89#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
90#endif
91
92/* data type for block offset of block group */
93typedef int ext4_grpblk_t;
94
95/* data type for filesystem-wide blocks number */
96typedef unsigned long long ext4_fsblk_t;
97
98/* data type for file logical block number */
99typedef __u32 ext4_lblk_t;
100
101/* data type for block group number */
102typedef unsigned int ext4_group_t;
103
104enum SHIFT_DIRECTION {
105 SHIFT_LEFT = 0,
106 SHIFT_RIGHT,
107};
108
109/*
110 * Flags used in mballoc's allocation_context flags field.
111 *
112 * Also used to show what's going on for debugging purposes when the
113 * flag field is exported via the traceport interface
114 */
115
116/* prefer goal again. length */
117#define EXT4_MB_HINT_MERGE 0x0001
118/* blocks already reserved */
119#define EXT4_MB_HINT_RESERVED 0x0002
120/* metadata is being allocated */
121#define EXT4_MB_HINT_METADATA 0x0004
122/* first blocks in the file */
123#define EXT4_MB_HINT_FIRST 0x0008
124/* search for the best chunk */
125#define EXT4_MB_HINT_BEST 0x0010
126/* data is being allocated */
127#define EXT4_MB_HINT_DATA 0x0020
128/* don't preallocate (for tails) */
129#define EXT4_MB_HINT_NOPREALLOC 0x0040
130/* allocate for locality group */
131#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
132/* allocate goal blocks or none */
133#define EXT4_MB_HINT_GOAL_ONLY 0x0100
134/* goal is meaningful */
135#define EXT4_MB_HINT_TRY_GOAL 0x0200
136/* blocks already pre-reserved by delayed allocation */
137#define EXT4_MB_DELALLOC_RESERVED 0x0400
138/* We are doing stream allocation */
139#define EXT4_MB_STREAM_ALLOC 0x0800
140/* Use reserved root blocks if needed */
141#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
142/* Use blocks from reserved pool */
143#define EXT4_MB_USE_RESERVED 0x2000
144
145struct ext4_allocation_request {
146 /* target inode for block we're allocating */
147 struct inode *inode;
148 /* how many blocks we want to allocate */
149 unsigned int len;
150 /* logical block in target inode */
151 ext4_lblk_t logical;
152 /* the closest logical allocated block to the left */
153 ext4_lblk_t lleft;
154 /* the closest logical allocated block to the right */
155 ext4_lblk_t lright;
156 /* phys. target (a hint) */
157 ext4_fsblk_t goal;
158 /* phys. block for the closest logical allocated block to the left */
159 ext4_fsblk_t pleft;
160 /* phys. block for the closest logical allocated block to the right */
161 ext4_fsblk_t pright;
162 /* flags. see above EXT4_MB_HINT_* */
163 unsigned int flags;
164};
165
166/*
167 * Logical to physical block mapping, used by ext4_map_blocks()
168 *
169 * This structure is used to pass requests into ext4_map_blocks() as
170 * well as to store the information returned by ext4_map_blocks(). It
171 * takes less room on the stack than a struct buffer_head.
172 */
173#define EXT4_MAP_NEW (1 << BH_New)
174#define EXT4_MAP_MAPPED (1 << BH_Mapped)
175#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
176#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
177#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
179
180struct ext4_map_blocks {
181 ext4_fsblk_t m_pblk;
182 ext4_lblk_t m_lblk;
183 unsigned int m_len;
184 unsigned int m_flags;
185};
186
187/*
188 * Flags for ext4_io_end->flags
189 */
190#define EXT4_IO_END_UNWRITTEN 0x0001
191
192/*
193 * For converting unwritten extents on a work queue. 'handle' is used for
194 * buffered writeback.
195 */
196typedef struct ext4_io_end {
197 struct list_head list; /* per-file finished IO list */
198 handle_t *handle; /* handle reserved for extent
199 * conversion */
200 struct inode *inode; /* file being written to */
201 struct bio *bio; /* Linked list of completed
202 * bios covering the extent */
203 unsigned int flag; /* unwritten or not */
204 atomic_t count; /* reference counter */
205 loff_t offset; /* offset in the file */
206 ssize_t size; /* size of the extent */
207} ext4_io_end_t;
208
209struct ext4_io_submit {
210 struct writeback_control *io_wbc;
211 struct bio *io_bio;
212 ext4_io_end_t *io_end;
213 sector_t io_next_block;
214};
215
216/*
217 * Special inodes numbers
218 */
219#define EXT4_BAD_INO 1 /* Bad blocks inode */
220#define EXT4_ROOT_INO 2 /* Root inode */
221#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
222#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
223#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
224#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
225#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
226#define EXT4_JOURNAL_INO 8 /* Journal inode */
227
228/* First non-reserved inode for old ext4 filesystems */
229#define EXT4_GOOD_OLD_FIRST_INO 11
230
231/*
232 * Maximal count of links to a file
233 */
234#define EXT4_LINK_MAX 65000
235
236/*
237 * Macro-instructions used to manage several block sizes
238 */
239#define EXT4_MIN_BLOCK_SIZE 1024
240#define EXT4_MAX_BLOCK_SIZE 65536
241#define EXT4_MIN_BLOCK_LOG_SIZE 10
242#define EXT4_MAX_BLOCK_LOG_SIZE 16
243#define EXT4_MAX_CLUSTER_LOG_SIZE 30
244#ifdef __KERNEL__
245# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
246#else
247# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
248#endif
249#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
250#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
251 EXT4_SB(s)->s_cluster_bits)
252#ifdef __KERNEL__
253# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
254# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
255#else
256# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
257#endif
258#ifdef __KERNEL__
259#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
260#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
261#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
262#else
263#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 EXT4_GOOD_OLD_INODE_SIZE : \
265 (s)->s_inode_size)
266#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 EXT4_GOOD_OLD_FIRST_INO : \
268 (s)->s_first_ino)
269#endif
270#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
271#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
273 blkbits))
274
275/* Translate a block number to a cluster number */
276#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
277/* Translate a cluster number to a block number */
278#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
279/* Translate # of blks to # of clusters */
280#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
281 (sbi)->s_cluster_bits)
282/* Mask out the low bits to get the starting block of the cluster */
283#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
285#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
287/* Get the cluster offset */
288#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
290#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
292
293/*
294 * Structure of a blocks group descriptor
295 */
296struct ext4_group_desc
297{
298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
300 __le32 bg_inode_table_lo; /* Inodes table block */
301 __le16 bg_free_blocks_count_lo;/* Free blocks count */
302 __le16 bg_free_inodes_count_lo;/* Free inodes count */
303 __le16 bg_used_dirs_count_lo; /* Directories count */
304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
308 __le16 bg_itable_unused_lo; /* Unused inodes count */
309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
312 __le32 bg_inode_table_hi; /* Inodes table block MSB */
313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
320 __u32 bg_reserved;
321};
322
323#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
325 sizeof(__le16))
326#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
328 sizeof(__le16))
329
330/*
331 * Structure of a flex block group info
332 */
333
334struct flex_groups {
335 atomic64_t free_clusters;
336 atomic_t free_inodes;
337 atomic_t used_dirs;
338};
339
340#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
341#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
342#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
343
344/*
345 * Macro-instructions used to manage group descriptors
346 */
347#define EXT4_MIN_DESC_SIZE 32
348#define EXT4_MIN_DESC_SIZE_64BIT 64
349#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
350#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
351#ifdef __KERNEL__
352# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
353# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
354# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
355# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
356# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
357#else
358# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
359# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
360# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
361#endif
362
363/*
364 * Constants relative to the data blocks
365 */
366#define EXT4_NDIR_BLOCKS 12
367#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
368#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
369#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
370#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
371
372/*
373 * Inode flags
374 */
375#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
376#define EXT4_UNRM_FL 0x00000002 /* Undelete */
377#define EXT4_COMPR_FL 0x00000004 /* Compress file */
378#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
379#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
380#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
381#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
382#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
383/* Reserved for compression usage... */
384#define EXT4_DIRTY_FL 0x00000100
385#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
386#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
387 /* nb: was previously EXT2_ECOMPR_FL */
388#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
389/* End compression flags --- maybe not all used */
390#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
391#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
392#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
393#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
394#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
395#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
396#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
397#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
398#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
399#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
400#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
401#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
402#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
403
404#define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
405#define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
406
407/* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
408#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
409 EXT4_IMMUTABLE_FL | \
410 EXT4_APPEND_FL | \
411 EXT4_NODUMP_FL | \
412 EXT4_NOATIME_FL | \
413 EXT4_PROJINHERIT_FL)
414
415/* Flags that should be inherited by new inodes from their parent. */
416#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
417 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
418 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
419 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
420 EXT4_PROJINHERIT_FL)
421
422/* Flags that are appropriate for regular files (all but dir-specific ones). */
423#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
424
425/* Flags that are appropriate for non-directories/regular files. */
426#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
427
428/* The only flags that should be swapped */
429#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
430
431/* Mask out flags that are inappropriate for the given type of inode. */
432static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
433{
434 if (S_ISDIR(mode))
435 return flags;
436 else if (S_ISREG(mode))
437 return flags & EXT4_REG_FLMASK;
438 else
439 return flags & EXT4_OTHER_FLMASK;
440}
441
442/*
443 * Inode flags used for atomic set/get
444 */
445enum {
446 EXT4_INODE_SECRM = 0, /* Secure deletion */
447 EXT4_INODE_UNRM = 1, /* Undelete */
448 EXT4_INODE_COMPR = 2, /* Compress file */
449 EXT4_INODE_SYNC = 3, /* Synchronous updates */
450 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
451 EXT4_INODE_APPEND = 5, /* writes to file may only append */
452 EXT4_INODE_NODUMP = 6, /* do not dump file */
453 EXT4_INODE_NOATIME = 7, /* do not update atime */
454/* Reserved for compression usage... */
455 EXT4_INODE_DIRTY = 8,
456 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
457 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
458 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
459/* End compression flags --- maybe not all used */
460 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
461 EXT4_INODE_IMAGIC = 13, /* AFS directory */
462 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
463 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
464 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
465 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
466 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
467 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
468 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
469 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
470 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
471 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
472 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
473};
474
475/*
476 * Since it's pretty easy to mix up bit numbers and hex values, we use a
477 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
478 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
479 * any extra space in the compiled kernel image, otherwise, the build will fail.
480 * It's important that these values are the same, since we are using
481 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
482 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
483 * values found in ext2, ext3 and ext4 filesystems, and of course the values
484 * defined in e2fsprogs.
485 *
486 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
487 */
488#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
489#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
490
491static inline void ext4_check_flag_values(void)
492{
493 CHECK_FLAG_VALUE(SECRM);
494 CHECK_FLAG_VALUE(UNRM);
495 CHECK_FLAG_VALUE(COMPR);
496 CHECK_FLAG_VALUE(SYNC);
497 CHECK_FLAG_VALUE(IMMUTABLE);
498 CHECK_FLAG_VALUE(APPEND);
499 CHECK_FLAG_VALUE(NODUMP);
500 CHECK_FLAG_VALUE(NOATIME);
501 CHECK_FLAG_VALUE(DIRTY);
502 CHECK_FLAG_VALUE(COMPRBLK);
503 CHECK_FLAG_VALUE(NOCOMPR);
504 CHECK_FLAG_VALUE(ENCRYPT);
505 CHECK_FLAG_VALUE(INDEX);
506 CHECK_FLAG_VALUE(IMAGIC);
507 CHECK_FLAG_VALUE(JOURNAL_DATA);
508 CHECK_FLAG_VALUE(NOTAIL);
509 CHECK_FLAG_VALUE(DIRSYNC);
510 CHECK_FLAG_VALUE(TOPDIR);
511 CHECK_FLAG_VALUE(HUGE_FILE);
512 CHECK_FLAG_VALUE(EXTENTS);
513 CHECK_FLAG_VALUE(EA_INODE);
514 CHECK_FLAG_VALUE(EOFBLOCKS);
515 CHECK_FLAG_VALUE(INLINE_DATA);
516 CHECK_FLAG_VALUE(PROJINHERIT);
517 CHECK_FLAG_VALUE(RESERVED);
518}
519
520/* Used to pass group descriptor data when online resize is done */
521struct ext4_new_group_input {
522 __u32 group; /* Group number for this data */
523 __u64 block_bitmap; /* Absolute block number of block bitmap */
524 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
525 __u64 inode_table; /* Absolute block number of inode table start */
526 __u32 blocks_count; /* Total number of blocks in this group */
527 __u16 reserved_blocks; /* Number of reserved blocks in this group */
528 __u16 unused;
529};
530
531#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
532struct compat_ext4_new_group_input {
533 u32 group;
534 compat_u64 block_bitmap;
535 compat_u64 inode_bitmap;
536 compat_u64 inode_table;
537 u32 blocks_count;
538 u16 reserved_blocks;
539 u16 unused;
540};
541#endif
542
543/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
544struct ext4_new_group_data {
545 __u32 group;
546 __u64 block_bitmap;
547 __u64 inode_bitmap;
548 __u64 inode_table;
549 __u32 blocks_count;
550 __u16 reserved_blocks;
551 __u16 mdata_blocks;
552 __u32 free_clusters_count;
553};
554
555/* Indexes used to index group tables in ext4_new_group_data */
556enum {
557 BLOCK_BITMAP = 0, /* block bitmap */
558 INODE_BITMAP, /* inode bitmap */
559 INODE_TABLE, /* inode tables */
560 GROUP_TABLE_COUNT,
561};
562
563/*
564 * Flags used by ext4_map_blocks()
565 */
566 /* Allocate any needed blocks and/or convert an unwritten
567 extent to be an initialized ext4 */
568#define EXT4_GET_BLOCKS_CREATE 0x0001
569 /* Request the creation of an unwritten extent */
570#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
571#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
572 EXT4_GET_BLOCKS_CREATE)
573 /* Caller is from the delayed allocation writeout path
574 * finally doing the actual allocation of delayed blocks */
575#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
576 /* caller is from the direct IO path, request to creation of an
577 unwritten extents if not allocated, split the unwritten
578 extent if blocks has been preallocated already*/
579#define EXT4_GET_BLOCKS_PRE_IO 0x0008
580#define EXT4_GET_BLOCKS_CONVERT 0x0010
581#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
582 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
583 /* Convert extent to initialized after IO complete */
584#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
585 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
586 /* Eventual metadata allocation (due to growing extent tree)
587 * should not fail, so try to use reserved blocks for that.*/
588#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
589 /* Don't normalize allocation size (used for fallocate) */
590#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
591 /* Request will not result in inode size update (user for fallocate) */
592#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
593 /* Convert written extents to unwritten */
594#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
595 /* Write zeros to newly created written extents */
596#define EXT4_GET_BLOCKS_ZERO 0x0200
597#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
598 EXT4_GET_BLOCKS_ZERO)
599 /* Caller will submit data before dropping transaction handle. This
600 * allows jbd2 to avoid submitting data before commit. */
601#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
602
603/*
604 * The bit position of these flags must not overlap with any of the
605 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
606 * read_extent_tree_block(), ext4_split_extent_at(),
607 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
608 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
609 * caching the extents when reading from the extent tree while a
610 * truncate or punch hole operation is in progress.
611 */
612#define EXT4_EX_NOCACHE 0x40000000
613#define EXT4_EX_FORCE_CACHE 0x20000000
614
615/*
616 * Flags used by ext4_free_blocks
617 */
618#define EXT4_FREE_BLOCKS_METADATA 0x0001
619#define EXT4_FREE_BLOCKS_FORGET 0x0002
620#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
621#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
622#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
623#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
624#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
625
626/*
627 * ioctl commands
628 */
629#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
630#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
631#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
632#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
633#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
634#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
635#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
636#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
637#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
638#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
639#define EXT4_IOC_MIGRATE _IO('f', 9)
640 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
641 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
642#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
643#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
644#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
645#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
646#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
647#define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
648#define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
649#define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
650
651#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
652#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
653
654#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
655
656/*
657 * Flags for going down operation
658 */
659#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
660#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
661#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
662
663
664#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
665/*
666 * ioctl commands in 32 bit emulation
667 */
668#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
669#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
670#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
671#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
672#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
673#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
674#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
675#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
676#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
677#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
678#endif
679
680/* Max physical block we can address w/o extents */
681#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
682
683/* Max logical block we can support */
684#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
685
686/*
687 * Structure of an inode on the disk
688 */
689struct ext4_inode {
690 __le16 i_mode; /* File mode */
691 __le16 i_uid; /* Low 16 bits of Owner Uid */
692 __le32 i_size_lo; /* Size in bytes */
693 __le32 i_atime; /* Access time */
694 __le32 i_ctime; /* Inode Change time */
695 __le32 i_mtime; /* Modification time */
696 __le32 i_dtime; /* Deletion Time */
697 __le16 i_gid; /* Low 16 bits of Group Id */
698 __le16 i_links_count; /* Links count */
699 __le32 i_blocks_lo; /* Blocks count */
700 __le32 i_flags; /* File flags */
701 union {
702 struct {
703 __le32 l_i_version;
704 } linux1;
705 struct {
706 __u32 h_i_translator;
707 } hurd1;
708 struct {
709 __u32 m_i_reserved1;
710 } masix1;
711 } osd1; /* OS dependent 1 */
712 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
713 __le32 i_generation; /* File version (for NFS) */
714 __le32 i_file_acl_lo; /* File ACL */
715 __le32 i_size_high;
716 __le32 i_obso_faddr; /* Obsoleted fragment address */
717 union {
718 struct {
719 __le16 l_i_blocks_high; /* were l_i_reserved1 */
720 __le16 l_i_file_acl_high;
721 __le16 l_i_uid_high; /* these 2 fields */
722 __le16 l_i_gid_high; /* were reserved2[0] */
723 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
724 __le16 l_i_reserved;
725 } linux2;
726 struct {
727 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
728 __u16 h_i_mode_high;
729 __u16 h_i_uid_high;
730 __u16 h_i_gid_high;
731 __u32 h_i_author;
732 } hurd2;
733 struct {
734 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
735 __le16 m_i_file_acl_high;
736 __u32 m_i_reserved2[2];
737 } masix2;
738 } osd2; /* OS dependent 2 */
739 __le16 i_extra_isize;
740 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
741 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
742 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
743 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
744 __le32 i_crtime; /* File Creation time */
745 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
746 __le32 i_version_hi; /* high 32 bits for 64-bit version */
747 __le32 i_projid; /* Project ID */
748};
749
750struct move_extent {
751 __u32 reserved; /* should be zero */
752 __u32 donor_fd; /* donor file descriptor */
753 __u64 orig_start; /* logical start offset in block for orig */
754 __u64 donor_start; /* logical start offset in block for donor */
755 __u64 len; /* block length to be moved */
756 __u64 moved_len; /* moved block length */
757};
758
759#define EXT4_EPOCH_BITS 2
760#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
761#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
762
763/*
764 * Extended fields will fit into an inode if the filesystem was formatted
765 * with large inodes (-I 256 or larger) and there are not currently any EAs
766 * consuming all of the available space. For new inodes we always reserve
767 * enough space for the kernel's known extended fields, but for inodes
768 * created with an old kernel this might not have been the case. None of
769 * the extended inode fields is critical for correct filesystem operation.
770 * This macro checks if a certain field fits in the inode. Note that
771 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
772 */
773#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
774 ((offsetof(typeof(*ext4_inode), field) + \
775 sizeof((ext4_inode)->field)) \
776 <= (EXT4_GOOD_OLD_INODE_SIZE + \
777 (einode)->i_extra_isize)) \
778
779/*
780 * We use an encoding that preserves the times for extra epoch "00":
781 *
782 * extra msb of adjust for signed
783 * epoch 32-bit 32-bit tv_sec to
784 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
785 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
786 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
787 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
788 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
789 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
790 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
791 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
792 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
793 *
794 * Note that previous versions of the kernel on 64-bit systems would
795 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
796 * 1970. e2fsck will correct this, assuming that it is run on the
797 * affected filesystem before 2242.
798 */
799
800static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
801{
802 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
803 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
804}
805
806static inline void ext4_decode_extra_time(struct timespec64 *time,
807 __le32 extra)
808{
809 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) {
810
811#if 1
812 /* Handle legacy encoding of pre-1970 dates with epoch
813 * bits 1,1. (This backwards compatibility may be removed
814 * at the discretion of the ext4 developers.)
815 */
816 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
817 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
818 extra_bits = 0;
819 time->tv_sec += extra_bits << 32;
820#else
821 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
822#endif
823 }
824 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
825}
826
827#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
828do { \
829 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
830 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
831 (raw_inode)->xtime ## _extra = \
832 ext4_encode_extra_time(&(inode)->xtime); \
833 } \
834} while (0)
835
836#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
837do { \
838 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
839 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
840 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
841 (raw_inode)->xtime ## _extra = \
842 ext4_encode_extra_time(&(einode)->xtime); \
843} while (0)
844
845#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
846do { \
847 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
848 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
849 ext4_decode_extra_time(&(inode)->xtime, \
850 raw_inode->xtime ## _extra); \
851 } \
852 else \
853 (inode)->xtime.tv_nsec = 0; \
854} while (0)
855
856
857#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
858do { \
859 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
860 (einode)->xtime.tv_sec = \
861 (signed)le32_to_cpu((raw_inode)->xtime); \
862 else \
863 (einode)->xtime.tv_sec = 0; \
864 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
865 ext4_decode_extra_time(&(einode)->xtime, \
866 raw_inode->xtime ## _extra); \
867 else \
868 (einode)->xtime.tv_nsec = 0; \
869} while (0)
870
871#define i_disk_version osd1.linux1.l_i_version
872
873#if defined(__KERNEL__) || defined(__linux__)
874#define i_reserved1 osd1.linux1.l_i_reserved1
875#define i_file_acl_high osd2.linux2.l_i_file_acl_high
876#define i_blocks_high osd2.linux2.l_i_blocks_high
877#define i_uid_low i_uid
878#define i_gid_low i_gid
879#define i_uid_high osd2.linux2.l_i_uid_high
880#define i_gid_high osd2.linux2.l_i_gid_high
881#define i_checksum_lo osd2.linux2.l_i_checksum_lo
882
883#elif defined(__GNU__)
884
885#define i_translator osd1.hurd1.h_i_translator
886#define i_uid_high osd2.hurd2.h_i_uid_high
887#define i_gid_high osd2.hurd2.h_i_gid_high
888#define i_author osd2.hurd2.h_i_author
889
890#elif defined(__masix__)
891
892#define i_reserved1 osd1.masix1.m_i_reserved1
893#define i_file_acl_high osd2.masix2.m_i_file_acl_high
894#define i_reserved2 osd2.masix2.m_i_reserved2
895
896#endif /* defined(__KERNEL__) || defined(__linux__) */
897
898#include "extents_status.h"
899
900/*
901 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
902 *
903 * These are needed to avoid lockdep false positives when we need to
904 * allocate blocks to the quota inode during ext4_map_blocks(), while
905 * holding i_data_sem for a normal (non-quota) inode. Since we don't
906 * do quota tracking for the quota inode, this avoids deadlock (as
907 * well as infinite recursion, since it isn't turtles all the way
908 * down...)
909 *
910 * I_DATA_SEM_NORMAL - Used for most inodes
911 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
912 * where the second inode has larger inode number
913 * than the first
914 * I_DATA_SEM_QUOTA - Used for quota inodes only
915 */
916enum {
917 I_DATA_SEM_NORMAL = 0,
918 I_DATA_SEM_OTHER,
919 I_DATA_SEM_QUOTA,
920};
921
922
923/*
924 * fourth extended file system inode data in memory
925 */
926struct ext4_inode_info {
927 __le32 i_data[15]; /* unconverted */
928 __u32 i_dtime;
929 ext4_fsblk_t i_file_acl;
930
931 /*
932 * i_block_group is the number of the block group which contains
933 * this file's inode. Constant across the lifetime of the inode,
934 * it is used for making block allocation decisions - we try to
935 * place a file's data blocks near its inode block, and new inodes
936 * near to their parent directory's inode.
937 */
938 ext4_group_t i_block_group;
939 ext4_lblk_t i_dir_start_lookup;
940#if (BITS_PER_LONG < 64)
941 unsigned long i_state_flags; /* Dynamic state flags */
942#endif
943 unsigned long i_flags;
944
945 /*
946 * Extended attributes can be read independently of the main file
947 * data. Taking i_mutex even when reading would cause contention
948 * between readers of EAs and writers of regular file data, so
949 * instead we synchronize on xattr_sem when reading or changing
950 * EAs.
951 */
952 struct rw_semaphore xattr_sem;
953
954 struct list_head i_orphan; /* unlinked but open inodes */
955
956 /*
957 * i_disksize keeps track of what the inode size is ON DISK, not
958 * in memory. During truncate, i_size is set to the new size by
959 * the VFS prior to calling ext4_truncate(), but the filesystem won't
960 * set i_disksize to 0 until the truncate is actually under way.
961 *
962 * The intent is that i_disksize always represents the blocks which
963 * are used by this file. This allows recovery to restart truncate
964 * on orphans if we crash during truncate. We actually write i_disksize
965 * into the on-disk inode when writing inodes out, instead of i_size.
966 *
967 * The only time when i_disksize and i_size may be different is when
968 * a truncate is in progress. The only things which change i_disksize
969 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
970 */
971 loff_t i_disksize;
972
973 /*
974 * i_data_sem is for serialising ext4_truncate() against
975 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
976 * data tree are chopped off during truncate. We can't do that in
977 * ext4 because whenever we perform intermediate commits during
978 * truncate, the inode and all the metadata blocks *must* be in a
979 * consistent state which allows truncation of the orphans to restart
980 * during recovery. Hence we must fix the get_block-vs-truncate race
981 * by other means, so we have i_data_sem.
982 */
983 struct rw_semaphore i_data_sem;
984 /*
985 * i_mmap_sem is for serializing page faults with truncate / punch hole
986 * operations. We have to make sure that new page cannot be faulted in
987 * a section of the inode that is being punched. We cannot easily use
988 * i_data_sem for this since we need protection for the whole punch
989 * operation and i_data_sem ranks below transaction start so we have
990 * to occasionally drop it.
991 */
992 struct rw_semaphore i_mmap_sem;
993 struct inode vfs_inode;
994 struct jbd2_inode *jinode;
995
996 spinlock_t i_raw_lock; /* protects updates to the raw inode */
997
998 /*
999 * File creation time. Its function is same as that of
1000 * struct timespec64 i_{a,c,m}time in the generic inode.
1001 */
1002 struct timespec64 i_crtime;
1003
1004 /* mballoc */
1005 struct list_head i_prealloc_list;
1006 spinlock_t i_prealloc_lock;
1007
1008 /* extents status tree */
1009 struct ext4_es_tree i_es_tree;
1010 rwlock_t i_es_lock;
1011 struct list_head i_es_list;
1012 unsigned int i_es_all_nr; /* protected by i_es_lock */
1013 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1014 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1015 extents to shrink. Protected by
1016 i_es_lock */
1017
1018 /* ialloc */
1019 ext4_group_t i_last_alloc_group;
1020
1021 /* allocation reservation info for delalloc */
1022 /* In case of bigalloc, this refer to clusters rather than blocks */
1023 unsigned int i_reserved_data_blocks;
1024 ext4_lblk_t i_da_metadata_calc_last_lblock;
1025 int i_da_metadata_calc_len;
1026
1027 /* pending cluster reservations for bigalloc file systems */
1028 struct ext4_pending_tree i_pending_tree;
1029
1030 /* on-disk additional length */
1031 __u16 i_extra_isize;
1032
1033 /* Indicate the inline data space. */
1034 u16 i_inline_off;
1035 u16 i_inline_size;
1036
1037#ifdef CONFIG_QUOTA
1038 /* quota space reservation, managed internally by quota code */
1039 qsize_t i_reserved_quota;
1040#endif
1041
1042 /* Lock protecting lists below */
1043 spinlock_t i_completed_io_lock;
1044 /*
1045 * Completed IOs that need unwritten extents handling and have
1046 * transaction reserved
1047 */
1048 struct list_head i_rsv_conversion_list;
1049 struct work_struct i_rsv_conversion_work;
1050 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1051
1052 spinlock_t i_block_reservation_lock;
1053
1054 /*
1055 * Transactions that contain inode's metadata needed to complete
1056 * fsync and fdatasync, respectively.
1057 */
1058 tid_t i_sync_tid;
1059 tid_t i_datasync_tid;
1060
1061#ifdef CONFIG_QUOTA
1062 struct dquot *i_dquot[MAXQUOTAS];
1063#endif
1064
1065 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1066 __u32 i_csum_seed;
1067
1068 kprojid_t i_projid;
1069};
1070
1071/*
1072 * File system states
1073 */
1074#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1075#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1076#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1077
1078/*
1079 * Misc. filesystem flags
1080 */
1081#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1082#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1083#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1084
1085/*
1086 * Mount flags set via mount options or defaults
1087 */
1088#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1089#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1090#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1091#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1092#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1093#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1094#define EXT4_MOUNT_ERRORS_MASK 0x00070
1095#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1096#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1097#ifdef CONFIG_FS_DAX
1098#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1099#else
1100#define EXT4_MOUNT_DAX 0
1101#endif
1102#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1103#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1104#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1105#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1106#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1107#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1108#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1109#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1110#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1111#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1112#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1113#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1114 * enable enforcement for hidden
1115 * quota files */
1116#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1117 * enforcement for hidden quota
1118 * files */
1119#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1120 * enforcement */
1121#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1122#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1123#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1124#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1125#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1126#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1127#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1128#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1129#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1130
1131/*
1132 * Mount flags set either automatically (could not be set by mount option)
1133 * based on per file system feature or property or in special cases such as
1134 * distinguishing between explicit mount option definition and default.
1135 */
1136#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1137 specified delalloc */
1138#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1139 size of blocksize * 8
1140 blocks */
1141#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1142 file systems */
1143
1144#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1145 specified journal checksum */
1146
1147#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1148 ~EXT4_MOUNT_##opt
1149#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1150 EXT4_MOUNT_##opt
1151#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1152 EXT4_MOUNT_##opt)
1153
1154#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1155 ~EXT4_MOUNT2_##opt
1156#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1157 EXT4_MOUNT2_##opt
1158#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1159 EXT4_MOUNT2_##opt)
1160
1161#define ext4_test_and_set_bit __test_and_set_bit_le
1162#define ext4_set_bit __set_bit_le
1163#define ext4_set_bit_atomic ext2_set_bit_atomic
1164#define ext4_test_and_clear_bit __test_and_clear_bit_le
1165#define ext4_clear_bit __clear_bit_le
1166#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1167#define ext4_test_bit test_bit_le
1168#define ext4_find_next_zero_bit find_next_zero_bit_le
1169#define ext4_find_next_bit find_next_bit_le
1170
1171extern void ext4_set_bits(void *bm, int cur, int len);
1172
1173/*
1174 * Maximal mount counts between two filesystem checks
1175 */
1176#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1177#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1178
1179/*
1180 * Behaviour when detecting errors
1181 */
1182#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1183#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1184#define EXT4_ERRORS_PANIC 3 /* Panic */
1185#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1186
1187/* Metadata checksum algorithm codes */
1188#define EXT4_CRC32C_CHKSUM 1
1189
1190/*
1191 * Structure of the super block
1192 */
1193struct ext4_super_block {
1194/*00*/ __le32 s_inodes_count; /* Inodes count */
1195 __le32 s_blocks_count_lo; /* Blocks count */
1196 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1197 __le32 s_free_blocks_count_lo; /* Free blocks count */
1198/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1199 __le32 s_first_data_block; /* First Data Block */
1200 __le32 s_log_block_size; /* Block size */
1201 __le32 s_log_cluster_size; /* Allocation cluster size */
1202/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1203 __le32 s_clusters_per_group; /* # Clusters per group */
1204 __le32 s_inodes_per_group; /* # Inodes per group */
1205 __le32 s_mtime; /* Mount time */
1206/*30*/ __le32 s_wtime; /* Write time */
1207 __le16 s_mnt_count; /* Mount count */
1208 __le16 s_max_mnt_count; /* Maximal mount count */
1209 __le16 s_magic; /* Magic signature */
1210 __le16 s_state; /* File system state */
1211 __le16 s_errors; /* Behaviour when detecting errors */
1212 __le16 s_minor_rev_level; /* minor revision level */
1213/*40*/ __le32 s_lastcheck; /* time of last check */
1214 __le32 s_checkinterval; /* max. time between checks */
1215 __le32 s_creator_os; /* OS */
1216 __le32 s_rev_level; /* Revision level */
1217/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1218 __le16 s_def_resgid; /* Default gid for reserved blocks */
1219 /*
1220 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1221 *
1222 * Note: the difference between the compatible feature set and
1223 * the incompatible feature set is that if there is a bit set
1224 * in the incompatible feature set that the kernel doesn't
1225 * know about, it should refuse to mount the filesystem.
1226 *
1227 * e2fsck's requirements are more strict; if it doesn't know
1228 * about a feature in either the compatible or incompatible
1229 * feature set, it must abort and not try to meddle with
1230 * things it doesn't understand...
1231 */
1232 __le32 s_first_ino; /* First non-reserved inode */
1233 __le16 s_inode_size; /* size of inode structure */
1234 __le16 s_block_group_nr; /* block group # of this superblock */
1235 __le32 s_feature_compat; /* compatible feature set */
1236/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1237 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1238/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1239/*78*/ char s_volume_name[16]; /* volume name */
1240/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1241/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1242 /*
1243 * Performance hints. Directory preallocation should only
1244 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1245 */
1246 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1247 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1248 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1249 /*
1250 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1251 */
1252/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1253/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1254 __le32 s_journal_dev; /* device number of journal file */
1255 __le32 s_last_orphan; /* start of list of inodes to delete */
1256 __le32 s_hash_seed[4]; /* HTREE hash seed */
1257 __u8 s_def_hash_version; /* Default hash version to use */
1258 __u8 s_jnl_backup_type;
1259 __le16 s_desc_size; /* size of group descriptor */
1260/*100*/ __le32 s_default_mount_opts;
1261 __le32 s_first_meta_bg; /* First metablock block group */
1262 __le32 s_mkfs_time; /* When the filesystem was created */
1263 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1264 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1265/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1266 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1267 __le32 s_free_blocks_count_hi; /* Free blocks count */
1268 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1269 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1270 __le32 s_flags; /* Miscellaneous flags */
1271 __le16 s_raid_stride; /* RAID stride */
1272 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1273 __le64 s_mmp_block; /* Block for multi-mount protection */
1274 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1275 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1276 __u8 s_checksum_type; /* metadata checksum algorithm used */
1277 __u8 s_encryption_level; /* versioning level for encryption */
1278 __u8 s_reserved_pad; /* Padding to next 32bits */
1279 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1280 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1281 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1282 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1283 snapshot's future use */
1284 __le32 s_snapshot_list; /* inode number of the head of the
1285 on-disk snapshot list */
1286#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1287 __le32 s_error_count; /* number of fs errors */
1288 __le32 s_first_error_time; /* first time an error happened */
1289 __le32 s_first_error_ino; /* inode involved in first error */
1290 __le64 s_first_error_block; /* block involved of first error */
1291 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1292 __le32 s_first_error_line; /* line number where error happened */
1293 __le32 s_last_error_time; /* most recent time of an error */
1294 __le32 s_last_error_ino; /* inode involved in last error */
1295 __le32 s_last_error_line; /* line number where error happened */
1296 __le64 s_last_error_block; /* block involved of last error */
1297 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1298#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1299 __u8 s_mount_opts[64];
1300 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1301 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1302 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1303 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1304 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1305 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1306 __le32 s_lpf_ino; /* Location of the lost+found inode */
1307 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1308 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1309 __u8 s_wtime_hi;
1310 __u8 s_mtime_hi;
1311 __u8 s_mkfs_time_hi;
1312 __u8 s_lastcheck_hi;
1313 __u8 s_first_error_time_hi;
1314 __u8 s_last_error_time_hi;
1315 __u8 s_pad[2];
1316 __le32 s_reserved[96]; /* Padding to the end of the block */
1317 __le32 s_checksum; /* crc32c(superblock) */
1318};
1319
1320#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1321
1322#ifdef __KERNEL__
1323
1324/*
1325 * run-time mount flags
1326 */
1327#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1328#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1329#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1330
1331#ifdef CONFIG_FS_ENCRYPTION
1332#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1333 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1334#else
1335#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1336#endif
1337
1338/* Number of quota types we support */
1339#define EXT4_MAXQUOTAS 3
1340
1341/*
1342 * fourth extended-fs super-block data in memory
1343 */
1344struct ext4_sb_info {
1345 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1346 unsigned long s_inodes_per_block;/* Number of inodes per block */
1347 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1348 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1349 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1350 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1351 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1352 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1353 ext4_group_t s_groups_count; /* Number of groups in the fs */
1354 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1355 unsigned long s_overhead; /* # of fs overhead clusters */
1356 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1357 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1358 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1359 struct buffer_head * s_sbh; /* Buffer containing the super block */
1360 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1361 struct buffer_head **s_group_desc;
1362 unsigned int s_mount_opt;
1363 unsigned int s_mount_opt2;
1364 unsigned int s_mount_flags;
1365 unsigned int s_def_mount_opt;
1366 ext4_fsblk_t s_sb_block;
1367 atomic64_t s_resv_clusters;
1368 kuid_t s_resuid;
1369 kgid_t s_resgid;
1370 unsigned short s_mount_state;
1371 unsigned short s_pad;
1372 int s_addr_per_block_bits;
1373 int s_desc_per_block_bits;
1374 int s_inode_size;
1375 int s_first_ino;
1376 unsigned int s_inode_readahead_blks;
1377 unsigned int s_inode_goal;
1378 u32 s_hash_seed[4];
1379 int s_def_hash_version;
1380 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1381 struct percpu_counter s_freeclusters_counter;
1382 struct percpu_counter s_freeinodes_counter;
1383 struct percpu_counter s_dirs_counter;
1384 struct percpu_counter s_dirtyclusters_counter;
1385 struct blockgroup_lock *s_blockgroup_lock;
1386 struct proc_dir_entry *s_proc;
1387 struct kobject s_kobj;
1388 struct completion s_kobj_unregister;
1389 struct super_block *s_sb;
1390
1391 /* Journaling */
1392 struct journal_s *s_journal;
1393 struct list_head s_orphan;
1394 struct mutex s_orphan_lock;
1395 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1396 unsigned long s_commit_interval;
1397 u32 s_max_batch_time;
1398 u32 s_min_batch_time;
1399 struct block_device *journal_bdev;
1400#ifdef CONFIG_QUOTA
1401 /* Names of quota files with journalled quota */
1402 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1403 int s_jquota_fmt; /* Format of quota to use */
1404#endif
1405 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1406 struct rb_root system_blks;
1407
1408#ifdef EXTENTS_STATS
1409 /* ext4 extents stats */
1410 unsigned long s_ext_min;
1411 unsigned long s_ext_max;
1412 unsigned long s_depth_max;
1413 spinlock_t s_ext_stats_lock;
1414 unsigned long s_ext_blocks;
1415 unsigned long s_ext_extents;
1416#endif
1417
1418 /* for buddy allocator */
1419 struct ext4_group_info ***s_group_info;
1420 struct inode *s_buddy_cache;
1421 spinlock_t s_md_lock;
1422 unsigned short *s_mb_offsets;
1423 unsigned int *s_mb_maxs;
1424 unsigned int s_group_info_size;
1425 unsigned int s_mb_free_pending;
1426 struct list_head s_freed_data_list; /* List of blocks to be freed
1427 after commit completed */
1428
1429 /* tunables */
1430 unsigned long s_stripe;
1431 unsigned int s_mb_stream_request;
1432 unsigned int s_mb_max_to_scan;
1433 unsigned int s_mb_min_to_scan;
1434 unsigned int s_mb_stats;
1435 unsigned int s_mb_order2_reqs;
1436 unsigned int s_mb_group_prealloc;
1437 unsigned int s_max_dir_size_kb;
1438 /* where last allocation was done - for stream allocation */
1439 unsigned long s_mb_last_group;
1440 unsigned long s_mb_last_start;
1441
1442 /* stats for buddy allocator */
1443 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1444 atomic_t s_bal_success; /* we found long enough chunks */
1445 atomic_t s_bal_allocated; /* in blocks */
1446 atomic_t s_bal_ex_scanned; /* total extents scanned */
1447 atomic_t s_bal_goals; /* goal hits */
1448 atomic_t s_bal_breaks; /* too long searches */
1449 atomic_t s_bal_2orders; /* 2^order hits */
1450 spinlock_t s_bal_lock;
1451 unsigned long s_mb_buddies_generated;
1452 unsigned long long s_mb_generation_time;
1453 atomic_t s_mb_lost_chunks;
1454 atomic_t s_mb_preallocated;
1455 atomic_t s_mb_discarded;
1456 atomic_t s_lock_busy;
1457
1458 /* locality groups */
1459 struct ext4_locality_group __percpu *s_locality_groups;
1460
1461 /* for write statistics */
1462 unsigned long s_sectors_written_start;
1463 u64 s_kbytes_written;
1464
1465 /* the size of zero-out chunk */
1466 unsigned int s_extent_max_zeroout_kb;
1467
1468 unsigned int s_log_groups_per_flex;
1469 struct flex_groups *s_flex_groups;
1470 ext4_group_t s_flex_groups_allocated;
1471
1472 /* workqueue for reserved extent conversions (buffered io) */
1473 struct workqueue_struct *rsv_conversion_wq;
1474
1475 /* timer for periodic error stats printing */
1476 struct timer_list s_err_report;
1477
1478 /* Lazy inode table initialization info */
1479 struct ext4_li_request *s_li_request;
1480 /* Wait multiplier for lazy initialization thread */
1481 unsigned int s_li_wait_mult;
1482
1483 /* Kernel thread for multiple mount protection */
1484 struct task_struct *s_mmp_tsk;
1485
1486 /* record the last minlen when FITRIM is called. */
1487 atomic_t s_last_trim_minblks;
1488
1489 /* Reference to checksum algorithm driver via cryptoapi */
1490 struct crypto_shash *s_chksum_driver;
1491
1492 /* Precomputed FS UUID checksum for seeding other checksums */
1493 __u32 s_csum_seed;
1494
1495 /* Reclaim extents from extent status tree */
1496 struct shrinker s_es_shrinker;
1497 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1498 long s_es_nr_inode;
1499 struct ext4_es_stats s_es_stats;
1500 struct mb_cache *s_ea_block_cache;
1501 struct mb_cache *s_ea_inode_cache;
1502 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1503
1504 /* Ratelimit ext4 messages. */
1505 struct ratelimit_state s_err_ratelimit_state;
1506 struct ratelimit_state s_warning_ratelimit_state;
1507 struct ratelimit_state s_msg_ratelimit_state;
1508
1509 /* Barrier between changing inodes' journal flags and writepages ops. */
1510 struct percpu_rw_semaphore s_journal_flag_rwsem;
1511 struct dax_device *s_daxdev;
1512};
1513
1514static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1515{
1516 return sb->s_fs_info;
1517}
1518static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1519{
1520 return container_of(inode, struct ext4_inode_info, vfs_inode);
1521}
1522
1523static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1524{
1525 return ino == EXT4_ROOT_INO ||
1526 (ino >= EXT4_FIRST_INO(sb) &&
1527 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1528}
1529
1530/*
1531 * Inode dynamic state flags
1532 */
1533enum {
1534 EXT4_STATE_JDATA, /* journaled data exists */
1535 EXT4_STATE_NEW, /* inode is newly created */
1536 EXT4_STATE_XATTR, /* has in-inode xattrs */
1537 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1538 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1539 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1540 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1541 EXT4_STATE_NEWENTRY, /* File just added to dir */
1542 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1543 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1544 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1545};
1546
1547#define EXT4_INODE_BIT_FNS(name, field, offset) \
1548static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1549{ \
1550 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1551} \
1552static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1553{ \
1554 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1555} \
1556static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1557{ \
1558 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1559}
1560
1561/* Add these declarations here only so that these functions can be
1562 * found by name. Otherwise, they are very hard to locate. */
1563static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1564static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1565static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1566EXT4_INODE_BIT_FNS(flag, flags, 0)
1567
1568/* Add these declarations here only so that these functions can be
1569 * found by name. Otherwise, they are very hard to locate. */
1570static inline int ext4_test_inode_state(struct inode *inode, int bit);
1571static inline void ext4_set_inode_state(struct inode *inode, int bit);
1572static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1573#if (BITS_PER_LONG < 64)
1574EXT4_INODE_BIT_FNS(state, state_flags, 0)
1575
1576static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1577{
1578 (ei)->i_state_flags = 0;
1579}
1580#else
1581EXT4_INODE_BIT_FNS(state, flags, 32)
1582
1583static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1584{
1585 /* We depend on the fact that callers will set i_flags */
1586}
1587#endif
1588#else
1589/* Assume that user mode programs are passing in an ext4fs superblock, not
1590 * a kernel struct super_block. This will allow us to call the feature-test
1591 * macros from user land. */
1592#define EXT4_SB(sb) (sb)
1593#endif
1594
1595/*
1596 * Returns true if the inode is inode is encrypted
1597 */
1598#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1599
1600/*
1601 * Codes for operating systems
1602 */
1603#define EXT4_OS_LINUX 0
1604#define EXT4_OS_HURD 1
1605#define EXT4_OS_MASIX 2
1606#define EXT4_OS_FREEBSD 3
1607#define EXT4_OS_LITES 4
1608
1609/*
1610 * Revision levels
1611 */
1612#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1613#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1614
1615#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1616#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1617
1618#define EXT4_GOOD_OLD_INODE_SIZE 128
1619
1620/*
1621 * Feature set definitions
1622 */
1623
1624#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1625#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1626#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1627#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1628#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1629#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1630#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1631
1632#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1633#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1634#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1635#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1636#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1637#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1638#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1639#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1640#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1641/*
1642 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1643 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1644 * all other data structures' checksums. However, the METADATA_CSUM and
1645 * GDT_CSUM bits are mutually exclusive.
1646 */
1647#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1648#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1649#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1650
1651#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1652#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1653#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1654#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1655#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1656#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1657#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1658#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1659#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1660#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1661#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1662#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1663#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1664#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1665#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1666
1667extern void ext4_update_dynamic_rev(struct super_block *sb);
1668
1669#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1670static inline bool ext4_has_feature_##name(struct super_block *sb) \
1671{ \
1672 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1673 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1674} \
1675static inline void ext4_set_feature_##name(struct super_block *sb) \
1676{ \
1677 ext4_update_dynamic_rev(sb); \
1678 EXT4_SB(sb)->s_es->s_feature_compat |= \
1679 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1680} \
1681static inline void ext4_clear_feature_##name(struct super_block *sb) \
1682{ \
1683 EXT4_SB(sb)->s_es->s_feature_compat &= \
1684 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1685}
1686
1687#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1688static inline bool ext4_has_feature_##name(struct super_block *sb) \
1689{ \
1690 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1691 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1692} \
1693static inline void ext4_set_feature_##name(struct super_block *sb) \
1694{ \
1695 ext4_update_dynamic_rev(sb); \
1696 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1697 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1698} \
1699static inline void ext4_clear_feature_##name(struct super_block *sb) \
1700{ \
1701 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1702 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1703}
1704
1705#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1706static inline bool ext4_has_feature_##name(struct super_block *sb) \
1707{ \
1708 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1709 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1710} \
1711static inline void ext4_set_feature_##name(struct super_block *sb) \
1712{ \
1713 ext4_update_dynamic_rev(sb); \
1714 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1715 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1716} \
1717static inline void ext4_clear_feature_##name(struct super_block *sb) \
1718{ \
1719 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1720 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1721}
1722
1723EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1724EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1725EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1726EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1727EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1728EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1729EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1730
1731EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1732EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1733EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1734EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1735EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1736EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1737EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1738EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1739EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1740EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1741EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1742EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1743
1744EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1745EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1746EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1747EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1748EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1749EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1750EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1751EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1752EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1753EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1754EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1755EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1756EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1757EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1758EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1759
1760#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1761#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1762 EXT4_FEATURE_INCOMPAT_META_BG)
1763#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1764 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1765 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1766
1767#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1768#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1769 EXT4_FEATURE_INCOMPAT_RECOVER| \
1770 EXT4_FEATURE_INCOMPAT_META_BG)
1771#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1772 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1773 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1774
1775#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1776#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1777 EXT4_FEATURE_INCOMPAT_RECOVER| \
1778 EXT4_FEATURE_INCOMPAT_META_BG| \
1779 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1780 EXT4_FEATURE_INCOMPAT_64BIT| \
1781 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1782 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1783 EXT4_FEATURE_INCOMPAT_MMP | \
1784 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1785 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1786 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1787 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1788#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1789 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1790 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1791 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1792 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1793 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1794 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1795 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1796 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1797 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1798 EXT4_FEATURE_RO_COMPAT_PROJECT)
1799
1800#define EXTN_FEATURE_FUNCS(ver) \
1801static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1802{ \
1803 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1804 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1805} \
1806static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1807{ \
1808 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1809 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1810} \
1811static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1812{ \
1813 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1814 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1815}
1816
1817EXTN_FEATURE_FUNCS(2)
1818EXTN_FEATURE_FUNCS(3)
1819EXTN_FEATURE_FUNCS(4)
1820
1821static inline bool ext4_has_compat_features(struct super_block *sb)
1822{
1823 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1824}
1825static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1826{
1827 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1828}
1829static inline bool ext4_has_incompat_features(struct super_block *sb)
1830{
1831 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1832}
1833
1834/*
1835 * Superblock flags
1836 */
1837#define EXT4_FLAGS_RESIZING 0
1838#define EXT4_FLAGS_SHUTDOWN 1
1839
1840static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1841{
1842 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1843}
1844
1845
1846/*
1847 * Default values for user and/or group using reserved blocks
1848 */
1849#define EXT4_DEF_RESUID 0
1850#define EXT4_DEF_RESGID 0
1851
1852/*
1853 * Default project ID
1854 */
1855#define EXT4_DEF_PROJID 0
1856
1857#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1858
1859/*
1860 * Default mount options
1861 */
1862#define EXT4_DEFM_DEBUG 0x0001
1863#define EXT4_DEFM_BSDGROUPS 0x0002
1864#define EXT4_DEFM_XATTR_USER 0x0004
1865#define EXT4_DEFM_ACL 0x0008
1866#define EXT4_DEFM_UID16 0x0010
1867#define EXT4_DEFM_JMODE 0x0060
1868#define EXT4_DEFM_JMODE_DATA 0x0020
1869#define EXT4_DEFM_JMODE_ORDERED 0x0040
1870#define EXT4_DEFM_JMODE_WBACK 0x0060
1871#define EXT4_DEFM_NOBARRIER 0x0100
1872#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1873#define EXT4_DEFM_DISCARD 0x0400
1874#define EXT4_DEFM_NODELALLOC 0x0800
1875
1876/*
1877 * Default journal batch times
1878 */
1879#define EXT4_DEF_MIN_BATCH_TIME 0
1880#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1881
1882/*
1883 * Minimum number of groups in a flexgroup before we separate out
1884 * directories into the first block group of a flexgroup
1885 */
1886#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1887
1888/*
1889 * Structure of a directory entry
1890 */
1891#define EXT4_NAME_LEN 255
1892
1893struct ext4_dir_entry {
1894 __le32 inode; /* Inode number */
1895 __le16 rec_len; /* Directory entry length */
1896 __le16 name_len; /* Name length */
1897 char name[EXT4_NAME_LEN]; /* File name */
1898};
1899
1900/*
1901 * The new version of the directory entry. Since EXT4 structures are
1902 * stored in intel byte order, and the name_len field could never be
1903 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1904 * file_type field.
1905 */
1906struct ext4_dir_entry_2 {
1907 __le32 inode; /* Inode number */
1908 __le16 rec_len; /* Directory entry length */
1909 __u8 name_len; /* Name length */
1910 __u8 file_type;
1911 char name[EXT4_NAME_LEN]; /* File name */
1912};
1913
1914/*
1915 * This is a bogus directory entry at the end of each leaf block that
1916 * records checksums.
1917 */
1918struct ext4_dir_entry_tail {
1919 __le32 det_reserved_zero1; /* Pretend to be unused */
1920 __le16 det_rec_len; /* 12 */
1921 __u8 det_reserved_zero2; /* Zero name length */
1922 __u8 det_reserved_ft; /* 0xDE, fake file type */
1923 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1924};
1925
1926#define EXT4_DIRENT_TAIL(block, blocksize) \
1927 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1928 ((blocksize) - \
1929 sizeof(struct ext4_dir_entry_tail))))
1930
1931/*
1932 * Ext4 directory file types. Only the low 3 bits are used. The
1933 * other bits are reserved for now.
1934 */
1935#define EXT4_FT_UNKNOWN 0
1936#define EXT4_FT_REG_FILE 1
1937#define EXT4_FT_DIR 2
1938#define EXT4_FT_CHRDEV 3
1939#define EXT4_FT_BLKDEV 4
1940#define EXT4_FT_FIFO 5
1941#define EXT4_FT_SOCK 6
1942#define EXT4_FT_SYMLINK 7
1943
1944#define EXT4_FT_MAX 8
1945
1946#define EXT4_FT_DIR_CSUM 0xDE
1947
1948/*
1949 * EXT4_DIR_PAD defines the directory entries boundaries
1950 *
1951 * NOTE: It must be a multiple of 4
1952 */
1953#define EXT4_DIR_PAD 4
1954#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1955#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1956 ~EXT4_DIR_ROUND)
1957#define EXT4_MAX_REC_LEN ((1<<16)-1)
1958
1959/*
1960 * If we ever get support for fs block sizes > page_size, we'll need
1961 * to remove the #if statements in the next two functions...
1962 */
1963static inline unsigned int
1964ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1965{
1966 unsigned len = le16_to_cpu(dlen);
1967
1968#if (PAGE_SIZE >= 65536)
1969 if (len == EXT4_MAX_REC_LEN || len == 0)
1970 return blocksize;
1971 return (len & 65532) | ((len & 3) << 16);
1972#else
1973 return len;
1974#endif
1975}
1976
1977static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1978{
1979 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1980 BUG();
1981#if (PAGE_SIZE >= 65536)
1982 if (len < 65536)
1983 return cpu_to_le16(len);
1984 if (len == blocksize) {
1985 if (blocksize == 65536)
1986 return cpu_to_le16(EXT4_MAX_REC_LEN);
1987 else
1988 return cpu_to_le16(0);
1989 }
1990 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1991#else
1992 return cpu_to_le16(len);
1993#endif
1994}
1995
1996/*
1997 * Hash Tree Directory indexing
1998 * (c) Daniel Phillips, 2001
1999 */
2000
2001#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2002 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2003#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2004 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2005#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2006
2007/* Legal values for the dx_root hash_version field: */
2008
2009#define DX_HASH_LEGACY 0
2010#define DX_HASH_HALF_MD4 1
2011#define DX_HASH_TEA 2
2012#define DX_HASH_LEGACY_UNSIGNED 3
2013#define DX_HASH_HALF_MD4_UNSIGNED 4
2014#define DX_HASH_TEA_UNSIGNED 5
2015
2016static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2017 const void *address, unsigned int length)
2018{
2019 struct {
2020 struct shash_desc shash;
2021 char ctx[4];
2022 } desc;
2023
2024 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2025
2026 desc.shash.tfm = sbi->s_chksum_driver;
2027 desc.shash.flags = 0;
2028 *(u32 *)desc.ctx = crc;
2029
2030 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2031
2032 return *(u32 *)desc.ctx;
2033}
2034
2035#ifdef __KERNEL__
2036
2037/* hash info structure used by the directory hash */
2038struct dx_hash_info
2039{
2040 u32 hash;
2041 u32 minor_hash;
2042 int hash_version;
2043 u32 *seed;
2044};
2045
2046
2047/* 32 and 64 bit signed EOF for dx directories */
2048#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2049#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2050
2051
2052/*
2053 * Control parameters used by ext4_htree_next_block
2054 */
2055#define HASH_NB_ALWAYS 1
2056
2057struct ext4_filename {
2058 const struct qstr *usr_fname;
2059 struct fscrypt_str disk_name;
2060 struct dx_hash_info hinfo;
2061#ifdef CONFIG_FS_ENCRYPTION
2062 struct fscrypt_str crypto_buf;
2063#endif
2064};
2065
2066#define fname_name(p) ((p)->disk_name.name)
2067#define fname_len(p) ((p)->disk_name.len)
2068
2069/*
2070 * Describe an inode's exact location on disk and in memory
2071 */
2072struct ext4_iloc
2073{
2074 struct buffer_head *bh;
2075 unsigned long offset;
2076 ext4_group_t block_group;
2077};
2078
2079static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2080{
2081 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2082}
2083
2084static inline bool ext4_is_quota_file(struct inode *inode)
2085{
2086 return IS_NOQUOTA(inode) &&
2087 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2088}
2089
2090/*
2091 * This structure is stuffed into the struct file's private_data field
2092 * for directories. It is where we put information so that we can do
2093 * readdir operations in hash tree order.
2094 */
2095struct dir_private_info {
2096 struct rb_root root;
2097 struct rb_node *curr_node;
2098 struct fname *extra_fname;
2099 loff_t last_pos;
2100 __u32 curr_hash;
2101 __u32 curr_minor_hash;
2102 __u32 next_hash;
2103};
2104
2105/* calculate the first block number of the group */
2106static inline ext4_fsblk_t
2107ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2108{
2109 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2110 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2111}
2112
2113/*
2114 * Special error return code only used by dx_probe() and its callers.
2115 */
2116#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2117
2118/* htree levels for ext4 */
2119#define EXT4_HTREE_LEVEL_COMPAT 2
2120#define EXT4_HTREE_LEVEL 3
2121
2122static inline int ext4_dir_htree_level(struct super_block *sb)
2123{
2124 return ext4_has_feature_largedir(sb) ?
2125 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2126}
2127
2128/*
2129 * Timeout and state flag for lazy initialization inode thread.
2130 */
2131#define EXT4_DEF_LI_WAIT_MULT 10
2132#define EXT4_DEF_LI_MAX_START_DELAY 5
2133#define EXT4_LAZYINIT_QUIT 0x0001
2134#define EXT4_LAZYINIT_RUNNING 0x0002
2135
2136/*
2137 * Lazy inode table initialization info
2138 */
2139struct ext4_lazy_init {
2140 unsigned long li_state;
2141 struct list_head li_request_list;
2142 struct mutex li_list_mtx;
2143};
2144
2145struct ext4_li_request {
2146 struct super_block *lr_super;
2147 struct ext4_sb_info *lr_sbi;
2148 ext4_group_t lr_next_group;
2149 struct list_head lr_request;
2150 unsigned long lr_next_sched;
2151 unsigned long lr_timeout;
2152};
2153
2154struct ext4_features {
2155 struct kobject f_kobj;
2156 struct completion f_kobj_unregister;
2157};
2158
2159/*
2160 * This structure will be used for multiple mount protection. It will be
2161 * written into the block number saved in the s_mmp_block field in the
2162 * superblock. Programs that check MMP should assume that if
2163 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2164 * to use the filesystem, regardless of how old the timestamp is.
2165 */
2166#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2167#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2168#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2169#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2170
2171struct mmp_struct {
2172 __le32 mmp_magic; /* Magic number for MMP */
2173 __le32 mmp_seq; /* Sequence no. updated periodically */
2174
2175 /*
2176 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2177 * purposes and do not affect the correctness of the algorithm
2178 */
2179 __le64 mmp_time; /* Time last updated */
2180 char mmp_nodename[64]; /* Node which last updated MMP block */
2181 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2182
2183 /*
2184 * mmp_check_interval is used to verify if the MMP block has been
2185 * updated on the block device. The value is updated based on the
2186 * maximum time to write the MMP block during an update cycle.
2187 */
2188 __le16 mmp_check_interval;
2189
2190 __le16 mmp_pad1;
2191 __le32 mmp_pad2[226];
2192 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2193};
2194
2195/* arguments passed to the mmp thread */
2196struct mmpd_data {
2197 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2198 struct super_block *sb; /* super block of the fs */
2199};
2200
2201/*
2202 * Check interval multiplier
2203 * The MMP block is written every update interval and initially checked every
2204 * update interval x the multiplier (the value is then adapted based on the
2205 * write latency). The reason is that writes can be delayed under load and we
2206 * don't want readers to incorrectly assume that the filesystem is no longer
2207 * in use.
2208 */
2209#define EXT4_MMP_CHECK_MULT 2UL
2210
2211/*
2212 * Minimum interval for MMP checking in seconds.
2213 */
2214#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2215
2216/*
2217 * Maximum interval for MMP checking in seconds.
2218 */
2219#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2220
2221/*
2222 * Function prototypes
2223 */
2224
2225/*
2226 * Ok, these declarations are also in <linux/kernel.h> but none of the
2227 * ext4 source programs needs to include it so they are duplicated here.
2228 */
2229# define NORET_TYPE /**/
2230# define ATTRIB_NORET __attribute__((noreturn))
2231# define NORET_AND noreturn,
2232
2233/* bitmap.c */
2234extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2235void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2236 struct ext4_group_desc *gdp,
2237 struct buffer_head *bh, int sz);
2238int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2239 struct ext4_group_desc *gdp,
2240 struct buffer_head *bh, int sz);
2241void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2242 struct ext4_group_desc *gdp,
2243 struct buffer_head *bh);
2244int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2245 struct ext4_group_desc *gdp,
2246 struct buffer_head *bh);
2247
2248/* balloc.c */
2249extern void ext4_get_group_no_and_offset(struct super_block *sb,
2250 ext4_fsblk_t blocknr,
2251 ext4_group_t *blockgrpp,
2252 ext4_grpblk_t *offsetp);
2253extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2254 ext4_fsblk_t block);
2255
2256extern unsigned int ext4_block_group(struct super_block *sb,
2257 ext4_fsblk_t blocknr);
2258extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2259 ext4_fsblk_t blocknr);
2260extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2261extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2262 ext4_group_t group);
2263extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2264 ext4_fsblk_t goal,
2265 unsigned int flags,
2266 unsigned long *count,
2267 int *errp);
2268extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2269 s64 nclusters, unsigned int flags);
2270extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2271extern void ext4_check_blocks_bitmap(struct super_block *);
2272extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2273 ext4_group_t block_group,
2274 struct buffer_head ** bh);
2275extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2276
2277extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2278 ext4_group_t block_group);
2279extern int ext4_wait_block_bitmap(struct super_block *sb,
2280 ext4_group_t block_group,
2281 struct buffer_head *bh);
2282extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2283 ext4_group_t block_group);
2284extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2285 ext4_group_t block_group,
2286 struct ext4_group_desc *gdp);
2287ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2288
2289#ifdef CONFIG_FS_ENCRYPTION
2290static inline int ext4_fname_setup_filename(struct inode *dir,
2291 const struct qstr *iname,
2292 int lookup, struct ext4_filename *fname)
2293{
2294 struct fscrypt_name name;
2295 int err;
2296
2297 memset(fname, 0, sizeof(struct ext4_filename));
2298
2299 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2300
2301 fname->usr_fname = name.usr_fname;
2302 fname->disk_name = name.disk_name;
2303 fname->hinfo.hash = name.hash;
2304 fname->hinfo.minor_hash = name.minor_hash;
2305 fname->crypto_buf = name.crypto_buf;
2306 return err;
2307}
2308
2309static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2310{
2311 struct fscrypt_name name;
2312
2313 name.crypto_buf = fname->crypto_buf;
2314 fscrypt_free_filename(&name);
2315
2316 fname->crypto_buf.name = NULL;
2317 fname->usr_fname = NULL;
2318 fname->disk_name.name = NULL;
2319}
2320#else
2321static inline int ext4_fname_setup_filename(struct inode *dir,
2322 const struct qstr *iname,
2323 int lookup, struct ext4_filename *fname)
2324{
2325 fname->usr_fname = iname;
2326 fname->disk_name.name = (unsigned char *) iname->name;
2327 fname->disk_name.len = iname->len;
2328 return 0;
2329}
2330static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2331
2332#endif
2333
2334/* dir.c */
2335extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2336 struct file *,
2337 struct ext4_dir_entry_2 *,
2338 struct buffer_head *, char *, int,
2339 unsigned int);
2340#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2341 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2342 (de), (bh), (buf), (size), (offset)))
2343extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2344 __u32 minor_hash,
2345 struct ext4_dir_entry_2 *dirent,
2346 struct fscrypt_str *ent_name);
2347extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2348extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2349 struct buffer_head *bh,
2350 void *buf, int buf_size,
2351 struct ext4_filename *fname,
2352 struct ext4_dir_entry_2 **dest_de);
2353void ext4_insert_dentry(struct inode *inode,
2354 struct ext4_dir_entry_2 *de,
2355 int buf_size,
2356 struct ext4_filename *fname);
2357static inline void ext4_update_dx_flag(struct inode *inode)
2358{
2359 if (!ext4_has_feature_dir_index(inode->i_sb))
2360 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2361}
2362static const unsigned char ext4_filetype_table[] = {
2363 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2364};
2365
2366static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2367{
2368 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2369 return DT_UNKNOWN;
2370
2371 return ext4_filetype_table[filetype];
2372}
2373extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2374 void *buf, int buf_size);
2375
2376/* fsync.c */
2377extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2378
2379/* hash.c */
2380extern int ext4fs_dirhash(const char *name, int len, struct
2381 dx_hash_info *hinfo);
2382
2383/* ialloc.c */
2384extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2385 const struct qstr *qstr, __u32 goal,
2386 uid_t *owner, __u32 i_flags,
2387 int handle_type, unsigned int line_no,
2388 int nblocks);
2389
2390#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2391 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2392 i_flags, 0, 0, 0)
2393#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2394 type, nblocks) \
2395 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2396 0, (type), __LINE__, (nblocks))
2397
2398
2399extern void ext4_free_inode(handle_t *, struct inode *);
2400extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2401extern unsigned long ext4_count_free_inodes(struct super_block *);
2402extern unsigned long ext4_count_dirs(struct super_block *);
2403extern void ext4_check_inodes_bitmap(struct super_block *);
2404extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2405extern int ext4_init_inode_table(struct super_block *sb,
2406 ext4_group_t group, int barrier);
2407extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2408
2409/* mballoc.c */
2410extern const struct seq_operations ext4_mb_seq_groups_ops;
2411extern long ext4_mb_stats;
2412extern long ext4_mb_max_to_scan;
2413extern int ext4_mb_init(struct super_block *);
2414extern int ext4_mb_release(struct super_block *);
2415extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2416 struct ext4_allocation_request *, int *);
2417extern int ext4_mb_reserve_blocks(struct super_block *, int);
2418extern void ext4_discard_preallocations(struct inode *);
2419extern int __init ext4_init_mballoc(void);
2420extern void ext4_exit_mballoc(void);
2421extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2422 struct buffer_head *bh, ext4_fsblk_t block,
2423 unsigned long count, int flags);
2424extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2425 ext4_group_t ngroups);
2426extern int ext4_mb_add_groupinfo(struct super_block *sb,
2427 ext4_group_t i, struct ext4_group_desc *desc);
2428extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2429 ext4_fsblk_t block, unsigned long count);
2430extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2431extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2432
2433/* inode.c */
2434int ext4_inode_is_fast_symlink(struct inode *inode);
2435struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2436struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2437int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2438 bool wait, struct buffer_head **bhs);
2439int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2440 struct buffer_head *bh_result, int create);
2441int ext4_get_block(struct inode *inode, sector_t iblock,
2442 struct buffer_head *bh_result, int create);
2443int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2444 struct buffer_head *bh_result, int create);
2445int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2446 struct buffer_head *bh, int create);
2447int ext4_walk_page_buffers(handle_t *handle,
2448 struct buffer_head *head,
2449 unsigned from,
2450 unsigned to,
2451 int *partial,
2452 int (*fn)(handle_t *handle,
2453 struct buffer_head *bh));
2454int do_journal_get_write_access(handle_t *handle,
2455 struct buffer_head *bh);
2456#define FALL_BACK_TO_NONDELALLOC 1
2457#define CONVERT_INLINE_DATA 2
2458
2459typedef enum {
2460 EXT4_IGET_NORMAL = 0,
2461 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2462 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2463} ext4_iget_flags;
2464
2465extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2466 ext4_iget_flags flags, const char *function,
2467 unsigned int line);
2468
2469#define ext4_iget(sb, ino, flags) \
2470 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2471
2472extern int ext4_write_inode(struct inode *, struct writeback_control *);
2473extern int ext4_setattr(struct dentry *, struct iattr *);
2474extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2475extern void ext4_evict_inode(struct inode *);
2476extern void ext4_clear_inode(struct inode *);
2477extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2478extern int ext4_sync_inode(handle_t *, struct inode *);
2479extern void ext4_dirty_inode(struct inode *, int);
2480extern int ext4_change_inode_journal_flag(struct inode *, int);
2481extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2482extern int ext4_inode_attach_jinode(struct inode *inode);
2483extern int ext4_can_truncate(struct inode *inode);
2484extern int ext4_truncate(struct inode *);
2485extern int ext4_break_layouts(struct inode *);
2486extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2487extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2488extern void ext4_set_inode_flags(struct inode *);
2489extern int ext4_alloc_da_blocks(struct inode *inode);
2490extern void ext4_set_aops(struct inode *inode);
2491extern int ext4_writepage_trans_blocks(struct inode *);
2492extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2493extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2494 loff_t lstart, loff_t lend);
2495extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2496extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2497extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2498extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2499extern void ext4_da_release_space(struct inode *inode, int to_free);
2500extern void ext4_da_update_reserve_space(struct inode *inode,
2501 int used, int quota_claim);
2502extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2503 ext4_fsblk_t pblk, ext4_lblk_t len);
2504
2505/* indirect.c */
2506extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2507 struct ext4_map_blocks *map, int flags);
2508extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2509extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2510extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2511extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2512 ext4_lblk_t start, ext4_lblk_t end);
2513
2514/* ioctl.c */
2515extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2516extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2517
2518/* migrate.c */
2519extern int ext4_ext_migrate(struct inode *);
2520extern int ext4_ind_migrate(struct inode *inode);
2521
2522/* namei.c */
2523extern int ext4_dirent_csum_verify(struct inode *inode,
2524 struct ext4_dir_entry *dirent);
2525extern int ext4_orphan_add(handle_t *, struct inode *);
2526extern int ext4_orphan_del(handle_t *, struct inode *);
2527extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2528 __u32 start_minor_hash, __u32 *next_hash);
2529extern int ext4_search_dir(struct buffer_head *bh,
2530 char *search_buf,
2531 int buf_size,
2532 struct inode *dir,
2533 struct ext4_filename *fname,
2534 unsigned int offset,
2535 struct ext4_dir_entry_2 **res_dir);
2536extern int ext4_generic_delete_entry(handle_t *handle,
2537 struct inode *dir,
2538 struct ext4_dir_entry_2 *de_del,
2539 struct buffer_head *bh,
2540 void *entry_buf,
2541 int buf_size,
2542 int csum_size);
2543extern bool ext4_empty_dir(struct inode *inode);
2544
2545/* resize.c */
2546extern int ext4_group_add(struct super_block *sb,
2547 struct ext4_new_group_data *input);
2548extern int ext4_group_extend(struct super_block *sb,
2549 struct ext4_super_block *es,
2550 ext4_fsblk_t n_blocks_count);
2551extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2552
2553/* super.c */
2554extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2555 sector_t block, int op_flags);
2556extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2557extern int ext4_calculate_overhead(struct super_block *sb);
2558extern void ext4_superblock_csum_set(struct super_block *sb);
2559extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2560extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2561extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2562 ext4_group_t ngroup);
2563extern const char *ext4_decode_error(struct super_block *sb, int errno,
2564 char nbuf[16]);
2565extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2566 ext4_group_t block_group,
2567 unsigned int flags);
2568
2569extern __printf(4, 5)
2570void __ext4_error(struct super_block *, const char *, unsigned int,
2571 const char *, ...);
2572extern __printf(5, 6)
2573void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2574 const char *, ...);
2575extern __printf(5, 6)
2576void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2577 const char *, ...);
2578extern void __ext4_std_error(struct super_block *, const char *,
2579 unsigned int, int);
2580extern __printf(4, 5)
2581void __ext4_abort(struct super_block *, const char *, unsigned int,
2582 const char *, ...);
2583extern __printf(4, 5)
2584void __ext4_warning(struct super_block *, const char *, unsigned int,
2585 const char *, ...);
2586extern __printf(4, 5)
2587void __ext4_warning_inode(const struct inode *inode, const char *function,
2588 unsigned int line, const char *fmt, ...);
2589extern __printf(3, 4)
2590void __ext4_msg(struct super_block *, const char *, const char *, ...);
2591extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2592 const char *, unsigned int, const char *);
2593extern __printf(7, 8)
2594void __ext4_grp_locked_error(const char *, unsigned int,
2595 struct super_block *, ext4_group_t,
2596 unsigned long, ext4_fsblk_t,
2597 const char *, ...);
2598
2599#define EXT4_ERROR_INODE(inode, fmt, a...) \
2600 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2601
2602#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2603 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2604
2605#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2606 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2607
2608#ifdef CONFIG_PRINTK
2609
2610#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2611 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2612#define ext4_error_file(file, func, line, block, fmt, ...) \
2613 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2614#define ext4_error(sb, fmt, ...) \
2615 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2616#define ext4_abort(sb, fmt, ...) \
2617 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2618#define ext4_warning(sb, fmt, ...) \
2619 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2620#define ext4_warning_inode(inode, fmt, ...) \
2621 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2622#define ext4_msg(sb, level, fmt, ...) \
2623 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2624#define dump_mmp_msg(sb, mmp, msg) \
2625 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2626#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2627 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2628 fmt, ##__VA_ARGS__)
2629
2630#else
2631
2632#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2633do { \
2634 no_printk(fmt, ##__VA_ARGS__); \
2635 __ext4_error_inode(inode, "", 0, block, " "); \
2636} while (0)
2637#define ext4_error_file(file, func, line, block, fmt, ...) \
2638do { \
2639 no_printk(fmt, ##__VA_ARGS__); \
2640 __ext4_error_file(file, "", 0, block, " "); \
2641} while (0)
2642#define ext4_error(sb, fmt, ...) \
2643do { \
2644 no_printk(fmt, ##__VA_ARGS__); \
2645 __ext4_error(sb, "", 0, " "); \
2646} while (0)
2647#define ext4_abort(sb, fmt, ...) \
2648do { \
2649 no_printk(fmt, ##__VA_ARGS__); \
2650 __ext4_abort(sb, "", 0, " "); \
2651} while (0)
2652#define ext4_warning(sb, fmt, ...) \
2653do { \
2654 no_printk(fmt, ##__VA_ARGS__); \
2655 __ext4_warning(sb, "", 0, " "); \
2656} while (0)
2657#define ext4_warning_inode(inode, fmt, ...) \
2658do { \
2659 no_printk(fmt, ##__VA_ARGS__); \
2660 __ext4_warning_inode(inode, "", 0, " "); \
2661} while (0)
2662#define ext4_msg(sb, level, fmt, ...) \
2663do { \
2664 no_printk(fmt, ##__VA_ARGS__); \
2665 __ext4_msg(sb, "", " "); \
2666} while (0)
2667#define dump_mmp_msg(sb, mmp, msg) \
2668 __dump_mmp_msg(sb, mmp, "", 0, "")
2669#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2670do { \
2671 no_printk(fmt, ##__VA_ARGS__); \
2672 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2673} while (0)
2674
2675#endif
2676
2677extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2678 __u32 compat);
2679extern int ext4_update_rocompat_feature(handle_t *handle,
2680 struct super_block *sb, __u32 rocompat);
2681extern int ext4_update_incompat_feature(handle_t *handle,
2682 struct super_block *sb, __u32 incompat);
2683extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2684 struct ext4_group_desc *bg);
2685extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2686 struct ext4_group_desc *bg);
2687extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2688 struct ext4_group_desc *bg);
2689extern __u32 ext4_free_group_clusters(struct super_block *sb,
2690 struct ext4_group_desc *bg);
2691extern __u32 ext4_free_inodes_count(struct super_block *sb,
2692 struct ext4_group_desc *bg);
2693extern __u32 ext4_used_dirs_count(struct super_block *sb,
2694 struct ext4_group_desc *bg);
2695extern __u32 ext4_itable_unused_count(struct super_block *sb,
2696 struct ext4_group_desc *bg);
2697extern void ext4_block_bitmap_set(struct super_block *sb,
2698 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2699extern void ext4_inode_bitmap_set(struct super_block *sb,
2700 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2701extern void ext4_inode_table_set(struct super_block *sb,
2702 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2703extern void ext4_free_group_clusters_set(struct super_block *sb,
2704 struct ext4_group_desc *bg,
2705 __u32 count);
2706extern void ext4_free_inodes_set(struct super_block *sb,
2707 struct ext4_group_desc *bg, __u32 count);
2708extern void ext4_used_dirs_set(struct super_block *sb,
2709 struct ext4_group_desc *bg, __u32 count);
2710extern void ext4_itable_unused_set(struct super_block *sb,
2711 struct ext4_group_desc *bg, __u32 count);
2712extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2713 struct ext4_group_desc *gdp);
2714extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2715 struct ext4_group_desc *gdp);
2716extern int ext4_register_li_request(struct super_block *sb,
2717 ext4_group_t first_not_zeroed);
2718
2719static inline int ext4_has_metadata_csum(struct super_block *sb)
2720{
2721 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2722 !EXT4_SB(sb)->s_chksum_driver);
2723
2724 return ext4_has_feature_metadata_csum(sb) &&
2725 (EXT4_SB(sb)->s_chksum_driver != NULL);
2726}
2727
2728static inline int ext4_has_group_desc_csum(struct super_block *sb)
2729{
2730 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2731}
2732
2733static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2734{
2735 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2736 le32_to_cpu(es->s_blocks_count_lo);
2737}
2738
2739static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2740{
2741 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2742 le32_to_cpu(es->s_r_blocks_count_lo);
2743}
2744
2745static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2746{
2747 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2748 le32_to_cpu(es->s_free_blocks_count_lo);
2749}
2750
2751static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2752 ext4_fsblk_t blk)
2753{
2754 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2755 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2756}
2757
2758static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2759 ext4_fsblk_t blk)
2760{
2761 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2762 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2763}
2764
2765static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2766 ext4_fsblk_t blk)
2767{
2768 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2769 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2770}
2771
2772static inline loff_t ext4_isize(struct super_block *sb,
2773 struct ext4_inode *raw_inode)
2774{
2775 if (ext4_has_feature_largedir(sb) ||
2776 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2777 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2778 le32_to_cpu(raw_inode->i_size_lo);
2779
2780 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2781}
2782
2783static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2784{
2785 raw_inode->i_size_lo = cpu_to_le32(i_size);
2786 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2787}
2788
2789static inline
2790struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2791 ext4_group_t group)
2792{
2793 struct ext4_group_info ***grp_info;
2794 long indexv, indexh;
2795 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2796 grp_info = EXT4_SB(sb)->s_group_info;
2797 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2798 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2799 return grp_info[indexv][indexh];
2800}
2801
2802/*
2803 * Reading s_groups_count requires using smp_rmb() afterwards. See
2804 * the locking protocol documented in the comments of ext4_group_add()
2805 * in resize.c
2806 */
2807static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2808{
2809 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2810
2811 smp_rmb();
2812 return ngroups;
2813}
2814
2815static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2816 ext4_group_t block_group)
2817{
2818 return block_group >> sbi->s_log_groups_per_flex;
2819}
2820
2821static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2822{
2823 return 1 << sbi->s_log_groups_per_flex;
2824}
2825
2826#define ext4_std_error(sb, errno) \
2827do { \
2828 if ((errno)) \
2829 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2830} while (0)
2831
2832#ifdef CONFIG_SMP
2833/* Each CPU can accumulate percpu_counter_batch clusters in their local
2834 * counters. So we need to make sure we have free clusters more
2835 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2836 */
2837#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2838#else
2839#define EXT4_FREECLUSTERS_WATERMARK 0
2840#endif
2841
2842/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2843static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2844{
2845 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2846 !inode_is_locked(inode));
2847 down_write(&EXT4_I(inode)->i_data_sem);
2848 if (newsize > EXT4_I(inode)->i_disksize)
2849 EXT4_I(inode)->i_disksize = newsize;
2850 up_write(&EXT4_I(inode)->i_data_sem);
2851}
2852
2853/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2854static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2855{
2856 int changed = 0;
2857
2858 if (newsize > inode->i_size) {
2859 i_size_write(inode, newsize);
2860 changed = 1;
2861 }
2862 if (newsize > EXT4_I(inode)->i_disksize) {
2863 ext4_update_i_disksize(inode, newsize);
2864 changed |= 2;
2865 }
2866 return changed;
2867}
2868
2869int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2870 loff_t len);
2871
2872struct ext4_group_info {
2873 unsigned long bb_state;
2874 struct rb_root bb_free_root;
2875 ext4_grpblk_t bb_first_free; /* first free block */
2876 ext4_grpblk_t bb_free; /* total free blocks */
2877 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2878 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2879 struct list_head bb_prealloc_list;
2880#ifdef DOUBLE_CHECK
2881 void *bb_bitmap;
2882#endif
2883 struct rw_semaphore alloc_sem;
2884 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2885 * regions, index is order.
2886 * bb_counters[3] = 5 means
2887 * 5 free 8-block regions. */
2888};
2889
2890#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2891#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2892#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2893#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2894#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
2895 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
2896#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
2897 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
2898
2899#define EXT4_MB_GRP_NEED_INIT(grp) \
2900 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2901#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2902 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2903#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2904 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2905
2906#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2907 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2908#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2909 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2910#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2911 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2912
2913#define EXT4_MAX_CONTENTION 8
2914#define EXT4_CONTENTION_THRESHOLD 2
2915
2916static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2917 ext4_group_t group)
2918{
2919 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2920}
2921
2922/*
2923 * Returns true if the filesystem is busy enough that attempts to
2924 * access the block group locks has run into contention.
2925 */
2926static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2927{
2928 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2929}
2930
2931static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2932{
2933 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2934 if (spin_trylock(lock))
2935 /*
2936 * We're able to grab the lock right away, so drop the
2937 * lock contention counter.
2938 */
2939 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2940 else {
2941 /*
2942 * The lock is busy, so bump the contention counter,
2943 * and then wait on the spin lock.
2944 */
2945 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2946 EXT4_MAX_CONTENTION);
2947 spin_lock(lock);
2948 }
2949}
2950
2951static inline void ext4_unlock_group(struct super_block *sb,
2952 ext4_group_t group)
2953{
2954 spin_unlock(ext4_group_lock_ptr(sb, group));
2955}
2956
2957/*
2958 * Block validity checking
2959 */
2960#define ext4_check_indirect_blockref(inode, bh) \
2961 ext4_check_blockref(__func__, __LINE__, inode, \
2962 (__le32 *)(bh)->b_data, \
2963 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2964
2965#define ext4_ind_check_inode(inode) \
2966 ext4_check_blockref(__func__, __LINE__, inode, \
2967 EXT4_I(inode)->i_data, \
2968 EXT4_NDIR_BLOCKS)
2969
2970/*
2971 * Inodes and files operations
2972 */
2973
2974/* dir.c */
2975extern const struct file_operations ext4_dir_operations;
2976
2977/* file.c */
2978extern const struct inode_operations ext4_file_inode_operations;
2979extern const struct file_operations ext4_file_operations;
2980extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2981
2982/* inline.c */
2983extern int ext4_get_max_inline_size(struct inode *inode);
2984extern int ext4_find_inline_data_nolock(struct inode *inode);
2985extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2986 unsigned int len);
2987extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2988
2989extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2990extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2991 struct inode *inode,
2992 loff_t pos, unsigned len,
2993 unsigned flags,
2994 struct page **pagep);
2995extern int ext4_write_inline_data_end(struct inode *inode,
2996 loff_t pos, unsigned len,
2997 unsigned copied,
2998 struct page *page);
2999extern struct buffer_head *
3000ext4_journalled_write_inline_data(struct inode *inode,
3001 unsigned len,
3002 struct page *page);
3003extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3004 struct inode *inode,
3005 loff_t pos, unsigned len,
3006 unsigned flags,
3007 struct page **pagep,
3008 void **fsdata);
3009extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3010 unsigned len, unsigned copied,
3011 struct page *page);
3012extern int ext4_try_add_inline_entry(handle_t *handle,
3013 struct ext4_filename *fname,
3014 struct inode *dir, struct inode *inode);
3015extern int ext4_try_create_inline_dir(handle_t *handle,
3016 struct inode *parent,
3017 struct inode *inode);
3018extern int ext4_read_inline_dir(struct file *filp,
3019 struct dir_context *ctx,
3020 int *has_inline_data);
3021extern int htree_inlinedir_to_tree(struct file *dir_file,
3022 struct inode *dir, ext4_lblk_t block,
3023 struct dx_hash_info *hinfo,
3024 __u32 start_hash, __u32 start_minor_hash,
3025 int *has_inline_data);
3026extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3027 struct ext4_filename *fname,
3028 struct ext4_dir_entry_2 **res_dir,
3029 int *has_inline_data);
3030extern int ext4_delete_inline_entry(handle_t *handle,
3031 struct inode *dir,
3032 struct ext4_dir_entry_2 *de_del,
3033 struct buffer_head *bh,
3034 int *has_inline_data);
3035extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3036extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3037 struct ext4_dir_entry_2 **parent_de,
3038 int *retval);
3039extern int ext4_inline_data_fiemap(struct inode *inode,
3040 struct fiemap_extent_info *fieinfo,
3041 int *has_inline, __u64 start, __u64 len);
3042
3043struct iomap;
3044extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3045
3046extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3047
3048extern int ext4_convert_inline_data(struct inode *inode);
3049
3050static inline int ext4_has_inline_data(struct inode *inode)
3051{
3052 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3053 EXT4_I(inode)->i_inline_off;
3054}
3055
3056/* namei.c */
3057extern const struct inode_operations ext4_dir_inode_operations;
3058extern const struct inode_operations ext4_special_inode_operations;
3059extern struct dentry *ext4_get_parent(struct dentry *child);
3060extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3061 struct ext4_dir_entry_2 *de,
3062 int blocksize, int csum_size,
3063 unsigned int parent_ino, int dotdot_real_len);
3064extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3065 unsigned int blocksize);
3066extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3067 struct inode *inode,
3068 struct buffer_head *bh);
3069#define S_SHIFT 12
3070static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3071 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3072 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3073 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3074 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3075 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3076 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3077 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3078};
3079
3080static inline void ext4_set_de_type(struct super_block *sb,
3081 struct ext4_dir_entry_2 *de,
3082 umode_t mode) {
3083 if (ext4_has_feature_filetype(sb))
3084 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3085}
3086
3087/* readpages.c */
3088extern int ext4_mpage_readpages(struct address_space *mapping,
3089 struct list_head *pages, struct page *page,
3090 unsigned nr_pages, bool is_readahead);
3091
3092/* symlink.c */
3093extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3094extern const struct inode_operations ext4_symlink_inode_operations;
3095extern const struct inode_operations ext4_fast_symlink_inode_operations;
3096
3097/* sysfs.c */
3098extern int ext4_register_sysfs(struct super_block *sb);
3099extern void ext4_unregister_sysfs(struct super_block *sb);
3100extern int __init ext4_init_sysfs(void);
3101extern void ext4_exit_sysfs(void);
3102
3103/* block_validity */
3104extern void ext4_release_system_zone(struct super_block *sb);
3105extern int ext4_setup_system_zone(struct super_block *sb);
3106extern int __init ext4_init_system_zone(void);
3107extern void ext4_exit_system_zone(void);
3108extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3109 ext4_fsblk_t start_blk,
3110 unsigned int count);
3111extern int ext4_check_blockref(const char *, unsigned int,
3112 struct inode *, __le32 *, unsigned int);
3113
3114/* extents.c */
3115struct ext4_ext_path;
3116struct ext4_extent;
3117
3118/*
3119 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3120 * __le32.
3121 */
3122#define EXT_MAX_BLOCKS 0xffffffff
3123
3124extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3125extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3126extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3127extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3128 struct ext4_map_blocks *map, int flags);
3129extern int ext4_ext_truncate(handle_t *, struct inode *);
3130extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3131 ext4_lblk_t end);
3132extern void ext4_ext_init(struct super_block *);
3133extern void ext4_ext_release(struct super_block *);
3134extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3135 loff_t len);
3136extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3137 loff_t offset, ssize_t len);
3138extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3139 struct ext4_map_blocks *map, int flags);
3140extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3141 ext4_lblk_t lblocks);
3142extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3143 int num,
3144 struct ext4_ext_path *path);
3145extern int ext4_can_extents_be_merged(struct inode *inode,
3146 struct ext4_extent *ex1,
3147 struct ext4_extent *ex2);
3148extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3149 struct ext4_ext_path **,
3150 struct ext4_extent *, int);
3151extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3152 struct ext4_ext_path **,
3153 int flags);
3154extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3155extern int ext4_ext_check_inode(struct inode *inode);
3156extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3157extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3158 __u64 start, __u64 len);
3159extern int ext4_ext_precache(struct inode *inode);
3160extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3161extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3162extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3163 struct inode *inode2, ext4_lblk_t lblk1,
3164 ext4_lblk_t lblk2, ext4_lblk_t count,
3165 int mark_unwritten,int *err);
3166extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3167
3168/* move_extent.c */
3169extern void ext4_double_down_write_data_sem(struct inode *first,
3170 struct inode *second);
3171extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3172 struct inode *donor_inode);
3173extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3174 __u64 start_orig, __u64 start_donor,
3175 __u64 len, __u64 *moved_len);
3176
3177/* page-io.c */
3178extern int __init ext4_init_pageio(void);
3179extern void ext4_exit_pageio(void);
3180extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3181extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3182extern int ext4_put_io_end(ext4_io_end_t *io_end);
3183extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3184extern void ext4_io_submit_init(struct ext4_io_submit *io,
3185 struct writeback_control *wbc);
3186extern void ext4_end_io_rsv_work(struct work_struct *work);
3187extern void ext4_io_submit(struct ext4_io_submit *io);
3188extern int ext4_bio_write_page(struct ext4_io_submit *io,
3189 struct page *page,
3190 int len,
3191 struct writeback_control *wbc,
3192 bool keep_towrite);
3193
3194/* mmp.c */
3195extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3196
3197/*
3198 * Add new method to test whether block and inode bitmaps are properly
3199 * initialized. With uninit_bg reading the block from disk is not enough
3200 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3201 */
3202#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3203
3204static inline int bitmap_uptodate(struct buffer_head *bh)
3205{
3206 return (buffer_uptodate(bh) &&
3207 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3208}
3209static inline void set_bitmap_uptodate(struct buffer_head *bh)
3210{
3211 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3212}
3213
3214#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3215
3216/* For ioend & aio unwritten conversion wait queues */
3217#define EXT4_WQ_HASH_SZ 37
3218#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3219 EXT4_WQ_HASH_SZ])
3220extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3221
3222extern int ext4_resize_begin(struct super_block *sb);
3223extern void ext4_resize_end(struct super_block *sb);
3224
3225static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3226 struct ext4_io_end *io_end)
3227{
3228 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3229 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3230 atomic_inc(&EXT4_I(inode)->i_unwritten);
3231 }
3232}
3233
3234static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3235{
3236 struct inode *inode = io_end->inode;
3237
3238 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3239 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3240 /* Wake up anyone waiting on unwritten extent conversion */
3241 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3242 wake_up_all(ext4_ioend_wq(inode));
3243 }
3244}
3245
3246extern const struct iomap_ops ext4_iomap_ops;
3247
3248#endif /* __KERNEL__ */
3249
3250#define EFSBADCRC EBADMSG /* Bad CRC detected */
3251#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3252
3253#endif /* _EXT4_H */
3254