Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
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#include <linux/fsverity.h>
45
46#include <linux/compiler.h>
47
48/*
49 * The fourth extended filesystem constants/structures
50 */
51
52/*
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
55 */
56#define AGGRESSIVE_CHECK__
57
58/*
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
61 */
62#define DOUBLE_CHECK__
63
64/*
65 * Define EXT4FS_DEBUG to produce debug messages
66 */
67#undef EXT4FS_DEBUG
68
69/*
70 * Debug code
71 */
72#ifdef EXT4FS_DEBUG
73#define ext4_debug(f, a...) \
74 do { \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
78 } while (0)
79#else
80#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
81#endif
82
83/*
84 * Turn on EXT_DEBUG to get lots of info about extents operations.
85 */
86#define EXT_DEBUG__
87#ifdef EXT_DEBUG
88#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
89#else
90#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
91#endif
92
93/* data type for block offset of block group */
94typedef int ext4_grpblk_t;
95
96/* data type for filesystem-wide blocks number */
97typedef unsigned long long ext4_fsblk_t;
98
99/* data type for file logical block number */
100typedef __u32 ext4_lblk_t;
101
102/* data type for block group number */
103typedef unsigned int ext4_group_t;
104
105enum SHIFT_DIRECTION {
106 SHIFT_LEFT = 0,
107 SHIFT_RIGHT,
108};
109
110/*
111 * Flags used in mballoc's allocation_context flags field.
112 *
113 * Also used to show what's going on for debugging purposes when the
114 * flag field is exported via the traceport interface
115 */
116
117/* prefer goal again. length */
118#define EXT4_MB_HINT_MERGE 0x0001
119/* blocks already reserved */
120#define EXT4_MB_HINT_RESERVED 0x0002
121/* metadata is being allocated */
122#define EXT4_MB_HINT_METADATA 0x0004
123/* first blocks in the file */
124#define EXT4_MB_HINT_FIRST 0x0008
125/* search for the best chunk */
126#define EXT4_MB_HINT_BEST 0x0010
127/* data is being allocated */
128#define EXT4_MB_HINT_DATA 0x0020
129/* don't preallocate (for tails) */
130#define EXT4_MB_HINT_NOPREALLOC 0x0040
131/* allocate for locality group */
132#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
133/* allocate goal blocks or none */
134#define EXT4_MB_HINT_GOAL_ONLY 0x0100
135/* goal is meaningful */
136#define EXT4_MB_HINT_TRY_GOAL 0x0200
137/* blocks already pre-reserved by delayed allocation */
138#define EXT4_MB_DELALLOC_RESERVED 0x0400
139/* We are doing stream allocation */
140#define EXT4_MB_STREAM_ALLOC 0x0800
141/* Use reserved root blocks if needed */
142#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
143/* Use blocks from reserved pool */
144#define EXT4_MB_USE_RESERVED 0x2000
145
146struct ext4_allocation_request {
147 /* target inode for block we're allocating */
148 struct inode *inode;
149 /* how many blocks we want to allocate */
150 unsigned int len;
151 /* logical block in target inode */
152 ext4_lblk_t logical;
153 /* the closest logical allocated block to the left */
154 ext4_lblk_t lleft;
155 /* the closest logical allocated block to the right */
156 ext4_lblk_t lright;
157 /* phys. target (a hint) */
158 ext4_fsblk_t goal;
159 /* phys. block for the closest logical allocated block to the left */
160 ext4_fsblk_t pleft;
161 /* phys. block for the closest logical allocated block to the right */
162 ext4_fsblk_t pright;
163 /* flags. see above EXT4_MB_HINT_* */
164 unsigned int flags;
165};
166
167/*
168 * Logical to physical block mapping, used by ext4_map_blocks()
169 *
170 * This structure is used to pass requests into ext4_map_blocks() as
171 * well as to store the information returned by ext4_map_blocks(). It
172 * takes less room on the stack than a struct buffer_head.
173 */
174#define EXT4_MAP_NEW (1 << BH_New)
175#define EXT4_MAP_MAPPED (1 << BH_Mapped)
176#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
177#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
178#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
180
181struct ext4_map_blocks {
182 ext4_fsblk_t m_pblk;
183 ext4_lblk_t m_lblk;
184 unsigned int m_len;
185 unsigned int m_flags;
186};
187
188/*
189 * Block validity checking, system zone rbtree.
190 */
191struct ext4_system_blocks {
192 struct rb_root root;
193 struct rcu_head rcu;
194};
195
196/*
197 * Flags for ext4_io_end->flags
198 */
199#define EXT4_IO_END_UNWRITTEN 0x0001
200
201struct ext4_io_end_vec {
202 struct list_head list; /* list of io_end_vec */
203 loff_t offset; /* offset in the file */
204 ssize_t size; /* size of the extent */
205};
206
207/*
208 * For converting unwritten extents on a work queue. 'handle' is used for
209 * buffered writeback.
210 */
211typedef struct ext4_io_end {
212 struct list_head list; /* per-file finished IO list */
213 handle_t *handle; /* handle reserved for extent
214 * conversion */
215 struct inode *inode; /* file being written to */
216 struct bio *bio; /* Linked list of completed
217 * bios covering the extent */
218 unsigned int flag; /* unwritten or not */
219 atomic_t count; /* reference counter */
220 struct list_head list_vec; /* list of ext4_io_end_vec */
221} ext4_io_end_t;
222
223struct ext4_io_submit {
224 struct writeback_control *io_wbc;
225 struct bio *io_bio;
226 ext4_io_end_t *io_end;
227 sector_t io_next_block;
228};
229
230/*
231 * Special inodes numbers
232 */
233#define EXT4_BAD_INO 1 /* Bad blocks inode */
234#define EXT4_ROOT_INO 2 /* Root inode */
235#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
236#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
237#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
238#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
239#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
240#define EXT4_JOURNAL_INO 8 /* Journal inode */
241
242/* First non-reserved inode for old ext4 filesystems */
243#define EXT4_GOOD_OLD_FIRST_INO 11
244
245/*
246 * Maximal count of links to a file
247 */
248#define EXT4_LINK_MAX 65000
249
250/*
251 * Macro-instructions used to manage several block sizes
252 */
253#define EXT4_MIN_BLOCK_SIZE 1024
254#define EXT4_MAX_BLOCK_SIZE 65536
255#define EXT4_MIN_BLOCK_LOG_SIZE 10
256#define EXT4_MAX_BLOCK_LOG_SIZE 16
257#define EXT4_MAX_CLUSTER_LOG_SIZE 30
258#ifdef __KERNEL__
259# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
260#else
261# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
262#endif
263#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
264#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
265 EXT4_SB(s)->s_cluster_bits)
266#ifdef __KERNEL__
267# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
268# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
269#else
270# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
271#endif
272#ifdef __KERNEL__
273#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
274#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
275#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
276#else
277#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
278 EXT4_GOOD_OLD_INODE_SIZE : \
279 (s)->s_inode_size)
280#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
281 EXT4_GOOD_OLD_FIRST_INO : \
282 (s)->s_first_ino)
283#endif
284#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
285#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
286 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
287 blkbits))
288
289/* Translate a block number to a cluster number */
290#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
291/* Translate a cluster number to a block number */
292#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
293/* Translate # of blks to # of clusters */
294#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
295 (sbi)->s_cluster_bits)
296/* Mask out the low bits to get the starting block of the cluster */
297#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
298 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
299#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
300 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
301/* Fill in the low bits to get the last block of the cluster */
302#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
303 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
304/* Get the cluster offset */
305#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
306 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
307#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
308 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
309
310/*
311 * Structure of a blocks group descriptor
312 */
313struct ext4_group_desc
314{
315 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
316 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
317 __le32 bg_inode_table_lo; /* Inodes table block */
318 __le16 bg_free_blocks_count_lo;/* Free blocks count */
319 __le16 bg_free_inodes_count_lo;/* Free inodes count */
320 __le16 bg_used_dirs_count_lo; /* Directories count */
321 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
322 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
323 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
324 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
325 __le16 bg_itable_unused_lo; /* Unused inodes count */
326 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
327 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
328 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
329 __le32 bg_inode_table_hi; /* Inodes table block MSB */
330 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
331 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
332 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
333 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
334 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
335 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
336 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
337 __u32 bg_reserved;
338};
339
340#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
341 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
342 sizeof(__le16))
343#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
344 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
345 sizeof(__le16))
346
347/*
348 * Structure of a flex block group info
349 */
350
351struct flex_groups {
352 atomic64_t free_clusters;
353 atomic_t free_inodes;
354 atomic_t used_dirs;
355};
356
357#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
358#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
359#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
360
361/*
362 * Macro-instructions used to manage group descriptors
363 */
364#define EXT4_MIN_DESC_SIZE 32
365#define EXT4_MIN_DESC_SIZE_64BIT 64
366#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
367#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
368#ifdef __KERNEL__
369# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
370# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
371# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
372# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
373# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
374#else
375# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
376# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
377# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
378#endif
379
380/*
381 * Constants relative to the data blocks
382 */
383#define EXT4_NDIR_BLOCKS 12
384#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
385#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
386#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
387#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
388
389/*
390 * Inode flags
391 */
392#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
393#define EXT4_UNRM_FL 0x00000002 /* Undelete */
394#define EXT4_COMPR_FL 0x00000004 /* Compress file */
395#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
396#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
397#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
398#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
399#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
400/* Reserved for compression usage... */
401#define EXT4_DIRTY_FL 0x00000100
402#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
403#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
404 /* nb: was previously EXT2_ECOMPR_FL */
405#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
406/* End compression flags --- maybe not all used */
407#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
408#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
409#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
410#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
411#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
412#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
413#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
414#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
415#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
416#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
417#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
418#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
419#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
420#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */
421#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
422
423#define EXT4_FL_USER_VISIBLE 0x705BDFFF /* User visible flags */
424#define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */
425
426/* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
427#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
428 EXT4_IMMUTABLE_FL | \
429 EXT4_APPEND_FL | \
430 EXT4_NODUMP_FL | \
431 EXT4_NOATIME_FL | \
432 EXT4_PROJINHERIT_FL)
433
434/* Flags that should be inherited by new inodes from their parent. */
435#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
436 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
437 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
438 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
439 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
440
441/* Flags that are appropriate for regular files (all but dir-specific ones). */
442#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
443 EXT4_PROJINHERIT_FL))
444
445/* Flags that are appropriate for non-directories/regular files. */
446#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
447
448/* The only flags that should be swapped */
449#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
450
451/* Mask out flags that are inappropriate for the given type of inode. */
452static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
453{
454 if (S_ISDIR(mode))
455 return flags;
456 else if (S_ISREG(mode))
457 return flags & EXT4_REG_FLMASK;
458 else
459 return flags & EXT4_OTHER_FLMASK;
460}
461
462/*
463 * Inode flags used for atomic set/get
464 */
465enum {
466 EXT4_INODE_SECRM = 0, /* Secure deletion */
467 EXT4_INODE_UNRM = 1, /* Undelete */
468 EXT4_INODE_COMPR = 2, /* Compress file */
469 EXT4_INODE_SYNC = 3, /* Synchronous updates */
470 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
471 EXT4_INODE_APPEND = 5, /* writes to file may only append */
472 EXT4_INODE_NODUMP = 6, /* do not dump file */
473 EXT4_INODE_NOATIME = 7, /* do not update atime */
474/* Reserved for compression usage... */
475 EXT4_INODE_DIRTY = 8,
476 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
477 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
478 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
479/* End compression flags --- maybe not all used */
480 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
481 EXT4_INODE_IMAGIC = 13, /* AFS directory */
482 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
483 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
484 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
485 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
486 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
487 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
488 EXT4_INODE_VERITY = 20, /* Verity protected inode */
489 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
490 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
491 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
492 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
493 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
494};
495
496/*
497 * Since it's pretty easy to mix up bit numbers and hex values, we use a
498 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
499 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
500 * any extra space in the compiled kernel image, otherwise, the build will fail.
501 * It's important that these values are the same, since we are using
502 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
503 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
504 * values found in ext2, ext3 and ext4 filesystems, and of course the values
505 * defined in e2fsprogs.
506 *
507 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
508 */
509#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
510#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
511
512static inline void ext4_check_flag_values(void)
513{
514 CHECK_FLAG_VALUE(SECRM);
515 CHECK_FLAG_VALUE(UNRM);
516 CHECK_FLAG_VALUE(COMPR);
517 CHECK_FLAG_VALUE(SYNC);
518 CHECK_FLAG_VALUE(IMMUTABLE);
519 CHECK_FLAG_VALUE(APPEND);
520 CHECK_FLAG_VALUE(NODUMP);
521 CHECK_FLAG_VALUE(NOATIME);
522 CHECK_FLAG_VALUE(DIRTY);
523 CHECK_FLAG_VALUE(COMPRBLK);
524 CHECK_FLAG_VALUE(NOCOMPR);
525 CHECK_FLAG_VALUE(ENCRYPT);
526 CHECK_FLAG_VALUE(INDEX);
527 CHECK_FLAG_VALUE(IMAGIC);
528 CHECK_FLAG_VALUE(JOURNAL_DATA);
529 CHECK_FLAG_VALUE(NOTAIL);
530 CHECK_FLAG_VALUE(DIRSYNC);
531 CHECK_FLAG_VALUE(TOPDIR);
532 CHECK_FLAG_VALUE(HUGE_FILE);
533 CHECK_FLAG_VALUE(EXTENTS);
534 CHECK_FLAG_VALUE(VERITY);
535 CHECK_FLAG_VALUE(EA_INODE);
536 CHECK_FLAG_VALUE(EOFBLOCKS);
537 CHECK_FLAG_VALUE(INLINE_DATA);
538 CHECK_FLAG_VALUE(PROJINHERIT);
539 CHECK_FLAG_VALUE(RESERVED);
540}
541
542/* Used to pass group descriptor data when online resize is done */
543struct ext4_new_group_input {
544 __u32 group; /* Group number for this data */
545 __u64 block_bitmap; /* Absolute block number of block bitmap */
546 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
547 __u64 inode_table; /* Absolute block number of inode table start */
548 __u32 blocks_count; /* Total number of blocks in this group */
549 __u16 reserved_blocks; /* Number of reserved blocks in this group */
550 __u16 unused;
551};
552
553#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
554struct compat_ext4_new_group_input {
555 u32 group;
556 compat_u64 block_bitmap;
557 compat_u64 inode_bitmap;
558 compat_u64 inode_table;
559 u32 blocks_count;
560 u16 reserved_blocks;
561 u16 unused;
562};
563#endif
564
565/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
566struct ext4_new_group_data {
567 __u32 group;
568 __u64 block_bitmap;
569 __u64 inode_bitmap;
570 __u64 inode_table;
571 __u32 blocks_count;
572 __u16 reserved_blocks;
573 __u16 mdata_blocks;
574 __u32 free_clusters_count;
575};
576
577/* Indexes used to index group tables in ext4_new_group_data */
578enum {
579 BLOCK_BITMAP = 0, /* block bitmap */
580 INODE_BITMAP, /* inode bitmap */
581 INODE_TABLE, /* inode tables */
582 GROUP_TABLE_COUNT,
583};
584
585/*
586 * Flags used by ext4_map_blocks()
587 */
588 /* Allocate any needed blocks and/or convert an unwritten
589 extent to be an initialized ext4 */
590#define EXT4_GET_BLOCKS_CREATE 0x0001
591 /* Request the creation of an unwritten extent */
592#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
593#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
594 EXT4_GET_BLOCKS_CREATE)
595 /* Caller is from the delayed allocation writeout path
596 * finally doing the actual allocation of delayed blocks */
597#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
598 /* caller is from the direct IO path, request to creation of an
599 unwritten extents if not allocated, split the unwritten
600 extent if blocks has been preallocated already*/
601#define EXT4_GET_BLOCKS_PRE_IO 0x0008
602#define EXT4_GET_BLOCKS_CONVERT 0x0010
603#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
604 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
605 /* Convert extent to initialized after IO complete */
606#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
607 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
608 /* Eventual metadata allocation (due to growing extent tree)
609 * should not fail, so try to use reserved blocks for that.*/
610#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
611 /* Don't normalize allocation size (used for fallocate) */
612#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
613 /* Request will not result in inode size update (user for fallocate) */
614#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
615 /* Convert written extents to unwritten */
616#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
617 /* Write zeros to newly created written extents */
618#define EXT4_GET_BLOCKS_ZERO 0x0200
619#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
620 EXT4_GET_BLOCKS_ZERO)
621 /* Caller will submit data before dropping transaction handle. This
622 * allows jbd2 to avoid submitting data before commit. */
623#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
624
625/*
626 * The bit position of these flags must not overlap with any of the
627 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
628 * read_extent_tree_block(), ext4_split_extent_at(),
629 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
630 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
631 * caching the extents when reading from the extent tree while a
632 * truncate or punch hole operation is in progress.
633 */
634#define EXT4_EX_NOCACHE 0x40000000
635#define EXT4_EX_FORCE_CACHE 0x20000000
636
637/*
638 * Flags used by ext4_free_blocks
639 */
640#define EXT4_FREE_BLOCKS_METADATA 0x0001
641#define EXT4_FREE_BLOCKS_FORGET 0x0002
642#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
643#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
644#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
645#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
646#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
647
648/*
649 * ioctl commands
650 */
651#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
652#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
653#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
654#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
655#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
656#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
657#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
658#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
659#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
660#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
661#define EXT4_IOC_MIGRATE _IO('f', 9)
662 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
663 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
664#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
665#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
666#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
667#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
668#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
669#define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
670#define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
671#define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
672/* ioctl codes 19--39 are reserved for fscrypt */
673#define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
674#define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
675#define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
676
677#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
678#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
679
680#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
681
682/*
683 * Flags for going down operation
684 */
685#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
686#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
687#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
688
689/*
690 * Flags returned by EXT4_IOC_GETSTATE
691 *
692 * We only expose to userspace a subset of the state flags in
693 * i_state_flags
694 */
695#define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
696#define EXT4_STATE_FLAG_NEW 0x00000002
697#define EXT4_STATE_FLAG_NEWENTRY 0x00000004
698#define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
699
700#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
701/*
702 * ioctl commands in 32 bit emulation
703 */
704#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
705#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
706#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
707#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
708#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
709#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
710#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
711#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
712#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
713#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
714#endif
715
716/*
717 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
718 * It indicates that the entry in extent status cache is for a hole.
719 */
720#define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
721
722/* Max physical block we can address w/o extents */
723#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
724
725/* Max logical block we can support */
726#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
727
728/*
729 * Structure of an inode on the disk
730 */
731struct ext4_inode {
732 __le16 i_mode; /* File mode */
733 __le16 i_uid; /* Low 16 bits of Owner Uid */
734 __le32 i_size_lo; /* Size in bytes */
735 __le32 i_atime; /* Access time */
736 __le32 i_ctime; /* Inode Change time */
737 __le32 i_mtime; /* Modification time */
738 __le32 i_dtime; /* Deletion Time */
739 __le16 i_gid; /* Low 16 bits of Group Id */
740 __le16 i_links_count; /* Links count */
741 __le32 i_blocks_lo; /* Blocks count */
742 __le32 i_flags; /* File flags */
743 union {
744 struct {
745 __le32 l_i_version;
746 } linux1;
747 struct {
748 __u32 h_i_translator;
749 } hurd1;
750 struct {
751 __u32 m_i_reserved1;
752 } masix1;
753 } osd1; /* OS dependent 1 */
754 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
755 __le32 i_generation; /* File version (for NFS) */
756 __le32 i_file_acl_lo; /* File ACL */
757 __le32 i_size_high;
758 __le32 i_obso_faddr; /* Obsoleted fragment address */
759 union {
760 struct {
761 __le16 l_i_blocks_high; /* were l_i_reserved1 */
762 __le16 l_i_file_acl_high;
763 __le16 l_i_uid_high; /* these 2 fields */
764 __le16 l_i_gid_high; /* were reserved2[0] */
765 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
766 __le16 l_i_reserved;
767 } linux2;
768 struct {
769 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
770 __u16 h_i_mode_high;
771 __u16 h_i_uid_high;
772 __u16 h_i_gid_high;
773 __u32 h_i_author;
774 } hurd2;
775 struct {
776 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
777 __le16 m_i_file_acl_high;
778 __u32 m_i_reserved2[2];
779 } masix2;
780 } osd2; /* OS dependent 2 */
781 __le16 i_extra_isize;
782 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
783 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
784 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
785 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
786 __le32 i_crtime; /* File Creation time */
787 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
788 __le32 i_version_hi; /* high 32 bits for 64-bit version */
789 __le32 i_projid; /* Project ID */
790};
791
792struct move_extent {
793 __u32 reserved; /* should be zero */
794 __u32 donor_fd; /* donor file descriptor */
795 __u64 orig_start; /* logical start offset in block for orig */
796 __u64 donor_start; /* logical start offset in block for donor */
797 __u64 len; /* block length to be moved */
798 __u64 moved_len; /* moved block length */
799};
800
801#define EXT4_EPOCH_BITS 2
802#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
803#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
804
805/*
806 * Extended fields will fit into an inode if the filesystem was formatted
807 * with large inodes (-I 256 or larger) and there are not currently any EAs
808 * consuming all of the available space. For new inodes we always reserve
809 * enough space for the kernel's known extended fields, but for inodes
810 * created with an old kernel this might not have been the case. None of
811 * the extended inode fields is critical for correct filesystem operation.
812 * This macro checks if a certain field fits in the inode. Note that
813 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
814 */
815#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
816 ((offsetof(typeof(*ext4_inode), field) + \
817 sizeof((ext4_inode)->field)) \
818 <= (EXT4_GOOD_OLD_INODE_SIZE + \
819 (einode)->i_extra_isize)) \
820
821/*
822 * We use an encoding that preserves the times for extra epoch "00":
823 *
824 * extra msb of adjust for signed
825 * epoch 32-bit 32-bit tv_sec to
826 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
827 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
828 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
829 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
830 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
831 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
832 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
833 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
834 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
835 *
836 * Note that previous versions of the kernel on 64-bit systems would
837 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
838 * 1970. e2fsck will correct this, assuming that it is run on the
839 * affected filesystem before 2242.
840 */
841
842static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
843{
844 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
845 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
846}
847
848static inline void ext4_decode_extra_time(struct timespec64 *time,
849 __le32 extra)
850{
851 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
852 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
853 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
854}
855
856#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
857do { \
858 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
859 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
860 (raw_inode)->xtime ## _extra = \
861 ext4_encode_extra_time(&(inode)->xtime); \
862 } \
863 else \
864 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
865} while (0)
866
867#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
868do { \
869 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
870 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
871 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
872 (raw_inode)->xtime ## _extra = \
873 ext4_encode_extra_time(&(einode)->xtime); \
874} while (0)
875
876#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
877do { \
878 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
879 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
880 ext4_decode_extra_time(&(inode)->xtime, \
881 raw_inode->xtime ## _extra); \
882 } \
883 else \
884 (inode)->xtime.tv_nsec = 0; \
885} while (0)
886
887
888#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
889do { \
890 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
891 (einode)->xtime.tv_sec = \
892 (signed)le32_to_cpu((raw_inode)->xtime); \
893 else \
894 (einode)->xtime.tv_sec = 0; \
895 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
896 ext4_decode_extra_time(&(einode)->xtime, \
897 raw_inode->xtime ## _extra); \
898 else \
899 (einode)->xtime.tv_nsec = 0; \
900} while (0)
901
902#define i_disk_version osd1.linux1.l_i_version
903
904#if defined(__KERNEL__) || defined(__linux__)
905#define i_reserved1 osd1.linux1.l_i_reserved1
906#define i_file_acl_high osd2.linux2.l_i_file_acl_high
907#define i_blocks_high osd2.linux2.l_i_blocks_high
908#define i_uid_low i_uid
909#define i_gid_low i_gid
910#define i_uid_high osd2.linux2.l_i_uid_high
911#define i_gid_high osd2.linux2.l_i_gid_high
912#define i_checksum_lo osd2.linux2.l_i_checksum_lo
913
914#elif defined(__GNU__)
915
916#define i_translator osd1.hurd1.h_i_translator
917#define i_uid_high osd2.hurd2.h_i_uid_high
918#define i_gid_high osd2.hurd2.h_i_gid_high
919#define i_author osd2.hurd2.h_i_author
920
921#elif defined(__masix__)
922
923#define i_reserved1 osd1.masix1.m_i_reserved1
924#define i_file_acl_high osd2.masix2.m_i_file_acl_high
925#define i_reserved2 osd2.masix2.m_i_reserved2
926
927#endif /* defined(__KERNEL__) || defined(__linux__) */
928
929#include "extents_status.h"
930
931/*
932 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
933 *
934 * These are needed to avoid lockdep false positives when we need to
935 * allocate blocks to the quota inode during ext4_map_blocks(), while
936 * holding i_data_sem for a normal (non-quota) inode. Since we don't
937 * do quota tracking for the quota inode, this avoids deadlock (as
938 * well as infinite recursion, since it isn't turtles all the way
939 * down...)
940 *
941 * I_DATA_SEM_NORMAL - Used for most inodes
942 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
943 * where the second inode has larger inode number
944 * than the first
945 * I_DATA_SEM_QUOTA - Used for quota inodes only
946 */
947enum {
948 I_DATA_SEM_NORMAL = 0,
949 I_DATA_SEM_OTHER,
950 I_DATA_SEM_QUOTA,
951};
952
953
954/*
955 * fourth extended file system inode data in memory
956 */
957struct ext4_inode_info {
958 __le32 i_data[15]; /* unconverted */
959 __u32 i_dtime;
960 ext4_fsblk_t i_file_acl;
961
962 /*
963 * i_block_group is the number of the block group which contains
964 * this file's inode. Constant across the lifetime of the inode,
965 * it is used for making block allocation decisions - we try to
966 * place a file's data blocks near its inode block, and new inodes
967 * near to their parent directory's inode.
968 */
969 ext4_group_t i_block_group;
970 ext4_lblk_t i_dir_start_lookup;
971#if (BITS_PER_LONG < 64)
972 unsigned long i_state_flags; /* Dynamic state flags */
973#endif
974 unsigned long i_flags;
975
976 /*
977 * Extended attributes can be read independently of the main file
978 * data. Taking i_mutex even when reading would cause contention
979 * between readers of EAs and writers of regular file data, so
980 * instead we synchronize on xattr_sem when reading or changing
981 * EAs.
982 */
983 struct rw_semaphore xattr_sem;
984
985 struct list_head i_orphan; /* unlinked but open inodes */
986
987 /*
988 * i_disksize keeps track of what the inode size is ON DISK, not
989 * in memory. During truncate, i_size is set to the new size by
990 * the VFS prior to calling ext4_truncate(), but the filesystem won't
991 * set i_disksize to 0 until the truncate is actually under way.
992 *
993 * The intent is that i_disksize always represents the blocks which
994 * are used by this file. This allows recovery to restart truncate
995 * on orphans if we crash during truncate. We actually write i_disksize
996 * into the on-disk inode when writing inodes out, instead of i_size.
997 *
998 * The only time when i_disksize and i_size may be different is when
999 * a truncate is in progress. The only things which change i_disksize
1000 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1001 */
1002 loff_t i_disksize;
1003
1004 /*
1005 * i_data_sem is for serialising ext4_truncate() against
1006 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1007 * data tree are chopped off during truncate. We can't do that in
1008 * ext4 because whenever we perform intermediate commits during
1009 * truncate, the inode and all the metadata blocks *must* be in a
1010 * consistent state which allows truncation of the orphans to restart
1011 * during recovery. Hence we must fix the get_block-vs-truncate race
1012 * by other means, so we have i_data_sem.
1013 */
1014 struct rw_semaphore i_data_sem;
1015 /*
1016 * i_mmap_sem is for serializing page faults with truncate / punch hole
1017 * operations. We have to make sure that new page cannot be faulted in
1018 * a section of the inode that is being punched. We cannot easily use
1019 * i_data_sem for this since we need protection for the whole punch
1020 * operation and i_data_sem ranks below transaction start so we have
1021 * to occasionally drop it.
1022 */
1023 struct rw_semaphore i_mmap_sem;
1024 struct inode vfs_inode;
1025 struct jbd2_inode *jinode;
1026
1027 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1028
1029 /*
1030 * File creation time. Its function is same as that of
1031 * struct timespec64 i_{a,c,m}time in the generic inode.
1032 */
1033 struct timespec64 i_crtime;
1034
1035 /* mballoc */
1036 struct list_head i_prealloc_list;
1037 spinlock_t i_prealloc_lock;
1038
1039 /* extents status tree */
1040 struct ext4_es_tree i_es_tree;
1041 rwlock_t i_es_lock;
1042 struct list_head i_es_list;
1043 unsigned int i_es_all_nr; /* protected by i_es_lock */
1044 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1045 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1046 extents to shrink. Protected by
1047 i_es_lock */
1048
1049 /* ialloc */
1050 ext4_group_t i_last_alloc_group;
1051
1052 /* allocation reservation info for delalloc */
1053 /* In case of bigalloc, this refer to clusters rather than blocks */
1054 unsigned int i_reserved_data_blocks;
1055 ext4_lblk_t i_da_metadata_calc_last_lblock;
1056 int i_da_metadata_calc_len;
1057
1058 /* pending cluster reservations for bigalloc file systems */
1059 struct ext4_pending_tree i_pending_tree;
1060
1061 /* on-disk additional length */
1062 __u16 i_extra_isize;
1063
1064 /* Indicate the inline data space. */
1065 u16 i_inline_off;
1066 u16 i_inline_size;
1067
1068#ifdef CONFIG_QUOTA
1069 /* quota space reservation, managed internally by quota code */
1070 qsize_t i_reserved_quota;
1071#endif
1072
1073 /* Lock protecting lists below */
1074 spinlock_t i_completed_io_lock;
1075 /*
1076 * Completed IOs that need unwritten extents handling and have
1077 * transaction reserved
1078 */
1079 struct list_head i_rsv_conversion_list;
1080 struct work_struct i_rsv_conversion_work;
1081 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1082
1083 spinlock_t i_block_reservation_lock;
1084
1085 /*
1086 * Transactions that contain inode's metadata needed to complete
1087 * fsync and fdatasync, respectively.
1088 */
1089 tid_t i_sync_tid;
1090 tid_t i_datasync_tid;
1091
1092#ifdef CONFIG_QUOTA
1093 struct dquot *i_dquot[MAXQUOTAS];
1094#endif
1095
1096 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1097 __u32 i_csum_seed;
1098
1099 kprojid_t i_projid;
1100};
1101
1102/*
1103 * File system states
1104 */
1105#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1106#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1107#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1108
1109/*
1110 * Misc. filesystem flags
1111 */
1112#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1113#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1114#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1115
1116/*
1117 * Mount flags set via mount options or defaults
1118 */
1119#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1120#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1121#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1122#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1123#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1124#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1125#define EXT4_MOUNT_ERRORS_MASK 0x00070
1126#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1127#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1128#ifdef CONFIG_FS_DAX
1129#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1130#else
1131#define EXT4_MOUNT_DAX 0
1132#endif
1133#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1134#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1135#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1136#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1137#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1138#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1139#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1140#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1141#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1142#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1143#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1144#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1145 * enable enforcement for hidden
1146 * quota files */
1147#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1148 * enforcement for hidden quota
1149 * files */
1150#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1151 * enforcement */
1152#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1153#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1154#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1155#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1156#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1157#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1158#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1159#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1160#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1161
1162/*
1163 * Mount flags set either automatically (could not be set by mount option)
1164 * based on per file system feature or property or in special cases such as
1165 * distinguishing between explicit mount option definition and default.
1166 */
1167#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1168 specified delalloc */
1169#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1170 size of blocksize * 8
1171 blocks */
1172#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1173 file systems */
1174
1175#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1176 specified journal checksum */
1177
1178#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1179 ~EXT4_MOUNT_##opt
1180#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1181 EXT4_MOUNT_##opt
1182#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1183 EXT4_MOUNT_##opt)
1184
1185#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1186 ~EXT4_MOUNT2_##opt
1187#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1188 EXT4_MOUNT2_##opt
1189#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1190 EXT4_MOUNT2_##opt)
1191
1192#define ext4_test_and_set_bit __test_and_set_bit_le
1193#define ext4_set_bit __set_bit_le
1194#define ext4_set_bit_atomic ext2_set_bit_atomic
1195#define ext4_test_and_clear_bit __test_and_clear_bit_le
1196#define ext4_clear_bit __clear_bit_le
1197#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1198#define ext4_test_bit test_bit_le
1199#define ext4_find_next_zero_bit find_next_zero_bit_le
1200#define ext4_find_next_bit find_next_bit_le
1201
1202extern void ext4_set_bits(void *bm, int cur, int len);
1203
1204/*
1205 * Maximal mount counts between two filesystem checks
1206 */
1207#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1208#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1209
1210/*
1211 * Behaviour when detecting errors
1212 */
1213#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1214#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1215#define EXT4_ERRORS_PANIC 3 /* Panic */
1216#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1217
1218/* Metadata checksum algorithm codes */
1219#define EXT4_CRC32C_CHKSUM 1
1220
1221/*
1222 * Structure of the super block
1223 */
1224struct ext4_super_block {
1225/*00*/ __le32 s_inodes_count; /* Inodes count */
1226 __le32 s_blocks_count_lo; /* Blocks count */
1227 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1228 __le32 s_free_blocks_count_lo; /* Free blocks count */
1229/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1230 __le32 s_first_data_block; /* First Data Block */
1231 __le32 s_log_block_size; /* Block size */
1232 __le32 s_log_cluster_size; /* Allocation cluster size */
1233/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1234 __le32 s_clusters_per_group; /* # Clusters per group */
1235 __le32 s_inodes_per_group; /* # Inodes per group */
1236 __le32 s_mtime; /* Mount time */
1237/*30*/ __le32 s_wtime; /* Write time */
1238 __le16 s_mnt_count; /* Mount count */
1239 __le16 s_max_mnt_count; /* Maximal mount count */
1240 __le16 s_magic; /* Magic signature */
1241 __le16 s_state; /* File system state */
1242 __le16 s_errors; /* Behaviour when detecting errors */
1243 __le16 s_minor_rev_level; /* minor revision level */
1244/*40*/ __le32 s_lastcheck; /* time of last check */
1245 __le32 s_checkinterval; /* max. time between checks */
1246 __le32 s_creator_os; /* OS */
1247 __le32 s_rev_level; /* Revision level */
1248/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1249 __le16 s_def_resgid; /* Default gid for reserved blocks */
1250 /*
1251 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1252 *
1253 * Note: the difference between the compatible feature set and
1254 * the incompatible feature set is that if there is a bit set
1255 * in the incompatible feature set that the kernel doesn't
1256 * know about, it should refuse to mount the filesystem.
1257 *
1258 * e2fsck's requirements are more strict; if it doesn't know
1259 * about a feature in either the compatible or incompatible
1260 * feature set, it must abort and not try to meddle with
1261 * things it doesn't understand...
1262 */
1263 __le32 s_first_ino; /* First non-reserved inode */
1264 __le16 s_inode_size; /* size of inode structure */
1265 __le16 s_block_group_nr; /* block group # of this superblock */
1266 __le32 s_feature_compat; /* compatible feature set */
1267/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1268 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1269/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1270/*78*/ char s_volume_name[16]; /* volume name */
1271/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1272/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1273 /*
1274 * Performance hints. Directory preallocation should only
1275 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1276 */
1277 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1278 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1279 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1280 /*
1281 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1282 */
1283/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1284/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1285 __le32 s_journal_dev; /* device number of journal file */
1286 __le32 s_last_orphan; /* start of list of inodes to delete */
1287 __le32 s_hash_seed[4]; /* HTREE hash seed */
1288 __u8 s_def_hash_version; /* Default hash version to use */
1289 __u8 s_jnl_backup_type;
1290 __le16 s_desc_size; /* size of group descriptor */
1291/*100*/ __le32 s_default_mount_opts;
1292 __le32 s_first_meta_bg; /* First metablock block group */
1293 __le32 s_mkfs_time; /* When the filesystem was created */
1294 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1295 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1296/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1297 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1298 __le32 s_free_blocks_count_hi; /* Free blocks count */
1299 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1300 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1301 __le32 s_flags; /* Miscellaneous flags */
1302 __le16 s_raid_stride; /* RAID stride */
1303 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1304 __le64 s_mmp_block; /* Block for multi-mount protection */
1305 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1306 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1307 __u8 s_checksum_type; /* metadata checksum algorithm used */
1308 __u8 s_encryption_level; /* versioning level for encryption */
1309 __u8 s_reserved_pad; /* Padding to next 32bits */
1310 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1311 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1312 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1313 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1314 snapshot's future use */
1315 __le32 s_snapshot_list; /* inode number of the head of the
1316 on-disk snapshot list */
1317#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1318 __le32 s_error_count; /* number of fs errors */
1319 __le32 s_first_error_time; /* first time an error happened */
1320 __le32 s_first_error_ino; /* inode involved in first error */
1321 __le64 s_first_error_block; /* block involved of first error */
1322 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1323 __le32 s_first_error_line; /* line number where error happened */
1324 __le32 s_last_error_time; /* most recent time of an error */
1325 __le32 s_last_error_ino; /* inode involved in last error */
1326 __le32 s_last_error_line; /* line number where error happened */
1327 __le64 s_last_error_block; /* block involved of last error */
1328 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1329#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1330 __u8 s_mount_opts[64];
1331 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1332 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1333 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1334 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1335 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1336 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1337 __le32 s_lpf_ino; /* Location of the lost+found inode */
1338 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1339 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1340 __u8 s_wtime_hi;
1341 __u8 s_mtime_hi;
1342 __u8 s_mkfs_time_hi;
1343 __u8 s_lastcheck_hi;
1344 __u8 s_first_error_time_hi;
1345 __u8 s_last_error_time_hi;
1346 __u8 s_pad[2];
1347 __le16 s_encoding; /* Filename charset encoding */
1348 __le16 s_encoding_flags; /* Filename charset encoding flags */
1349 __le32 s_reserved[95]; /* Padding to the end of the block */
1350 __le32 s_checksum; /* crc32c(superblock) */
1351};
1352
1353#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1354
1355#ifdef __KERNEL__
1356
1357/*
1358 * run-time mount flags
1359 */
1360#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1361#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1362#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1363
1364#ifdef CONFIG_FS_ENCRYPTION
1365#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1366 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1367#else
1368#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1369#endif
1370
1371/* Number of quota types we support */
1372#define EXT4_MAXQUOTAS 3
1373
1374#define EXT4_ENC_UTF8_12_1 1
1375
1376/*
1377 * Flags for ext4_sb_info.s_encoding_flags.
1378 */
1379#define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1380
1381#define ext4_has_strict_mode(sbi) \
1382 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1383
1384/*
1385 * fourth extended-fs super-block data in memory
1386 */
1387struct ext4_sb_info {
1388 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1389 unsigned long s_inodes_per_block;/* Number of inodes per block */
1390 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1391 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1392 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1393 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1394 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1395 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1396 ext4_group_t s_groups_count; /* Number of groups in the fs */
1397 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1398 unsigned long s_overhead; /* # of fs overhead clusters */
1399 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1400 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1401 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1402 struct buffer_head * s_sbh; /* Buffer containing the super block */
1403 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1404 struct buffer_head **s_group_desc;
1405 unsigned int s_mount_opt;
1406 unsigned int s_mount_opt2;
1407 unsigned int s_mount_flags;
1408 unsigned int s_def_mount_opt;
1409 ext4_fsblk_t s_sb_block;
1410 atomic64_t s_resv_clusters;
1411 kuid_t s_resuid;
1412 kgid_t s_resgid;
1413 unsigned short s_mount_state;
1414 unsigned short s_pad;
1415 int s_addr_per_block_bits;
1416 int s_desc_per_block_bits;
1417 int s_inode_size;
1418 int s_first_ino;
1419 unsigned int s_inode_readahead_blks;
1420 unsigned int s_inode_goal;
1421 u32 s_hash_seed[4];
1422 int s_def_hash_version;
1423 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1424 struct percpu_counter s_freeclusters_counter;
1425 struct percpu_counter s_freeinodes_counter;
1426 struct percpu_counter s_dirs_counter;
1427 struct percpu_counter s_dirtyclusters_counter;
1428 struct blockgroup_lock *s_blockgroup_lock;
1429 struct proc_dir_entry *s_proc;
1430 struct kobject s_kobj;
1431 struct completion s_kobj_unregister;
1432 struct super_block *s_sb;
1433#ifdef CONFIG_UNICODE
1434 struct unicode_map *s_encoding;
1435 __u16 s_encoding_flags;
1436#endif
1437
1438 /* Journaling */
1439 struct journal_s *s_journal;
1440 struct list_head s_orphan;
1441 struct mutex s_orphan_lock;
1442 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1443 unsigned long s_commit_interval;
1444 u32 s_max_batch_time;
1445 u32 s_min_batch_time;
1446 struct block_device *journal_bdev;
1447#ifdef CONFIG_QUOTA
1448 /* Names of quota files with journalled quota */
1449 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1450 int s_jquota_fmt; /* Format of quota to use */
1451#endif
1452 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1453 struct ext4_system_blocks __rcu *system_blks;
1454
1455#ifdef EXTENTS_STATS
1456 /* ext4 extents stats */
1457 unsigned long s_ext_min;
1458 unsigned long s_ext_max;
1459 unsigned long s_depth_max;
1460 spinlock_t s_ext_stats_lock;
1461 unsigned long s_ext_blocks;
1462 unsigned long s_ext_extents;
1463#endif
1464
1465 /* for buddy allocator */
1466 struct ext4_group_info ***s_group_info;
1467 struct inode *s_buddy_cache;
1468 spinlock_t s_md_lock;
1469 unsigned short *s_mb_offsets;
1470 unsigned int *s_mb_maxs;
1471 unsigned int s_group_info_size;
1472 unsigned int s_mb_free_pending;
1473 struct list_head s_freed_data_list; /* List of blocks to be freed
1474 after commit completed */
1475
1476 /* tunables */
1477 unsigned long s_stripe;
1478 unsigned int s_mb_stream_request;
1479 unsigned int s_mb_max_to_scan;
1480 unsigned int s_mb_min_to_scan;
1481 unsigned int s_mb_stats;
1482 unsigned int s_mb_order2_reqs;
1483 unsigned int s_mb_group_prealloc;
1484 unsigned int s_max_dir_size_kb;
1485 /* where last allocation was done - for stream allocation */
1486 unsigned long s_mb_last_group;
1487 unsigned long s_mb_last_start;
1488
1489 /* stats for buddy allocator */
1490 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1491 atomic_t s_bal_success; /* we found long enough chunks */
1492 atomic_t s_bal_allocated; /* in blocks */
1493 atomic_t s_bal_ex_scanned; /* total extents scanned */
1494 atomic_t s_bal_goals; /* goal hits */
1495 atomic_t s_bal_breaks; /* too long searches */
1496 atomic_t s_bal_2orders; /* 2^order hits */
1497 spinlock_t s_bal_lock;
1498 unsigned long s_mb_buddies_generated;
1499 unsigned long long s_mb_generation_time;
1500 atomic_t s_mb_lost_chunks;
1501 atomic_t s_mb_preallocated;
1502 atomic_t s_mb_discarded;
1503 atomic_t s_lock_busy;
1504
1505 /* locality groups */
1506 struct ext4_locality_group __percpu *s_locality_groups;
1507
1508 /* for write statistics */
1509 unsigned long s_sectors_written_start;
1510 u64 s_kbytes_written;
1511
1512 /* the size of zero-out chunk */
1513 unsigned int s_extent_max_zeroout_kb;
1514
1515 unsigned int s_log_groups_per_flex;
1516 struct flex_groups *s_flex_groups;
1517 ext4_group_t s_flex_groups_allocated;
1518
1519 /* workqueue for reserved extent conversions (buffered io) */
1520 struct workqueue_struct *rsv_conversion_wq;
1521
1522 /* timer for periodic error stats printing */
1523 struct timer_list s_err_report;
1524
1525 /* Lazy inode table initialization info */
1526 struct ext4_li_request *s_li_request;
1527 /* Wait multiplier for lazy initialization thread */
1528 unsigned int s_li_wait_mult;
1529
1530 /* Kernel thread for multiple mount protection */
1531 struct task_struct *s_mmp_tsk;
1532
1533 /* record the last minlen when FITRIM is called. */
1534 atomic_t s_last_trim_minblks;
1535
1536 /* Reference to checksum algorithm driver via cryptoapi */
1537 struct crypto_shash *s_chksum_driver;
1538
1539 /* Precomputed FS UUID checksum for seeding other checksums */
1540 __u32 s_csum_seed;
1541
1542 /* Reclaim extents from extent status tree */
1543 struct shrinker s_es_shrinker;
1544 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1545 long s_es_nr_inode;
1546 struct ext4_es_stats s_es_stats;
1547 struct mb_cache *s_ea_block_cache;
1548 struct mb_cache *s_ea_inode_cache;
1549 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1550
1551 /* Ratelimit ext4 messages. */
1552 struct ratelimit_state s_err_ratelimit_state;
1553 struct ratelimit_state s_warning_ratelimit_state;
1554 struct ratelimit_state s_msg_ratelimit_state;
1555
1556 /* Barrier between changing inodes' journal flags and writepages ops. */
1557 struct percpu_rw_semaphore s_journal_flag_rwsem;
1558 struct dax_device *s_daxdev;
1559};
1560
1561static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1562{
1563 return sb->s_fs_info;
1564}
1565static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1566{
1567 return container_of(inode, struct ext4_inode_info, vfs_inode);
1568}
1569
1570static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1571{
1572 return ino == EXT4_ROOT_INO ||
1573 (ino >= EXT4_FIRST_INO(sb) &&
1574 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1575}
1576
1577/*
1578 * Inode dynamic state flags
1579 */
1580enum {
1581 EXT4_STATE_JDATA, /* journaled data exists */
1582 EXT4_STATE_NEW, /* inode is newly created */
1583 EXT4_STATE_XATTR, /* has in-inode xattrs */
1584 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1585 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1586 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1587 EXT4_STATE_NEWENTRY, /* File just added to dir */
1588 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1589 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1590 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1591 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1592};
1593
1594#define EXT4_INODE_BIT_FNS(name, field, offset) \
1595static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1596{ \
1597 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1598} \
1599static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1600{ \
1601 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1602} \
1603static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1604{ \
1605 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1606}
1607
1608/* Add these declarations here only so that these functions can be
1609 * found by name. Otherwise, they are very hard to locate. */
1610static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1611static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1612static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1613EXT4_INODE_BIT_FNS(flag, flags, 0)
1614
1615/* Add these declarations here only so that these functions can be
1616 * found by name. Otherwise, they are very hard to locate. */
1617static inline int ext4_test_inode_state(struct inode *inode, int bit);
1618static inline void ext4_set_inode_state(struct inode *inode, int bit);
1619static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1620#if (BITS_PER_LONG < 64)
1621EXT4_INODE_BIT_FNS(state, state_flags, 0)
1622
1623static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1624{
1625 (ei)->i_state_flags = 0;
1626}
1627#else
1628EXT4_INODE_BIT_FNS(state, flags, 32)
1629
1630static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1631{
1632 /* We depend on the fact that callers will set i_flags */
1633}
1634#endif
1635#else
1636/* Assume that user mode programs are passing in an ext4fs superblock, not
1637 * a kernel struct super_block. This will allow us to call the feature-test
1638 * macros from user land. */
1639#define EXT4_SB(sb) (sb)
1640#endif
1641
1642static inline bool ext4_verity_in_progress(struct inode *inode)
1643{
1644 return IS_ENABLED(CONFIG_FS_VERITY) &&
1645 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1646}
1647
1648#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1649
1650/*
1651 * Codes for operating systems
1652 */
1653#define EXT4_OS_LINUX 0
1654#define EXT4_OS_HURD 1
1655#define EXT4_OS_MASIX 2
1656#define EXT4_OS_FREEBSD 3
1657#define EXT4_OS_LITES 4
1658
1659/*
1660 * Revision levels
1661 */
1662#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1663#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1664
1665#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1666#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1667
1668#define EXT4_GOOD_OLD_INODE_SIZE 128
1669
1670#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1671#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1672#define EXT4_TIMESTAMP_MIN S32_MIN
1673
1674/*
1675 * Feature set definitions
1676 */
1677
1678#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1679#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1680#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1681#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1682#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1683#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1684#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1685#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1686
1687#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1688#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1689#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1690#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1691#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1692#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1693#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1694#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1695#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1696/*
1697 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1698 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1699 * all other data structures' checksums. However, the METADATA_CSUM and
1700 * GDT_CSUM bits are mutually exclusive.
1701 */
1702#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1703#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1704#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1705#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1706
1707#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1708#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1709#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1710#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1711#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1712#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1713#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1714#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1715#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1716#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1717#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1718#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1719#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1720#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1721#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1722#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1723
1724extern void ext4_update_dynamic_rev(struct super_block *sb);
1725
1726#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1727static inline bool ext4_has_feature_##name(struct super_block *sb) \
1728{ \
1729 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1730 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1731} \
1732static inline void ext4_set_feature_##name(struct super_block *sb) \
1733{ \
1734 ext4_update_dynamic_rev(sb); \
1735 EXT4_SB(sb)->s_es->s_feature_compat |= \
1736 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1737} \
1738static inline void ext4_clear_feature_##name(struct super_block *sb) \
1739{ \
1740 EXT4_SB(sb)->s_es->s_feature_compat &= \
1741 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1742}
1743
1744#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1745static inline bool ext4_has_feature_##name(struct super_block *sb) \
1746{ \
1747 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1748 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1749} \
1750static inline void ext4_set_feature_##name(struct super_block *sb) \
1751{ \
1752 ext4_update_dynamic_rev(sb); \
1753 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1754 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1755} \
1756static inline void ext4_clear_feature_##name(struct super_block *sb) \
1757{ \
1758 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1759 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1760}
1761
1762#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1763static inline bool ext4_has_feature_##name(struct super_block *sb) \
1764{ \
1765 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1766 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1767} \
1768static inline void ext4_set_feature_##name(struct super_block *sb) \
1769{ \
1770 ext4_update_dynamic_rev(sb); \
1771 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1772 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1773} \
1774static inline void ext4_clear_feature_##name(struct super_block *sb) \
1775{ \
1776 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1777 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1778}
1779
1780EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1781EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1782EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1783EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1784EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1785EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1786EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1787EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
1788
1789EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1790EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1791EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1792EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1793EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1794EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1795EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1796EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1797EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1798EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1799EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1800EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1801EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
1802
1803EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1804EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1805EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1806EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1807EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1808EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1809EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1810EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1811EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1812EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1813EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1814EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1815EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1816EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1817EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1818EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
1819
1820#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1821#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1822 EXT4_FEATURE_INCOMPAT_META_BG)
1823#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1824 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1825 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1826
1827#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1828#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1829 EXT4_FEATURE_INCOMPAT_RECOVER| \
1830 EXT4_FEATURE_INCOMPAT_META_BG)
1831#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1832 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1833 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1834
1835#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1836#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1837 EXT4_FEATURE_INCOMPAT_RECOVER| \
1838 EXT4_FEATURE_INCOMPAT_META_BG| \
1839 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1840 EXT4_FEATURE_INCOMPAT_64BIT| \
1841 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1842 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1843 EXT4_FEATURE_INCOMPAT_MMP | \
1844 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1845 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1846 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1847 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1848 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1849#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1850 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1851 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1852 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1853 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1854 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1855 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1856 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1857 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1858 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1859 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1860 EXT4_FEATURE_RO_COMPAT_VERITY)
1861
1862#define EXTN_FEATURE_FUNCS(ver) \
1863static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1864{ \
1865 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1866 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1867} \
1868static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1869{ \
1870 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1871 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1872} \
1873static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1874{ \
1875 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1876 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1877}
1878
1879EXTN_FEATURE_FUNCS(2)
1880EXTN_FEATURE_FUNCS(3)
1881EXTN_FEATURE_FUNCS(4)
1882
1883static inline bool ext4_has_compat_features(struct super_block *sb)
1884{
1885 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1886}
1887static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1888{
1889 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1890}
1891static inline bool ext4_has_incompat_features(struct super_block *sb)
1892{
1893 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1894}
1895
1896/*
1897 * Superblock flags
1898 */
1899#define EXT4_FLAGS_RESIZING 0
1900#define EXT4_FLAGS_SHUTDOWN 1
1901
1902static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1903{
1904 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1905}
1906
1907
1908/*
1909 * Default values for user and/or group using reserved blocks
1910 */
1911#define EXT4_DEF_RESUID 0
1912#define EXT4_DEF_RESGID 0
1913
1914/*
1915 * Default project ID
1916 */
1917#define EXT4_DEF_PROJID 0
1918
1919#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1920
1921/*
1922 * Default mount options
1923 */
1924#define EXT4_DEFM_DEBUG 0x0001
1925#define EXT4_DEFM_BSDGROUPS 0x0002
1926#define EXT4_DEFM_XATTR_USER 0x0004
1927#define EXT4_DEFM_ACL 0x0008
1928#define EXT4_DEFM_UID16 0x0010
1929#define EXT4_DEFM_JMODE 0x0060
1930#define EXT4_DEFM_JMODE_DATA 0x0020
1931#define EXT4_DEFM_JMODE_ORDERED 0x0040
1932#define EXT4_DEFM_JMODE_WBACK 0x0060
1933#define EXT4_DEFM_NOBARRIER 0x0100
1934#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1935#define EXT4_DEFM_DISCARD 0x0400
1936#define EXT4_DEFM_NODELALLOC 0x0800
1937
1938/*
1939 * Default journal batch times
1940 */
1941#define EXT4_DEF_MIN_BATCH_TIME 0
1942#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1943
1944/*
1945 * Minimum number of groups in a flexgroup before we separate out
1946 * directories into the first block group of a flexgroup
1947 */
1948#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1949
1950/*
1951 * Structure of a directory entry
1952 */
1953#define EXT4_NAME_LEN 255
1954
1955struct ext4_dir_entry {
1956 __le32 inode; /* Inode number */
1957 __le16 rec_len; /* Directory entry length */
1958 __le16 name_len; /* Name length */
1959 char name[EXT4_NAME_LEN]; /* File name */
1960};
1961
1962/*
1963 * The new version of the directory entry. Since EXT4 structures are
1964 * stored in intel byte order, and the name_len field could never be
1965 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1966 * file_type field.
1967 */
1968struct ext4_dir_entry_2 {
1969 __le32 inode; /* Inode number */
1970 __le16 rec_len; /* Directory entry length */
1971 __u8 name_len; /* Name length */
1972 __u8 file_type;
1973 char name[EXT4_NAME_LEN]; /* File name */
1974};
1975
1976/*
1977 * This is a bogus directory entry at the end of each leaf block that
1978 * records checksums.
1979 */
1980struct ext4_dir_entry_tail {
1981 __le32 det_reserved_zero1; /* Pretend to be unused */
1982 __le16 det_rec_len; /* 12 */
1983 __u8 det_reserved_zero2; /* Zero name length */
1984 __u8 det_reserved_ft; /* 0xDE, fake file type */
1985 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1986};
1987
1988#define EXT4_DIRENT_TAIL(block, blocksize) \
1989 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1990 ((blocksize) - \
1991 sizeof(struct ext4_dir_entry_tail))))
1992
1993/*
1994 * Ext4 directory file types. Only the low 3 bits are used. The
1995 * other bits are reserved for now.
1996 */
1997#define EXT4_FT_UNKNOWN 0
1998#define EXT4_FT_REG_FILE 1
1999#define EXT4_FT_DIR 2
2000#define EXT4_FT_CHRDEV 3
2001#define EXT4_FT_BLKDEV 4
2002#define EXT4_FT_FIFO 5
2003#define EXT4_FT_SOCK 6
2004#define EXT4_FT_SYMLINK 7
2005
2006#define EXT4_FT_MAX 8
2007
2008#define EXT4_FT_DIR_CSUM 0xDE
2009
2010/*
2011 * EXT4_DIR_PAD defines the directory entries boundaries
2012 *
2013 * NOTE: It must be a multiple of 4
2014 */
2015#define EXT4_DIR_PAD 4
2016#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2017#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2018 ~EXT4_DIR_ROUND)
2019#define EXT4_MAX_REC_LEN ((1<<16)-1)
2020
2021/*
2022 * If we ever get support for fs block sizes > page_size, we'll need
2023 * to remove the #if statements in the next two functions...
2024 */
2025static inline unsigned int
2026ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2027{
2028 unsigned len = le16_to_cpu(dlen);
2029
2030#if (PAGE_SIZE >= 65536)
2031 if (len == EXT4_MAX_REC_LEN || len == 0)
2032 return blocksize;
2033 return (len & 65532) | ((len & 3) << 16);
2034#else
2035 return len;
2036#endif
2037}
2038
2039static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2040{
2041 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2042 BUG();
2043#if (PAGE_SIZE >= 65536)
2044 if (len < 65536)
2045 return cpu_to_le16(len);
2046 if (len == blocksize) {
2047 if (blocksize == 65536)
2048 return cpu_to_le16(EXT4_MAX_REC_LEN);
2049 else
2050 return cpu_to_le16(0);
2051 }
2052 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2053#else
2054 return cpu_to_le16(len);
2055#endif
2056}
2057
2058/*
2059 * Hash Tree Directory indexing
2060 * (c) Daniel Phillips, 2001
2061 */
2062
2063#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2064 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2065#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2066 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2067#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2068
2069/* Legal values for the dx_root hash_version field: */
2070
2071#define DX_HASH_LEGACY 0
2072#define DX_HASH_HALF_MD4 1
2073#define DX_HASH_TEA 2
2074#define DX_HASH_LEGACY_UNSIGNED 3
2075#define DX_HASH_HALF_MD4_UNSIGNED 4
2076#define DX_HASH_TEA_UNSIGNED 5
2077
2078static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2079 const void *address, unsigned int length)
2080{
2081 struct {
2082 struct shash_desc shash;
2083 char ctx[4];
2084 } desc;
2085
2086 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2087
2088 desc.shash.tfm = sbi->s_chksum_driver;
2089 *(u32 *)desc.ctx = crc;
2090
2091 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2092
2093 return *(u32 *)desc.ctx;
2094}
2095
2096#ifdef __KERNEL__
2097
2098/* hash info structure used by the directory hash */
2099struct dx_hash_info
2100{
2101 u32 hash;
2102 u32 minor_hash;
2103 int hash_version;
2104 u32 *seed;
2105};
2106
2107
2108/* 32 and 64 bit signed EOF for dx directories */
2109#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2110#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2111
2112
2113/*
2114 * Control parameters used by ext4_htree_next_block
2115 */
2116#define HASH_NB_ALWAYS 1
2117
2118struct ext4_filename {
2119 const struct qstr *usr_fname;
2120 struct fscrypt_str disk_name;
2121 struct dx_hash_info hinfo;
2122#ifdef CONFIG_FS_ENCRYPTION
2123 struct fscrypt_str crypto_buf;
2124#endif
2125#ifdef CONFIG_UNICODE
2126 struct fscrypt_str cf_name;
2127#endif
2128};
2129
2130#define fname_name(p) ((p)->disk_name.name)
2131#define fname_len(p) ((p)->disk_name.len)
2132
2133/*
2134 * Describe an inode's exact location on disk and in memory
2135 */
2136struct ext4_iloc
2137{
2138 struct buffer_head *bh;
2139 unsigned long offset;
2140 ext4_group_t block_group;
2141};
2142
2143static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2144{
2145 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2146}
2147
2148static inline bool ext4_is_quota_file(struct inode *inode)
2149{
2150 return IS_NOQUOTA(inode) &&
2151 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2152}
2153
2154/*
2155 * This structure is stuffed into the struct file's private_data field
2156 * for directories. It is where we put information so that we can do
2157 * readdir operations in hash tree order.
2158 */
2159struct dir_private_info {
2160 struct rb_root root;
2161 struct rb_node *curr_node;
2162 struct fname *extra_fname;
2163 loff_t last_pos;
2164 __u32 curr_hash;
2165 __u32 curr_minor_hash;
2166 __u32 next_hash;
2167};
2168
2169/* calculate the first block number of the group */
2170static inline ext4_fsblk_t
2171ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2172{
2173 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2174 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2175}
2176
2177/*
2178 * Special error return code only used by dx_probe() and its callers.
2179 */
2180#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2181
2182/* htree levels for ext4 */
2183#define EXT4_HTREE_LEVEL_COMPAT 2
2184#define EXT4_HTREE_LEVEL 3
2185
2186static inline int ext4_dir_htree_level(struct super_block *sb)
2187{
2188 return ext4_has_feature_largedir(sb) ?
2189 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2190}
2191
2192/*
2193 * Timeout and state flag for lazy initialization inode thread.
2194 */
2195#define EXT4_DEF_LI_WAIT_MULT 10
2196#define EXT4_DEF_LI_MAX_START_DELAY 5
2197#define EXT4_LAZYINIT_QUIT 0x0001
2198#define EXT4_LAZYINIT_RUNNING 0x0002
2199
2200/*
2201 * Lazy inode table initialization info
2202 */
2203struct ext4_lazy_init {
2204 unsigned long li_state;
2205 struct list_head li_request_list;
2206 struct mutex li_list_mtx;
2207};
2208
2209struct ext4_li_request {
2210 struct super_block *lr_super;
2211 struct ext4_sb_info *lr_sbi;
2212 ext4_group_t lr_next_group;
2213 struct list_head lr_request;
2214 unsigned long lr_next_sched;
2215 unsigned long lr_timeout;
2216};
2217
2218struct ext4_features {
2219 struct kobject f_kobj;
2220 struct completion f_kobj_unregister;
2221};
2222
2223/*
2224 * This structure will be used for multiple mount protection. It will be
2225 * written into the block number saved in the s_mmp_block field in the
2226 * superblock. Programs that check MMP should assume that if
2227 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2228 * to use the filesystem, regardless of how old the timestamp is.
2229 */
2230#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2231#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2232#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2233#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2234
2235struct mmp_struct {
2236 __le32 mmp_magic; /* Magic number for MMP */
2237 __le32 mmp_seq; /* Sequence no. updated periodically */
2238
2239 /*
2240 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2241 * purposes and do not affect the correctness of the algorithm
2242 */
2243 __le64 mmp_time; /* Time last updated */
2244 char mmp_nodename[64]; /* Node which last updated MMP block */
2245 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2246
2247 /*
2248 * mmp_check_interval is used to verify if the MMP block has been
2249 * updated on the block device. The value is updated based on the
2250 * maximum time to write the MMP block during an update cycle.
2251 */
2252 __le16 mmp_check_interval;
2253
2254 __le16 mmp_pad1;
2255 __le32 mmp_pad2[226];
2256 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2257};
2258
2259/* arguments passed to the mmp thread */
2260struct mmpd_data {
2261 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2262 struct super_block *sb; /* super block of the fs */
2263};
2264
2265/*
2266 * Check interval multiplier
2267 * The MMP block is written every update interval and initially checked every
2268 * update interval x the multiplier (the value is then adapted based on the
2269 * write latency). The reason is that writes can be delayed under load and we
2270 * don't want readers to incorrectly assume that the filesystem is no longer
2271 * in use.
2272 */
2273#define EXT4_MMP_CHECK_MULT 2UL
2274
2275/*
2276 * Minimum interval for MMP checking in seconds.
2277 */
2278#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2279
2280/*
2281 * Maximum interval for MMP checking in seconds.
2282 */
2283#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2284
2285/*
2286 * Function prototypes
2287 */
2288
2289/*
2290 * Ok, these declarations are also in <linux/kernel.h> but none of the
2291 * ext4 source programs needs to include it so they are duplicated here.
2292 */
2293# define NORET_TYPE /**/
2294# define ATTRIB_NORET __attribute__((noreturn))
2295# define NORET_AND noreturn,
2296
2297/* bitmap.c */
2298extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2299void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2300 struct ext4_group_desc *gdp,
2301 struct buffer_head *bh, int sz);
2302int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2303 struct ext4_group_desc *gdp,
2304 struct buffer_head *bh, int sz);
2305void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2306 struct ext4_group_desc *gdp,
2307 struct buffer_head *bh);
2308int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2309 struct ext4_group_desc *gdp,
2310 struct buffer_head *bh);
2311
2312/* balloc.c */
2313extern void ext4_get_group_no_and_offset(struct super_block *sb,
2314 ext4_fsblk_t blocknr,
2315 ext4_group_t *blockgrpp,
2316 ext4_grpblk_t *offsetp);
2317extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2318 ext4_fsblk_t block);
2319
2320extern unsigned int ext4_block_group(struct super_block *sb,
2321 ext4_fsblk_t blocknr);
2322extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2323 ext4_fsblk_t blocknr);
2324extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2325extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2326 ext4_group_t group);
2327extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2328 ext4_fsblk_t goal,
2329 unsigned int flags,
2330 unsigned long *count,
2331 int *errp);
2332extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2333 s64 nclusters, unsigned int flags);
2334extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2335extern void ext4_check_blocks_bitmap(struct super_block *);
2336extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2337 ext4_group_t block_group,
2338 struct buffer_head ** bh);
2339extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2340
2341extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2342 ext4_group_t block_group);
2343extern int ext4_wait_block_bitmap(struct super_block *sb,
2344 ext4_group_t block_group,
2345 struct buffer_head *bh);
2346extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2347 ext4_group_t block_group);
2348extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2349 ext4_group_t block_group,
2350 struct ext4_group_desc *gdp);
2351ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2352
2353#ifdef CONFIG_UNICODE
2354extern void ext4_fname_setup_ci_filename(struct inode *dir,
2355 const struct qstr *iname,
2356 struct fscrypt_str *fname);
2357#endif
2358
2359#ifdef CONFIG_FS_ENCRYPTION
2360static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2361 const struct fscrypt_name *src)
2362{
2363 memset(dst, 0, sizeof(*dst));
2364
2365 dst->usr_fname = src->usr_fname;
2366 dst->disk_name = src->disk_name;
2367 dst->hinfo.hash = src->hash;
2368 dst->hinfo.minor_hash = src->minor_hash;
2369 dst->crypto_buf = src->crypto_buf;
2370}
2371
2372static inline int ext4_fname_setup_filename(struct inode *dir,
2373 const struct qstr *iname,
2374 int lookup,
2375 struct ext4_filename *fname)
2376{
2377 struct fscrypt_name name;
2378 int err;
2379
2380 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2381 if (err)
2382 return err;
2383
2384 ext4_fname_from_fscrypt_name(fname, &name);
2385
2386#ifdef CONFIG_UNICODE
2387 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2388#endif
2389 return 0;
2390}
2391
2392static inline int ext4_fname_prepare_lookup(struct inode *dir,
2393 struct dentry *dentry,
2394 struct ext4_filename *fname)
2395{
2396 struct fscrypt_name name;
2397 int err;
2398
2399 err = fscrypt_prepare_lookup(dir, dentry, &name);
2400 if (err)
2401 return err;
2402
2403 ext4_fname_from_fscrypt_name(fname, &name);
2404
2405#ifdef CONFIG_UNICODE
2406 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2407#endif
2408 return 0;
2409}
2410
2411static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2412{
2413 struct fscrypt_name name;
2414
2415 name.crypto_buf = fname->crypto_buf;
2416 fscrypt_free_filename(&name);
2417
2418 fname->crypto_buf.name = NULL;
2419 fname->usr_fname = NULL;
2420 fname->disk_name.name = NULL;
2421
2422#ifdef CONFIG_UNICODE
2423 kfree(fname->cf_name.name);
2424 fname->cf_name.name = NULL;
2425#endif
2426}
2427#else /* !CONFIG_FS_ENCRYPTION */
2428static inline int ext4_fname_setup_filename(struct inode *dir,
2429 const struct qstr *iname,
2430 int lookup,
2431 struct ext4_filename *fname)
2432{
2433 fname->usr_fname = iname;
2434 fname->disk_name.name = (unsigned char *) iname->name;
2435 fname->disk_name.len = iname->len;
2436
2437#ifdef CONFIG_UNICODE
2438 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2439#endif
2440
2441 return 0;
2442}
2443
2444static inline int ext4_fname_prepare_lookup(struct inode *dir,
2445 struct dentry *dentry,
2446 struct ext4_filename *fname)
2447{
2448 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2449}
2450
2451static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2452{
2453#ifdef CONFIG_UNICODE
2454 kfree(fname->cf_name.name);
2455 fname->cf_name.name = NULL;
2456#endif
2457}
2458#endif /* !CONFIG_FS_ENCRYPTION */
2459
2460/* dir.c */
2461extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2462 struct file *,
2463 struct ext4_dir_entry_2 *,
2464 struct buffer_head *, char *, int,
2465 unsigned int);
2466#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2467 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2468 (de), (bh), (buf), (size), (offset)))
2469extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2470 __u32 minor_hash,
2471 struct ext4_dir_entry_2 *dirent,
2472 struct fscrypt_str *ent_name);
2473extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2474extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2475 struct buffer_head *bh,
2476 void *buf, int buf_size,
2477 struct ext4_filename *fname,
2478 struct ext4_dir_entry_2 **dest_de);
2479void ext4_insert_dentry(struct inode *inode,
2480 struct ext4_dir_entry_2 *de,
2481 int buf_size,
2482 struct ext4_filename *fname);
2483static inline void ext4_update_dx_flag(struct inode *inode)
2484{
2485 if (!ext4_has_feature_dir_index(inode->i_sb))
2486 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2487}
2488static const unsigned char ext4_filetype_table[] = {
2489 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2490};
2491
2492static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2493{
2494 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2495 return DT_UNKNOWN;
2496
2497 return ext4_filetype_table[filetype];
2498}
2499extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2500 void *buf, int buf_size);
2501
2502/* fsync.c */
2503extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2504
2505/* hash.c */
2506extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2507 struct dx_hash_info *hinfo);
2508
2509/* ialloc.c */
2510extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2511 const struct qstr *qstr, __u32 goal,
2512 uid_t *owner, __u32 i_flags,
2513 int handle_type, unsigned int line_no,
2514 int nblocks);
2515
2516#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2517 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2518 i_flags, 0, 0, 0)
2519#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2520 type, nblocks) \
2521 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2522 0, (type), __LINE__, (nblocks))
2523
2524
2525extern void ext4_free_inode(handle_t *, struct inode *);
2526extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2527extern unsigned long ext4_count_free_inodes(struct super_block *);
2528extern unsigned long ext4_count_dirs(struct super_block *);
2529extern void ext4_check_inodes_bitmap(struct super_block *);
2530extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2531extern int ext4_init_inode_table(struct super_block *sb,
2532 ext4_group_t group, int barrier);
2533extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2534
2535/* mballoc.c */
2536extern const struct seq_operations ext4_mb_seq_groups_ops;
2537extern long ext4_mb_stats;
2538extern long ext4_mb_max_to_scan;
2539extern int ext4_mb_init(struct super_block *);
2540extern int ext4_mb_release(struct super_block *);
2541extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2542 struct ext4_allocation_request *, int *);
2543extern int ext4_mb_reserve_blocks(struct super_block *, int);
2544extern void ext4_discard_preallocations(struct inode *);
2545extern int __init ext4_init_mballoc(void);
2546extern void ext4_exit_mballoc(void);
2547extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2548 struct buffer_head *bh, ext4_fsblk_t block,
2549 unsigned long count, int flags);
2550extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2551 ext4_group_t ngroups);
2552extern int ext4_mb_add_groupinfo(struct super_block *sb,
2553 ext4_group_t i, struct ext4_group_desc *desc);
2554extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2555 ext4_fsblk_t block, unsigned long count);
2556extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2557extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2558
2559/* inode.c */
2560int ext4_inode_is_fast_symlink(struct inode *inode);
2561struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2562struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2563int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2564 bool wait, struct buffer_head **bhs);
2565int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2566 struct buffer_head *bh_result, int create);
2567int ext4_get_block(struct inode *inode, sector_t iblock,
2568 struct buffer_head *bh_result, int create);
2569int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2570 struct buffer_head *bh, int create);
2571int ext4_walk_page_buffers(handle_t *handle,
2572 struct buffer_head *head,
2573 unsigned from,
2574 unsigned to,
2575 int *partial,
2576 int (*fn)(handle_t *handle,
2577 struct buffer_head *bh));
2578int do_journal_get_write_access(handle_t *handle,
2579 struct buffer_head *bh);
2580#define FALL_BACK_TO_NONDELALLOC 1
2581#define CONVERT_INLINE_DATA 2
2582
2583typedef enum {
2584 EXT4_IGET_NORMAL = 0,
2585 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2586 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2587} ext4_iget_flags;
2588
2589extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2590 ext4_iget_flags flags, const char *function,
2591 unsigned int line);
2592
2593#define ext4_iget(sb, ino, flags) \
2594 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2595
2596extern int ext4_write_inode(struct inode *, struct writeback_control *);
2597extern int ext4_setattr(struct dentry *, struct iattr *);
2598extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2599extern void ext4_evict_inode(struct inode *);
2600extern void ext4_clear_inode(struct inode *);
2601extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2602extern int ext4_sync_inode(handle_t *, struct inode *);
2603extern void ext4_dirty_inode(struct inode *, int);
2604extern int ext4_change_inode_journal_flag(struct inode *, int);
2605extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2606extern int ext4_inode_attach_jinode(struct inode *inode);
2607extern int ext4_can_truncate(struct inode *inode);
2608extern int ext4_truncate(struct inode *);
2609extern int ext4_break_layouts(struct inode *);
2610extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2611extern void ext4_set_inode_flags(struct inode *);
2612extern int ext4_alloc_da_blocks(struct inode *inode);
2613extern void ext4_set_aops(struct inode *inode);
2614extern int ext4_writepage_trans_blocks(struct inode *);
2615extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2616extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2617 loff_t lstart, loff_t lend);
2618extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2619extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2620extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2621extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2622extern void ext4_da_release_space(struct inode *inode, int to_free);
2623extern void ext4_da_update_reserve_space(struct inode *inode,
2624 int used, int quota_claim);
2625extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2626 ext4_fsblk_t pblk, ext4_lblk_t len);
2627
2628/* indirect.c */
2629extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2630 struct ext4_map_blocks *map, int flags);
2631extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2632extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2633extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2634extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2635 ext4_lblk_t start, ext4_lblk_t end);
2636
2637/* ioctl.c */
2638extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2639extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2640
2641/* migrate.c */
2642extern int ext4_ext_migrate(struct inode *);
2643extern int ext4_ind_migrate(struct inode *inode);
2644
2645/* namei.c */
2646extern int ext4_dirblock_csum_verify(struct inode *inode,
2647 struct buffer_head *bh);
2648extern int ext4_orphan_add(handle_t *, struct inode *);
2649extern int ext4_orphan_del(handle_t *, struct inode *);
2650extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2651 __u32 start_minor_hash, __u32 *next_hash);
2652extern int ext4_search_dir(struct buffer_head *bh,
2653 char *search_buf,
2654 int buf_size,
2655 struct inode *dir,
2656 struct ext4_filename *fname,
2657 unsigned int offset,
2658 struct ext4_dir_entry_2 **res_dir);
2659extern int ext4_generic_delete_entry(handle_t *handle,
2660 struct inode *dir,
2661 struct ext4_dir_entry_2 *de_del,
2662 struct buffer_head *bh,
2663 void *entry_buf,
2664 int buf_size,
2665 int csum_size);
2666extern bool ext4_empty_dir(struct inode *inode);
2667
2668/* resize.c */
2669extern int ext4_group_add(struct super_block *sb,
2670 struct ext4_new_group_data *input);
2671extern int ext4_group_extend(struct super_block *sb,
2672 struct ext4_super_block *es,
2673 ext4_fsblk_t n_blocks_count);
2674extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2675
2676/* super.c */
2677extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2678 sector_t block, int op_flags);
2679extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2680extern int ext4_calculate_overhead(struct super_block *sb);
2681extern void ext4_superblock_csum_set(struct super_block *sb);
2682extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2683extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2684extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2685 ext4_group_t ngroup);
2686extern const char *ext4_decode_error(struct super_block *sb, int errno,
2687 char nbuf[16]);
2688extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2689 ext4_group_t block_group,
2690 unsigned int flags);
2691
2692extern __printf(4, 5)
2693void __ext4_error(struct super_block *, const char *, unsigned int,
2694 const char *, ...);
2695extern __printf(5, 6)
2696void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2697 const char *, ...);
2698extern __printf(5, 6)
2699void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2700 const char *, ...);
2701extern void __ext4_std_error(struct super_block *, const char *,
2702 unsigned int, int);
2703extern __printf(4, 5)
2704void __ext4_abort(struct super_block *, const char *, unsigned int,
2705 const char *, ...);
2706extern __printf(4, 5)
2707void __ext4_warning(struct super_block *, const char *, unsigned int,
2708 const char *, ...);
2709extern __printf(4, 5)
2710void __ext4_warning_inode(const struct inode *inode, const char *function,
2711 unsigned int line, const char *fmt, ...);
2712extern __printf(3, 4)
2713void __ext4_msg(struct super_block *, const char *, const char *, ...);
2714extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2715 const char *, unsigned int, const char *);
2716extern __printf(7, 8)
2717void __ext4_grp_locked_error(const char *, unsigned int,
2718 struct super_block *, ext4_group_t,
2719 unsigned long, ext4_fsblk_t,
2720 const char *, ...);
2721
2722#define EXT4_ERROR_INODE(inode, fmt, a...) \
2723 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2724
2725#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2726 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2727
2728#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2729 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2730
2731#ifdef CONFIG_PRINTK
2732
2733#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2734 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2735#define ext4_error_file(file, func, line, block, fmt, ...) \
2736 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2737#define ext4_error(sb, fmt, ...) \
2738 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2739#define ext4_abort(sb, fmt, ...) \
2740 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2741#define ext4_warning(sb, fmt, ...) \
2742 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2743#define ext4_warning_inode(inode, fmt, ...) \
2744 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2745#define ext4_msg(sb, level, fmt, ...) \
2746 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2747#define dump_mmp_msg(sb, mmp, msg) \
2748 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2749#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2750 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2751 fmt, ##__VA_ARGS__)
2752
2753#else
2754
2755#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2756do { \
2757 no_printk(fmt, ##__VA_ARGS__); \
2758 __ext4_error_inode(inode, "", 0, block, " "); \
2759} while (0)
2760#define ext4_error_file(file, func, line, block, fmt, ...) \
2761do { \
2762 no_printk(fmt, ##__VA_ARGS__); \
2763 __ext4_error_file(file, "", 0, block, " "); \
2764} while (0)
2765#define ext4_error(sb, fmt, ...) \
2766do { \
2767 no_printk(fmt, ##__VA_ARGS__); \
2768 __ext4_error(sb, "", 0, " "); \
2769} while (0)
2770#define ext4_abort(sb, fmt, ...) \
2771do { \
2772 no_printk(fmt, ##__VA_ARGS__); \
2773 __ext4_abort(sb, "", 0, " "); \
2774} while (0)
2775#define ext4_warning(sb, fmt, ...) \
2776do { \
2777 no_printk(fmt, ##__VA_ARGS__); \
2778 __ext4_warning(sb, "", 0, " "); \
2779} while (0)
2780#define ext4_warning_inode(inode, fmt, ...) \
2781do { \
2782 no_printk(fmt, ##__VA_ARGS__); \
2783 __ext4_warning_inode(inode, "", 0, " "); \
2784} while (0)
2785#define ext4_msg(sb, level, fmt, ...) \
2786do { \
2787 no_printk(fmt, ##__VA_ARGS__); \
2788 __ext4_msg(sb, "", " "); \
2789} while (0)
2790#define dump_mmp_msg(sb, mmp, msg) \
2791 __dump_mmp_msg(sb, mmp, "", 0, "")
2792#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2793do { \
2794 no_printk(fmt, ##__VA_ARGS__); \
2795 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2796} while (0)
2797
2798#endif
2799
2800extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2801 __u32 compat);
2802extern int ext4_update_rocompat_feature(handle_t *handle,
2803 struct super_block *sb, __u32 rocompat);
2804extern int ext4_update_incompat_feature(handle_t *handle,
2805 struct super_block *sb, __u32 incompat);
2806extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2807 struct ext4_group_desc *bg);
2808extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2809 struct ext4_group_desc *bg);
2810extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2811 struct ext4_group_desc *bg);
2812extern __u32 ext4_free_group_clusters(struct super_block *sb,
2813 struct ext4_group_desc *bg);
2814extern __u32 ext4_free_inodes_count(struct super_block *sb,
2815 struct ext4_group_desc *bg);
2816extern __u32 ext4_used_dirs_count(struct super_block *sb,
2817 struct ext4_group_desc *bg);
2818extern __u32 ext4_itable_unused_count(struct super_block *sb,
2819 struct ext4_group_desc *bg);
2820extern void ext4_block_bitmap_set(struct super_block *sb,
2821 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2822extern void ext4_inode_bitmap_set(struct super_block *sb,
2823 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2824extern void ext4_inode_table_set(struct super_block *sb,
2825 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2826extern void ext4_free_group_clusters_set(struct super_block *sb,
2827 struct ext4_group_desc *bg,
2828 __u32 count);
2829extern void ext4_free_inodes_set(struct super_block *sb,
2830 struct ext4_group_desc *bg, __u32 count);
2831extern void ext4_used_dirs_set(struct super_block *sb,
2832 struct ext4_group_desc *bg, __u32 count);
2833extern void ext4_itable_unused_set(struct super_block *sb,
2834 struct ext4_group_desc *bg, __u32 count);
2835extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2836 struct ext4_group_desc *gdp);
2837extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2838 struct ext4_group_desc *gdp);
2839extern int ext4_register_li_request(struct super_block *sb,
2840 ext4_group_t first_not_zeroed);
2841
2842static inline int ext4_has_metadata_csum(struct super_block *sb)
2843{
2844 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2845 !EXT4_SB(sb)->s_chksum_driver);
2846
2847 return ext4_has_feature_metadata_csum(sb) &&
2848 (EXT4_SB(sb)->s_chksum_driver != NULL);
2849}
2850
2851static inline int ext4_has_group_desc_csum(struct super_block *sb)
2852{
2853 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2854}
2855
2856static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2857{
2858 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2859 le32_to_cpu(es->s_blocks_count_lo);
2860}
2861
2862static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2863{
2864 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2865 le32_to_cpu(es->s_r_blocks_count_lo);
2866}
2867
2868static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2869{
2870 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2871 le32_to_cpu(es->s_free_blocks_count_lo);
2872}
2873
2874static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2875 ext4_fsblk_t blk)
2876{
2877 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2878 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2879}
2880
2881static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2882 ext4_fsblk_t blk)
2883{
2884 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2885 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2886}
2887
2888static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2889 ext4_fsblk_t blk)
2890{
2891 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2892 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2893}
2894
2895static inline loff_t ext4_isize(struct super_block *sb,
2896 struct ext4_inode *raw_inode)
2897{
2898 if (ext4_has_feature_largedir(sb) ||
2899 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2900 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2901 le32_to_cpu(raw_inode->i_size_lo);
2902
2903 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2904}
2905
2906static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2907{
2908 raw_inode->i_size_lo = cpu_to_le32(i_size);
2909 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2910}
2911
2912static inline
2913struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2914 ext4_group_t group)
2915{
2916 struct ext4_group_info ***grp_info;
2917 long indexv, indexh;
2918 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2919 grp_info = EXT4_SB(sb)->s_group_info;
2920 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2921 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2922 return grp_info[indexv][indexh];
2923}
2924
2925/*
2926 * Reading s_groups_count requires using smp_rmb() afterwards. See
2927 * the locking protocol documented in the comments of ext4_group_add()
2928 * in resize.c
2929 */
2930static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2931{
2932 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2933
2934 smp_rmb();
2935 return ngroups;
2936}
2937
2938static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2939 ext4_group_t block_group)
2940{
2941 return block_group >> sbi->s_log_groups_per_flex;
2942}
2943
2944static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2945{
2946 return 1 << sbi->s_log_groups_per_flex;
2947}
2948
2949#define ext4_std_error(sb, errno) \
2950do { \
2951 if ((errno)) \
2952 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2953} while (0)
2954
2955#ifdef CONFIG_SMP
2956/* Each CPU can accumulate percpu_counter_batch clusters in their local
2957 * counters. So we need to make sure we have free clusters more
2958 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2959 */
2960#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2961#else
2962#define EXT4_FREECLUSTERS_WATERMARK 0
2963#endif
2964
2965/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2966static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2967{
2968 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2969 !inode_is_locked(inode));
2970 down_write(&EXT4_I(inode)->i_data_sem);
2971 if (newsize > EXT4_I(inode)->i_disksize)
2972 EXT4_I(inode)->i_disksize = newsize;
2973 up_write(&EXT4_I(inode)->i_data_sem);
2974}
2975
2976/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2977static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2978{
2979 int changed = 0;
2980
2981 if (newsize > inode->i_size) {
2982 i_size_write(inode, newsize);
2983 changed = 1;
2984 }
2985 if (newsize > EXT4_I(inode)->i_disksize) {
2986 ext4_update_i_disksize(inode, newsize);
2987 changed |= 2;
2988 }
2989 return changed;
2990}
2991
2992int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2993 loff_t len);
2994
2995struct ext4_group_info {
2996 unsigned long bb_state;
2997 struct rb_root bb_free_root;
2998 ext4_grpblk_t bb_first_free; /* first free block */
2999 ext4_grpblk_t bb_free; /* total free blocks */
3000 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3001 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3002 struct list_head bb_prealloc_list;
3003#ifdef DOUBLE_CHECK
3004 void *bb_bitmap;
3005#endif
3006 struct rw_semaphore alloc_sem;
3007 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3008 * regions, index is order.
3009 * bb_counters[3] = 5 means
3010 * 5 free 8-block regions. */
3011};
3012
3013#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3014#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3015#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3016#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3017#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3018 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3019#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3020 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3021
3022#define EXT4_MB_GRP_NEED_INIT(grp) \
3023 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3024#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3025 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3026#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3027 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3028
3029#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3030 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3031#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3032 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3033#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3034 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3035
3036#define EXT4_MAX_CONTENTION 8
3037#define EXT4_CONTENTION_THRESHOLD 2
3038
3039static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3040 ext4_group_t group)
3041{
3042 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3043}
3044
3045/*
3046 * Returns true if the filesystem is busy enough that attempts to
3047 * access the block group locks has run into contention.
3048 */
3049static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3050{
3051 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3052}
3053
3054static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3055{
3056 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3057 if (spin_trylock(lock))
3058 /*
3059 * We're able to grab the lock right away, so drop the
3060 * lock contention counter.
3061 */
3062 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3063 else {
3064 /*
3065 * The lock is busy, so bump the contention counter,
3066 * and then wait on the spin lock.
3067 */
3068 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3069 EXT4_MAX_CONTENTION);
3070 spin_lock(lock);
3071 }
3072}
3073
3074static inline void ext4_unlock_group(struct super_block *sb,
3075 ext4_group_t group)
3076{
3077 spin_unlock(ext4_group_lock_ptr(sb, group));
3078}
3079
3080/*
3081 * Block validity checking
3082 */
3083#define ext4_check_indirect_blockref(inode, bh) \
3084 ext4_check_blockref(__func__, __LINE__, inode, \
3085 (__le32 *)(bh)->b_data, \
3086 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3087
3088#define ext4_ind_check_inode(inode) \
3089 ext4_check_blockref(__func__, __LINE__, inode, \
3090 EXT4_I(inode)->i_data, \
3091 EXT4_NDIR_BLOCKS)
3092
3093/*
3094 * Inodes and files operations
3095 */
3096
3097/* dir.c */
3098extern const struct file_operations ext4_dir_operations;
3099
3100#ifdef CONFIG_UNICODE
3101extern const struct dentry_operations ext4_dentry_ops;
3102#endif
3103
3104/* file.c */
3105extern const struct inode_operations ext4_file_inode_operations;
3106extern const struct file_operations ext4_file_operations;
3107extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3108
3109/* inline.c */
3110extern int ext4_get_max_inline_size(struct inode *inode);
3111extern int ext4_find_inline_data_nolock(struct inode *inode);
3112extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3113 unsigned int len);
3114extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3115
3116extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3117extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3118 struct inode *inode,
3119 loff_t pos, unsigned len,
3120 unsigned flags,
3121 struct page **pagep);
3122extern int ext4_write_inline_data_end(struct inode *inode,
3123 loff_t pos, unsigned len,
3124 unsigned copied,
3125 struct page *page);
3126extern struct buffer_head *
3127ext4_journalled_write_inline_data(struct inode *inode,
3128 unsigned len,
3129 struct page *page);
3130extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3131 struct inode *inode,
3132 loff_t pos, unsigned len,
3133 unsigned flags,
3134 struct page **pagep,
3135 void **fsdata);
3136extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3137 unsigned len, unsigned copied,
3138 struct page *page);
3139extern int ext4_try_add_inline_entry(handle_t *handle,
3140 struct ext4_filename *fname,
3141 struct inode *dir, struct inode *inode);
3142extern int ext4_try_create_inline_dir(handle_t *handle,
3143 struct inode *parent,
3144 struct inode *inode);
3145extern int ext4_read_inline_dir(struct file *filp,
3146 struct dir_context *ctx,
3147 int *has_inline_data);
3148extern int ext4_inlinedir_to_tree(struct file *dir_file,
3149 struct inode *dir, ext4_lblk_t block,
3150 struct dx_hash_info *hinfo,
3151 __u32 start_hash, __u32 start_minor_hash,
3152 int *has_inline_data);
3153extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3154 struct ext4_filename *fname,
3155 struct ext4_dir_entry_2 **res_dir,
3156 int *has_inline_data);
3157extern int ext4_delete_inline_entry(handle_t *handle,
3158 struct inode *dir,
3159 struct ext4_dir_entry_2 *de_del,
3160 struct buffer_head *bh,
3161 int *has_inline_data);
3162extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3163extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3164 struct ext4_dir_entry_2 **parent_de,
3165 int *retval);
3166extern int ext4_inline_data_fiemap(struct inode *inode,
3167 struct fiemap_extent_info *fieinfo,
3168 int *has_inline, __u64 start, __u64 len);
3169
3170struct iomap;
3171extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3172
3173extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3174
3175extern int ext4_convert_inline_data(struct inode *inode);
3176
3177static inline int ext4_has_inline_data(struct inode *inode)
3178{
3179 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3180 EXT4_I(inode)->i_inline_off;
3181}
3182
3183/* namei.c */
3184extern const struct inode_operations ext4_dir_inode_operations;
3185extern const struct inode_operations ext4_special_inode_operations;
3186extern struct dentry *ext4_get_parent(struct dentry *child);
3187extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3188 struct ext4_dir_entry_2 *de,
3189 int blocksize, int csum_size,
3190 unsigned int parent_ino, int dotdot_real_len);
3191extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3192 unsigned int blocksize);
3193extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3194 struct buffer_head *bh);
3195extern int ext4_ci_compare(const struct inode *parent,
3196 const struct qstr *fname,
3197 const struct qstr *entry, bool quick);
3198
3199#define S_SHIFT 12
3200static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3201 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3202 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3203 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3204 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3205 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3206 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3207 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3208};
3209
3210static inline void ext4_set_de_type(struct super_block *sb,
3211 struct ext4_dir_entry_2 *de,
3212 umode_t mode) {
3213 if (ext4_has_feature_filetype(sb))
3214 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3215}
3216
3217/* readpages.c */
3218extern int ext4_mpage_readpages(struct address_space *mapping,
3219 struct list_head *pages, struct page *page,
3220 unsigned nr_pages, bool is_readahead);
3221extern int __init ext4_init_post_read_processing(void);
3222extern void ext4_exit_post_read_processing(void);
3223
3224/* symlink.c */
3225extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3226extern const struct inode_operations ext4_symlink_inode_operations;
3227extern const struct inode_operations ext4_fast_symlink_inode_operations;
3228
3229/* sysfs.c */
3230extern int ext4_register_sysfs(struct super_block *sb);
3231extern void ext4_unregister_sysfs(struct super_block *sb);
3232extern int __init ext4_init_sysfs(void);
3233extern void ext4_exit_sysfs(void);
3234
3235/* block_validity */
3236extern void ext4_release_system_zone(struct super_block *sb);
3237extern int ext4_setup_system_zone(struct super_block *sb);
3238extern int __init ext4_init_system_zone(void);
3239extern void ext4_exit_system_zone(void);
3240extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3241 ext4_fsblk_t start_blk,
3242 unsigned int count);
3243extern int ext4_check_blockref(const char *, unsigned int,
3244 struct inode *, __le32 *, unsigned int);
3245
3246/* extents.c */
3247struct ext4_ext_path;
3248struct ext4_extent;
3249
3250/*
3251 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3252 * __le32.
3253 */
3254#define EXT_MAX_BLOCKS 0xffffffff
3255
3256extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3257extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3258extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3259extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3260 struct ext4_map_blocks *map, int flags);
3261extern int ext4_ext_truncate(handle_t *, struct inode *);
3262extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3263 ext4_lblk_t end);
3264extern void ext4_ext_init(struct super_block *);
3265extern void ext4_ext_release(struct super_block *);
3266extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3267 loff_t len);
3268extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3269 loff_t offset, ssize_t len);
3270extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3271 ext4_io_end_t *io_end);
3272extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3273 struct ext4_map_blocks *map, int flags);
3274extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3275 ext4_lblk_t lblocks);
3276extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3277 int num,
3278 struct ext4_ext_path *path);
3279extern int ext4_can_extents_be_merged(struct inode *inode,
3280 struct ext4_extent *ex1,
3281 struct ext4_extent *ex2);
3282extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3283 struct ext4_ext_path **,
3284 struct ext4_extent *, int);
3285extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3286 struct ext4_ext_path **,
3287 int flags);
3288extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3289extern int ext4_ext_check_inode(struct inode *inode);
3290extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3291extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3292 __u64 start, __u64 len);
3293extern int ext4_get_es_cache(struct inode *inode,
3294 struct fiemap_extent_info *fieinfo,
3295 __u64 start, __u64 len);
3296extern int ext4_ext_precache(struct inode *inode);
3297extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3298extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3299extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3300 struct inode *inode2, ext4_lblk_t lblk1,
3301 ext4_lblk_t lblk2, ext4_lblk_t count,
3302 int mark_unwritten,int *err);
3303extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3304extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3305 int check_cred, int restart_cred,
3306 int revoke_cred);
3307
3308
3309/* move_extent.c */
3310extern void ext4_double_down_write_data_sem(struct inode *first,
3311 struct inode *second);
3312extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3313 struct inode *donor_inode);
3314extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3315 __u64 start_orig, __u64 start_donor,
3316 __u64 len, __u64 *moved_len);
3317
3318/* page-io.c */
3319extern int __init ext4_init_pageio(void);
3320extern void ext4_exit_pageio(void);
3321extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3322extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3323extern int ext4_put_io_end(ext4_io_end_t *io_end);
3324extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3325extern void ext4_io_submit_init(struct ext4_io_submit *io,
3326 struct writeback_control *wbc);
3327extern void ext4_end_io_rsv_work(struct work_struct *work);
3328extern void ext4_io_submit(struct ext4_io_submit *io);
3329extern int ext4_bio_write_page(struct ext4_io_submit *io,
3330 struct page *page,
3331 int len,
3332 struct writeback_control *wbc,
3333 bool keep_towrite);
3334extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3335extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3336
3337/* mmp.c */
3338extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3339
3340/* verity.c */
3341extern const struct fsverity_operations ext4_verityops;
3342
3343/*
3344 * Add new method to test whether block and inode bitmaps are properly
3345 * initialized. With uninit_bg reading the block from disk is not enough
3346 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3347 */
3348#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3349
3350static inline int bitmap_uptodate(struct buffer_head *bh)
3351{
3352 return (buffer_uptodate(bh) &&
3353 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3354}
3355static inline void set_bitmap_uptodate(struct buffer_head *bh)
3356{
3357 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3358}
3359
3360#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3361
3362/* For ioend & aio unwritten conversion wait queues */
3363#define EXT4_WQ_HASH_SZ 37
3364#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3365 EXT4_WQ_HASH_SZ])
3366extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3367
3368extern int ext4_resize_begin(struct super_block *sb);
3369extern void ext4_resize_end(struct super_block *sb);
3370
3371static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3372 struct ext4_io_end *io_end)
3373{
3374 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3375 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3376 atomic_inc(&EXT4_I(inode)->i_unwritten);
3377 }
3378}
3379
3380static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3381{
3382 struct inode *inode = io_end->inode;
3383
3384 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3385 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3386 /* Wake up anyone waiting on unwritten extent conversion */
3387 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3388 wake_up_all(ext4_ioend_wq(inode));
3389 }
3390}
3391
3392extern const struct iomap_ops ext4_iomap_ops;
3393extern const struct iomap_ops ext4_iomap_report_ops;
3394
3395static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3396{
3397 /*
3398 * If the buffer has the write error flag, we have failed
3399 * to write out data in the block. In this case, we don't
3400 * have to read the block because we may read the old data
3401 * successfully.
3402 */
3403 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3404 set_buffer_uptodate(bh);
3405 return buffer_uptodate(bh);
3406}
3407
3408#endif /* __KERNEL__ */
3409
3410#define EFSBADCRC EBADMSG /* Bad CRC detected */
3411#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3412
3413#endif /* _EXT4_H */