defconfig: exynos9610: Re-add dropped Wi-Fi AP options lost
[GitHub/LineageOS/android_kernel_motorola_exynos9610.git] / fs / ext4 / ext4.h
... / ...
CommitLineData
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#define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION)
44#include <linux/fscrypt.h>
45
46/*
47 * The fourth extended filesystem constants/structures
48 */
49
50/*
51 * with AGGRESSIVE_CHECK allocator runs consistency checks over
52 * structures. these checks slow things down a lot
53 */
54#define AGGRESSIVE_CHECK__
55
56/*
57 * with DOUBLE_CHECK defined mballoc creates persistent in-core
58 * bitmaps, maintains and uses them to check for double allocations
59 */
60#define DOUBLE_CHECK__
61
62/*
63 * Define EXT4FS_DEBUG to produce debug messages
64 */
65#undef EXT4FS_DEBUG
66
67/*
68 * Debug code
69 */
70#ifdef EXT4FS_DEBUG
71#define ext4_debug(f, a...) \
72 do { \
73 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
74 __FILE__, __LINE__, __func__); \
75 printk(KERN_DEBUG f, ## a); \
76 } while (0)
77#else
78#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
79#endif
80
81/*
82 * Turn on EXT_DEBUG to get lots of info about extents operations.
83 */
84#define EXT_DEBUG__
85#ifdef EXT_DEBUG
86#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
87#else
88#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
89#endif
90
91/* data type for block offset of block group */
92typedef int ext4_grpblk_t;
93
94/* data type for filesystem-wide blocks number */
95typedef unsigned long long ext4_fsblk_t;
96
97/* data type for file logical block number */
98typedef __u32 ext4_lblk_t;
99
100/* data type for block group number */
101typedef unsigned int ext4_group_t;
102
103enum SHIFT_DIRECTION {
104 SHIFT_LEFT = 0,
105 SHIFT_RIGHT,
106};
107
108/*
109 * Flags used in mballoc's allocation_context flags field.
110 *
111 * Also used to show what's going on for debugging purposes when the
112 * flag field is exported via the traceport interface
113 */
114
115/* prefer goal again. length */
116#define EXT4_MB_HINT_MERGE 0x0001
117/* blocks already reserved */
118#define EXT4_MB_HINT_RESERVED 0x0002
119/* metadata is being allocated */
120#define EXT4_MB_HINT_METADATA 0x0004
121/* first blocks in the file */
122#define EXT4_MB_HINT_FIRST 0x0008
123/* search for the best chunk */
124#define EXT4_MB_HINT_BEST 0x0010
125/* data is being allocated */
126#define EXT4_MB_HINT_DATA 0x0020
127/* don't preallocate (for tails) */
128#define EXT4_MB_HINT_NOPREALLOC 0x0040
129/* allocate for locality group */
130#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
131/* allocate goal blocks or none */
132#define EXT4_MB_HINT_GOAL_ONLY 0x0100
133/* goal is meaningful */
134#define EXT4_MB_HINT_TRY_GOAL 0x0200
135/* blocks already pre-reserved by delayed allocation */
136#define EXT4_MB_DELALLOC_RESERVED 0x0400
137/* We are doing stream allocation */
138#define EXT4_MB_STREAM_ALLOC 0x0800
139/* Use reserved root blocks if needed */
140#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
141/* Use blocks from reserved pool */
142#define EXT4_MB_USE_RESERVED 0x2000
143
144struct ext4_allocation_request {
145 /* target inode for block we're allocating */
146 struct inode *inode;
147 /* how many blocks we want to allocate */
148 unsigned int len;
149 /* logical block in target inode */
150 ext4_lblk_t logical;
151 /* the closest logical allocated block to the left */
152 ext4_lblk_t lleft;
153 /* the closest logical allocated block to the right */
154 ext4_lblk_t lright;
155 /* phys. target (a hint) */
156 ext4_fsblk_t goal;
157 /* phys. block for the closest logical allocated block to the left */
158 ext4_fsblk_t pleft;
159 /* phys. block for the closest logical allocated block to the right */
160 ext4_fsblk_t pright;
161 /* flags. see above EXT4_MB_HINT_* */
162 unsigned int flags;
163};
164
165/*
166 * Logical to physical block mapping, used by ext4_map_blocks()
167 *
168 * This structure is used to pass requests into ext4_map_blocks() as
169 * well as to store the information returned by ext4_map_blocks(). It
170 * takes less room on the stack than a struct buffer_head.
171 */
172#define EXT4_MAP_NEW (1 << BH_New)
173#define EXT4_MAP_MAPPED (1 << BH_Mapped)
174#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
175#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
176#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
177 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
178
179struct ext4_map_blocks {
180 ext4_fsblk_t m_pblk;
181 ext4_lblk_t m_lblk;
182 unsigned int m_len;
183 unsigned int m_flags;
184};
185
186/*
187 * Flags for ext4_io_end->flags
188 */
189#define EXT4_IO_END_UNWRITTEN 0x0001
190
191/*
192 * For converting unwritten extents on a work queue. 'handle' is used for
193 * buffered writeback.
194 */
195typedef struct ext4_io_end {
196 struct list_head list; /* per-file finished IO list */
197 handle_t *handle; /* handle reserved for extent
198 * conversion */
199 struct inode *inode; /* file being written to */
200 struct bio *bio; /* Linked list of completed
201 * bios covering the extent */
202 unsigned int flag; /* unwritten or not */
203 atomic_t count; /* reference counter */
204 loff_t offset; /* offset in the file */
205 ssize_t size; /* size of the extent */
206} ext4_io_end_t;
207
208#define EXT4_IO_ENCRYPTED 1
209
210struct ext4_io_submit {
211 unsigned int io_flags;
212 struct writeback_control *io_wbc;
213 struct bio *io_bio;
214 ext4_io_end_t *io_end;
215 sector_t io_next_block;
216};
217
218/*
219 * Special inodes numbers
220 */
221#define EXT4_BAD_INO 1 /* Bad blocks inode */
222#define EXT4_ROOT_INO 2 /* Root inode */
223#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
224#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
225#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
226#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
227#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
228#define EXT4_JOURNAL_INO 8 /* Journal inode */
229
230/* First non-reserved inode for old ext4 filesystems */
231#define EXT4_GOOD_OLD_FIRST_INO 11
232
233/*
234 * Maximal count of links to a file
235 */
236#define EXT4_LINK_MAX 65000
237
238/*
239 * Macro-instructions used to manage several block sizes
240 */
241#define EXT4_MIN_BLOCK_SIZE 1024
242#define EXT4_MAX_BLOCK_SIZE 65536
243#define EXT4_MIN_BLOCK_LOG_SIZE 10
244#define EXT4_MAX_BLOCK_LOG_SIZE 16
245#define EXT4_MAX_CLUSTER_LOG_SIZE 30
246#ifdef __KERNEL__
247# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
248#else
249# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
250#endif
251#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
252#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
253 EXT4_SB(s)->s_cluster_bits)
254#ifdef __KERNEL__
255# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
256# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
257#else
258# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
259#endif
260#ifdef __KERNEL__
261#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
262#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
263#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
264#else
265#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
266 EXT4_GOOD_OLD_INODE_SIZE : \
267 (s)->s_inode_size)
268#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
269 EXT4_GOOD_OLD_FIRST_INO : \
270 (s)->s_first_ino)
271#endif
272#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
273#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
274 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
275 blkbits))
276
277/* Translate a block number to a cluster number */
278#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
279/* Translate a cluster number to a block number */
280#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
281/* Translate # of blks to # of clusters */
282#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
283 (sbi)->s_cluster_bits)
284/* Mask out the low bits to get the starting block of the cluster */
285#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
286 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
287#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
288 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
289/* Get the cluster offset */
290#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
291 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
292#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
293 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
294
295/*
296 * Structure of a blocks group descriptor
297 */
298struct ext4_group_desc
299{
300 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
301 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
302 __le32 bg_inode_table_lo; /* Inodes table block */
303 __le16 bg_free_blocks_count_lo;/* Free blocks count */
304 __le16 bg_free_inodes_count_lo;/* Free inodes count */
305 __le16 bg_used_dirs_count_lo; /* Directories count */
306 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
307 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
308 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
309 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
310 __le16 bg_itable_unused_lo; /* Unused inodes count */
311 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
312 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
313 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
314 __le32 bg_inode_table_hi; /* Inodes table block MSB */
315 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
316 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
317 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
318 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
319 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
320 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
321 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
322 __u32 bg_reserved;
323};
324
325#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
326 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
327 sizeof(__le16))
328#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
329 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
330 sizeof(__le16))
331
332/*
333 * Structure of a flex block group info
334 */
335
336struct flex_groups {
337 atomic64_t free_clusters;
338 atomic_t free_inodes;
339 atomic_t used_dirs;
340};
341
342#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
343#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
344#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
345
346/*
347 * Macro-instructions used to manage group descriptors
348 */
349#define EXT4_MIN_DESC_SIZE 32
350#define EXT4_MIN_DESC_SIZE_64BIT 64
351#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
352#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
353#ifdef __KERNEL__
354# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
355# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
356# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
357# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
358# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
359#else
360# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
361# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
362# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
363#endif
364
365/*
366 * Constants relative to the data blocks
367 */
368#define EXT4_NDIR_BLOCKS 12
369#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
370#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
371#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
372#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
373
374/*
375 * Inode flags
376 */
377#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
378#define EXT4_UNRM_FL 0x00000002 /* Undelete */
379#define EXT4_COMPR_FL 0x00000004 /* Compress file */
380#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
381#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
382#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
383#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
384#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
385/* Reserved for compression usage... */
386#define EXT4_DIRTY_FL 0x00000100
387#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
388#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
389 /* nb: was previously EXT2_ECOMPR_FL */
390#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
391/* End compression flags --- maybe not all used */
392#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
393#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
394#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
395#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
396#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
397#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
398#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
399#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
400#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
401#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
402#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
403#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
404#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
405
406#define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
407#define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
408
409/* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
410#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
411 EXT4_IMMUTABLE_FL | \
412 EXT4_APPEND_FL | \
413 EXT4_NODUMP_FL | \
414 EXT4_NOATIME_FL | \
415 EXT4_PROJINHERIT_FL)
416
417/* Flags that should be inherited by new inodes from their parent. */
418#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
419 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
420 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
421 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
422 EXT4_PROJINHERIT_FL)
423
424/* Flags that are appropriate for regular files (all but dir-specific ones). */
425#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
426
427/* Flags that are appropriate for non-directories/regular files. */
428#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
429
430/* The only flags that should be swapped */
431#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
432
433/* Mask out flags that are inappropriate for the given type of inode. */
434static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
435{
436 if (S_ISDIR(mode))
437 return flags;
438 else if (S_ISREG(mode))
439 return flags & EXT4_REG_FLMASK;
440 else
441 return flags & EXT4_OTHER_FLMASK;
442}
443
444/*
445 * Inode flags used for atomic set/get
446 */
447enum {
448 EXT4_INODE_SECRM = 0, /* Secure deletion */
449 EXT4_INODE_UNRM = 1, /* Undelete */
450 EXT4_INODE_COMPR = 2, /* Compress file */
451 EXT4_INODE_SYNC = 3, /* Synchronous updates */
452 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
453 EXT4_INODE_APPEND = 5, /* writes to file may only append */
454 EXT4_INODE_NODUMP = 6, /* do not dump file */
455 EXT4_INODE_NOATIME = 7, /* do not update atime */
456/* Reserved for compression usage... */
457 EXT4_INODE_DIRTY = 8,
458 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
459 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
460 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
461/* End compression flags --- maybe not all used */
462 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
463 EXT4_INODE_IMAGIC = 13, /* AFS directory */
464 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
465 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
466 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
467 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
468 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
469 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
470 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
471 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
472 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
473 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
474 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
475};
476
477/*
478 * Since it's pretty easy to mix up bit numbers and hex values, we use a
479 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
480 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
481 * any extra space in the compiled kernel image, otherwise, the build will fail.
482 * It's important that these values are the same, since we are using
483 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
484 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
485 * values found in ext2, ext3 and ext4 filesystems, and of course the values
486 * defined in e2fsprogs.
487 *
488 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
489 */
490#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
491#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
492
493static inline void ext4_check_flag_values(void)
494{
495 CHECK_FLAG_VALUE(SECRM);
496 CHECK_FLAG_VALUE(UNRM);
497 CHECK_FLAG_VALUE(COMPR);
498 CHECK_FLAG_VALUE(SYNC);
499 CHECK_FLAG_VALUE(IMMUTABLE);
500 CHECK_FLAG_VALUE(APPEND);
501 CHECK_FLAG_VALUE(NODUMP);
502 CHECK_FLAG_VALUE(NOATIME);
503 CHECK_FLAG_VALUE(DIRTY);
504 CHECK_FLAG_VALUE(COMPRBLK);
505 CHECK_FLAG_VALUE(NOCOMPR);
506 CHECK_FLAG_VALUE(ENCRYPT);
507 CHECK_FLAG_VALUE(INDEX);
508 CHECK_FLAG_VALUE(IMAGIC);
509 CHECK_FLAG_VALUE(JOURNAL_DATA);
510 CHECK_FLAG_VALUE(NOTAIL);
511 CHECK_FLAG_VALUE(DIRSYNC);
512 CHECK_FLAG_VALUE(TOPDIR);
513 CHECK_FLAG_VALUE(HUGE_FILE);
514 CHECK_FLAG_VALUE(EXTENTS);
515 CHECK_FLAG_VALUE(EA_INODE);
516 CHECK_FLAG_VALUE(EOFBLOCKS);
517 CHECK_FLAG_VALUE(INLINE_DATA);
518 CHECK_FLAG_VALUE(PROJINHERIT);
519 CHECK_FLAG_VALUE(RESERVED);
520}
521
522/* Used to pass group descriptor data when online resize is done */
523struct ext4_new_group_input {
524 __u32 group; /* Group number for this data */
525 __u64 block_bitmap; /* Absolute block number of block bitmap */
526 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
527 __u64 inode_table; /* Absolute block number of inode table start */
528 __u32 blocks_count; /* Total number of blocks in this group */
529 __u16 reserved_blocks; /* Number of reserved blocks in this group */
530 __u16 unused;
531};
532
533#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
534struct compat_ext4_new_group_input {
535 u32 group;
536 compat_u64 block_bitmap;
537 compat_u64 inode_bitmap;
538 compat_u64 inode_table;
539 u32 blocks_count;
540 u16 reserved_blocks;
541 u16 unused;
542};
543#endif
544
545/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
546struct ext4_new_group_data {
547 __u32 group;
548 __u64 block_bitmap;
549 __u64 inode_bitmap;
550 __u64 inode_table;
551 __u32 blocks_count;
552 __u16 reserved_blocks;
553 __u16 unused;
554 __u32 free_blocks_count;
555};
556
557/* Indexes used to index group tables in ext4_new_group_data */
558enum {
559 BLOCK_BITMAP = 0, /* block bitmap */
560 INODE_BITMAP, /* inode bitmap */
561 INODE_TABLE, /* inode tables */
562 GROUP_TABLE_COUNT,
563};
564
565/*
566 * Flags used by ext4_map_blocks()
567 */
568 /* Allocate any needed blocks and/or convert an unwritten
569 extent to be an initialized ext4 */
570#define EXT4_GET_BLOCKS_CREATE 0x0001
571 /* Request the creation of an unwritten extent */
572#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
573#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
574 EXT4_GET_BLOCKS_CREATE)
575 /* Caller is from the delayed allocation writeout path
576 * finally doing the actual allocation of delayed blocks */
577#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
578 /* caller is from the direct IO path, request to creation of an
579 unwritten extents if not allocated, split the unwritten
580 extent if blocks has been preallocated already*/
581#define EXT4_GET_BLOCKS_PRE_IO 0x0008
582#define EXT4_GET_BLOCKS_CONVERT 0x0010
583#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
584 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
585 /* Convert extent to initialized after IO complete */
586#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
587 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
588 /* Eventual metadata allocation (due to growing extent tree)
589 * should not fail, so try to use reserved blocks for that.*/
590#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
591 /* Don't normalize allocation size (used for fallocate) */
592#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
593 /* Request will not result in inode size update (user for fallocate) */
594#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
595 /* Convert written extents to unwritten */
596#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
597 /* Write zeros to newly created written extents */
598#define EXT4_GET_BLOCKS_ZERO 0x0200
599#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
600 EXT4_GET_BLOCKS_ZERO)
601 /* Caller will submit data before dropping transaction handle. This
602 * allows jbd2 to avoid submitting data before commit. */
603#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
604
605/*
606 * The bit position of these flags must not overlap with any of the
607 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
608 * read_extent_tree_block(), ext4_split_extent_at(),
609 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
610 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
611 * caching the extents when reading from the extent tree while a
612 * truncate or punch hole operation is in progress.
613 */
614#define EXT4_EX_NOCACHE 0x40000000
615#define EXT4_EX_FORCE_CACHE 0x20000000
616
617/*
618 * Flags used by ext4_free_blocks
619 */
620#define EXT4_FREE_BLOCKS_METADATA 0x0001
621#define EXT4_FREE_BLOCKS_FORGET 0x0002
622#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
623#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
624#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
625#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
626
627/*
628 * ioctl commands
629 */
630#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
631#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
632#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
633#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
634#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
635#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
636#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
637#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
638#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
639#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
640#define EXT4_IOC_MIGRATE _IO('f', 9)
641 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
642 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
643#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
644#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
645#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
646#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
647#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
648#define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
649#define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
650#define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
651
652#ifndef FS_IOC_FSGETXATTR
653/* Until the uapi changes get merged for project quota... */
654
655#define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
656#define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
657
658/*
659 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
660 */
661struct fsxattr {
662 __u32 fsx_xflags; /* xflags field value (get/set) */
663 __u32 fsx_extsize; /* extsize field value (get/set)*/
664 __u32 fsx_nextents; /* nextents field value (get) */
665 __u32 fsx_projid; /* project identifier (get/set) */
666 unsigned char fsx_pad[12];
667};
668
669/*
670 * Flags for the fsx_xflags field
671 */
672#define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
673#define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
674#define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
675#define FS_XFLAG_APPEND 0x00000010 /* all writes append */
676#define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
677#define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
678#define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
679#define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
680#define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
681#define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
682#define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
683#define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
684#define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
685#define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
686#define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
687#endif /* !defined(FS_IOC_FSGETXATTR) */
688
689#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
690#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
691
692#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
693
694/*
695 * Flags for going down operation
696 */
697#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
698#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
699#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
700
701
702#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
703/*
704 * ioctl commands in 32 bit emulation
705 */
706#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
707#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
708#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
709#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
710#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
711#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
712#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
713#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
714#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
715#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
716#endif
717
718/* Max physical block we can address w/o extents */
719#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
720
721/* Max logical block we can support */
722#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
723
724/*
725 * Structure of an inode on the disk
726 */
727struct ext4_inode {
728 __le16 i_mode; /* File mode */
729 __le16 i_uid; /* Low 16 bits of Owner Uid */
730 __le32 i_size_lo; /* Size in bytes */
731 __le32 i_atime; /* Access time */
732 __le32 i_ctime; /* Inode Change time */
733 __le32 i_mtime; /* Modification time */
734 __le32 i_dtime; /* Deletion Time */
735 __le16 i_gid; /* Low 16 bits of Group Id */
736 __le16 i_links_count; /* Links count */
737 __le32 i_blocks_lo; /* Blocks count */
738 __le32 i_flags; /* File flags */
739 union {
740 struct {
741 __le32 l_i_version;
742 } linux1;
743 struct {
744 __u32 h_i_translator;
745 } hurd1;
746 struct {
747 __u32 m_i_reserved1;
748 } masix1;
749 } osd1; /* OS dependent 1 */
750 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
751 __le32 i_generation; /* File version (for NFS) */
752 __le32 i_file_acl_lo; /* File ACL */
753 __le32 i_size_high;
754 __le32 i_obso_faddr; /* Obsoleted fragment address */
755 union {
756 struct {
757 __le16 l_i_blocks_high; /* were l_i_reserved1 */
758 __le16 l_i_file_acl_high;
759 __le16 l_i_uid_high; /* these 2 fields */
760 __le16 l_i_gid_high; /* were reserved2[0] */
761 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
762 __le16 l_i_reserved;
763 } linux2;
764 struct {
765 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
766 __u16 h_i_mode_high;
767 __u16 h_i_uid_high;
768 __u16 h_i_gid_high;
769 __u32 h_i_author;
770 } hurd2;
771 struct {
772 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
773 __le16 m_i_file_acl_high;
774 __u32 m_i_reserved2[2];
775 } masix2;
776 } osd2; /* OS dependent 2 */
777 __le16 i_extra_isize;
778 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
779 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
780 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
781 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
782 __le32 i_crtime; /* File Creation time */
783 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
784 __le32 i_version_hi; /* high 32 bits for 64-bit version */
785 __le32 i_projid; /* Project ID */
786};
787
788struct move_extent {
789 __u32 reserved; /* should be zero */
790 __u32 donor_fd; /* donor file descriptor */
791 __u64 orig_start; /* logical start offset in block for orig */
792 __u64 donor_start; /* logical start offset in block for donor */
793 __u64 len; /* block length to be moved */
794 __u64 moved_len; /* moved block length */
795};
796
797#define EXT4_EPOCH_BITS 2
798#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
799#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
800
801/*
802 * Extended fields will fit into an inode if the filesystem was formatted
803 * with large inodes (-I 256 or larger) and there are not currently any EAs
804 * consuming all of the available space. For new inodes we always reserve
805 * enough space for the kernel's known extended fields, but for inodes
806 * created with an old kernel this might not have been the case. None of
807 * the extended inode fields is critical for correct filesystem operation.
808 * This macro checks if a certain field fits in the inode. Note that
809 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
810 */
811#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
812 ((offsetof(typeof(*ext4_inode), field) + \
813 sizeof((ext4_inode)->field)) \
814 <= (EXT4_GOOD_OLD_INODE_SIZE + \
815 (einode)->i_extra_isize)) \
816
817/*
818 * We use an encoding that preserves the times for extra epoch "00":
819 *
820 * extra msb of adjust for signed
821 * epoch 32-bit 32-bit tv_sec to
822 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
823 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
824 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
825 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
826 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
827 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
828 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
829 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
830 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
831 *
832 * Note that previous versions of the kernel on 64-bit systems would
833 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
834 * 1970. e2fsck will correct this, assuming that it is run on the
835 * affected filesystem before 2242.
836 */
837
838static inline __le32 ext4_encode_extra_time(struct timespec *time)
839{
840 u32 extra = sizeof(time->tv_sec) > 4 ?
841 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
842 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
843}
844
845static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
846{
847 if (unlikely(sizeof(time->tv_sec) > 4 &&
848 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
849
850#if 1
851 /* Handle legacy encoding of pre-1970 dates with epoch
852 * bits 1,1. (This backwards compatibility may be removed
853 * at the discretion of the ext4 developers.)
854 */
855 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
856 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
857 extra_bits = 0;
858 time->tv_sec += extra_bits << 32;
859#else
860 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
861#endif
862 }
863 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
864}
865
866#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
867do { \
868 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
869 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
870 (raw_inode)->xtime ## _extra = \
871 ext4_encode_extra_time(&(inode)->xtime); \
872} while (0)
873
874#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
875do { \
876 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
877 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
878 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
879 (raw_inode)->xtime ## _extra = \
880 ext4_encode_extra_time(&(einode)->xtime); \
881} while (0)
882
883#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
884do { \
885 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
886 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
887 ext4_decode_extra_time(&(inode)->xtime, \
888 raw_inode->xtime ## _extra); \
889 else \
890 (inode)->xtime.tv_nsec = 0; \
891} while (0)
892
893#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
894do { \
895 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
896 (einode)->xtime.tv_sec = \
897 (signed)le32_to_cpu((raw_inode)->xtime); \
898 else \
899 (einode)->xtime.tv_sec = 0; \
900 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
901 ext4_decode_extra_time(&(einode)->xtime, \
902 raw_inode->xtime ## _extra); \
903 else \
904 (einode)->xtime.tv_nsec = 0; \
905} while (0)
906
907#define i_disk_version osd1.linux1.l_i_version
908
909#if defined(__KERNEL__) || defined(__linux__)
910#define i_reserved1 osd1.linux1.l_i_reserved1
911#define i_file_acl_high osd2.linux2.l_i_file_acl_high
912#define i_blocks_high osd2.linux2.l_i_blocks_high
913#define i_uid_low i_uid
914#define i_gid_low i_gid
915#define i_uid_high osd2.linux2.l_i_uid_high
916#define i_gid_high osd2.linux2.l_i_gid_high
917#define i_checksum_lo osd2.linux2.l_i_checksum_lo
918
919#elif defined(__GNU__)
920
921#define i_translator osd1.hurd1.h_i_translator
922#define i_uid_high osd2.hurd2.h_i_uid_high
923#define i_gid_high osd2.hurd2.h_i_gid_high
924#define i_author osd2.hurd2.h_i_author
925
926#elif defined(__masix__)
927
928#define i_reserved1 osd1.masix1.m_i_reserved1
929#define i_file_acl_high osd2.masix2.m_i_file_acl_high
930#define i_reserved2 osd2.masix2.m_i_reserved2
931
932#endif /* defined(__KERNEL__) || defined(__linux__) */
933
934#include "extents_status.h"
935
936/*
937 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
938 *
939 * These are needed to avoid lockdep false positives when we need to
940 * allocate blocks to the quota inode during ext4_map_blocks(), while
941 * holding i_data_sem for a normal (non-quota) inode. Since we don't
942 * do quota tracking for the quota inode, this avoids deadlock (as
943 * well as infinite recursion, since it isn't turtles all the way
944 * down...)
945 *
946 * I_DATA_SEM_NORMAL - Used for most inodes
947 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
948 * where the second inode has larger inode number
949 * than the first
950 * I_DATA_SEM_QUOTA - Used for quota inodes only
951 */
952enum {
953 I_DATA_SEM_NORMAL = 0,
954 I_DATA_SEM_OTHER,
955 I_DATA_SEM_QUOTA,
956};
957
958
959/*
960 * fourth extended file system inode data in memory
961 */
962struct ext4_inode_info {
963 __le32 i_data[15]; /* unconverted */
964 __u32 i_dtime;
965 ext4_fsblk_t i_file_acl;
966
967 /*
968 * i_block_group is the number of the block group which contains
969 * this file's inode. Constant across the lifetime of the inode,
970 * it is used for making block allocation decisions - we try to
971 * place a file's data blocks near its inode block, and new inodes
972 * near to their parent directory's inode.
973 */
974 ext4_group_t i_block_group;
975 ext4_lblk_t i_dir_start_lookup;
976#if (BITS_PER_LONG < 64)
977 unsigned long i_state_flags; /* Dynamic state flags */
978#endif
979 unsigned long i_flags;
980
981 /*
982 * Extended attributes can be read independently of the main file
983 * data. Taking i_mutex even when reading would cause contention
984 * between readers of EAs and writers of regular file data, so
985 * instead we synchronize on xattr_sem when reading or changing
986 * EAs.
987 */
988 struct rw_semaphore xattr_sem;
989
990 struct list_head i_orphan; /* unlinked but open inodes */
991
992 /*
993 * i_disksize keeps track of what the inode size is ON DISK, not
994 * in memory. During truncate, i_size is set to the new size by
995 * the VFS prior to calling ext4_truncate(), but the filesystem won't
996 * set i_disksize to 0 until the truncate is actually under way.
997 *
998 * The intent is that i_disksize always represents the blocks which
999 * are used by this file. This allows recovery to restart truncate
1000 * on orphans if we crash during truncate. We actually write i_disksize
1001 * into the on-disk inode when writing inodes out, instead of i_size.
1002 *
1003 * The only time when i_disksize and i_size may be different is when
1004 * a truncate is in progress. The only things which change i_disksize
1005 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1006 */
1007 loff_t i_disksize;
1008
1009 /*
1010 * i_data_sem is for serialising ext4_truncate() against
1011 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1012 * data tree are chopped off during truncate. We can't do that in
1013 * ext4 because whenever we perform intermediate commits during
1014 * truncate, the inode and all the metadata blocks *must* be in a
1015 * consistent state which allows truncation of the orphans to restart
1016 * during recovery. Hence we must fix the get_block-vs-truncate race
1017 * by other means, so we have i_data_sem.
1018 */
1019 struct rw_semaphore i_data_sem;
1020 /*
1021 * i_mmap_sem is for serializing page faults with truncate / punch hole
1022 * operations. We have to make sure that new page cannot be faulted in
1023 * a section of the inode that is being punched. We cannot easily use
1024 * i_data_sem for this since we need protection for the whole punch
1025 * operation and i_data_sem ranks below transaction start so we have
1026 * to occasionally drop it.
1027 */
1028 struct rw_semaphore i_mmap_sem;
1029 struct inode vfs_inode;
1030 struct jbd2_inode *jinode;
1031
1032 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1033
1034 /*
1035 * File creation time. Its function is same as that of
1036 * struct timespec i_{a,c,m}time in the generic inode.
1037 */
1038 struct timespec i_crtime;
1039
1040 /* mballoc */
1041 struct list_head i_prealloc_list;
1042 spinlock_t i_prealloc_lock;
1043
1044 /* extents status tree */
1045 struct ext4_es_tree i_es_tree;
1046 rwlock_t i_es_lock;
1047 struct list_head i_es_list;
1048 unsigned int i_es_all_nr; /* protected by i_es_lock */
1049 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1050 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1051 extents to shrink. Protected by
1052 i_es_lock */
1053
1054 /* ialloc */
1055 ext4_group_t i_last_alloc_group;
1056
1057 /* allocation reservation info for delalloc */
1058 /* In case of bigalloc, this refer to clusters rather than blocks */
1059 unsigned int i_reserved_data_blocks;
1060 ext4_lblk_t i_da_metadata_calc_last_lblock;
1061 int i_da_metadata_calc_len;
1062
1063 /* on-disk additional length */
1064 __u16 i_extra_isize;
1065
1066 /* Indicate the inline data space. */
1067 u16 i_inline_off;
1068 u16 i_inline_size;
1069
1070#ifdef CONFIG_QUOTA
1071 /* quota space reservation, managed internally by quota code */
1072 qsize_t i_reserved_quota;
1073#endif
1074
1075 /* Lock protecting lists below */
1076 spinlock_t i_completed_io_lock;
1077 /*
1078 * Completed IOs that need unwritten extents handling and have
1079 * transaction reserved
1080 */
1081 struct list_head i_rsv_conversion_list;
1082 struct work_struct i_rsv_conversion_work;
1083 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1084
1085 spinlock_t i_block_reservation_lock;
1086
1087 /*
1088 * Transactions that contain inode's metadata needed to complete
1089 * fsync and fdatasync, respectively.
1090 */
1091 tid_t i_sync_tid;
1092 tid_t i_datasync_tid;
1093
1094#ifdef CONFIG_QUOTA
1095 struct dquot *i_dquot[MAXQUOTAS];
1096#endif
1097
1098 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1099 __u32 i_csum_seed;
1100
1101 kprojid_t i_projid;
1102};
1103
1104/*
1105 * File system states
1106 */
1107#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1108#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1109#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1110
1111/*
1112 * Misc. filesystem flags
1113 */
1114#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1115#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1116#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1117
1118/*
1119 * Mount flags set via mount options or defaults
1120 */
1121#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1122#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1123#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1124#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1125#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1126#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1127#define EXT4_MOUNT_ERRORS_MASK 0x00070
1128#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1129#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1130#ifdef CONFIG_FS_DAX
1131#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1132#else
1133#define EXT4_MOUNT_DAX 0
1134#endif
1135#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1136#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1137#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1138#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1139#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1140#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1141#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1142#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1143#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1144#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1145#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1146#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1147 * enable enforcement for hidden
1148 * quota files */
1149#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1150 * enforcement for hidden quota
1151 * files */
1152#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1153 * enforcement */
1154#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1155#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1156#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1157#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1158#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1159#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1160#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1161#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1162
1163/*
1164 * Mount flags set either automatically (could not be set by mount option)
1165 * based on per file system feature or property or in special cases such as
1166 * distinguishing between explicit mount option definition and default.
1167 */
1168#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1169 specified delalloc */
1170#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1171 size of blocksize * 8
1172 blocks */
1173#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1174 file systems */
1175
1176#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1177 specified journal checksum */
1178
1179#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1180 ~EXT4_MOUNT_##opt
1181#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1182 EXT4_MOUNT_##opt
1183#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1184 EXT4_MOUNT_##opt)
1185
1186#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1187 ~EXT4_MOUNT2_##opt
1188#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1189 EXT4_MOUNT2_##opt
1190#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1191 EXT4_MOUNT2_##opt)
1192
1193#define ext4_test_and_set_bit __test_and_set_bit_le
1194#define ext4_set_bit __set_bit_le
1195#define ext4_set_bit_atomic ext2_set_bit_atomic
1196#define ext4_test_and_clear_bit __test_and_clear_bit_le
1197#define ext4_clear_bit __clear_bit_le
1198#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1199#define ext4_test_bit test_bit_le
1200#define ext4_find_next_zero_bit find_next_zero_bit_le
1201#define ext4_find_next_bit find_next_bit_le
1202
1203extern void ext4_set_bits(void *bm, int cur, int len);
1204
1205/*
1206 * Maximal mount counts between two filesystem checks
1207 */
1208#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1209#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1210
1211/*
1212 * Behaviour when detecting errors
1213 */
1214#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1215#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1216#define EXT4_ERRORS_PANIC 3 /* Panic */
1217#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1218
1219/* Metadata checksum algorithm codes */
1220#define EXT4_CRC32C_CHKSUM 1
1221
1222/*
1223 * Structure of the super block
1224 */
1225struct ext4_super_block {
1226/*00*/ __le32 s_inodes_count; /* Inodes count */
1227 __le32 s_blocks_count_lo; /* Blocks count */
1228 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1229 __le32 s_free_blocks_count_lo; /* Free blocks count */
1230/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1231 __le32 s_first_data_block; /* First Data Block */
1232 __le32 s_log_block_size; /* Block size */
1233 __le32 s_log_cluster_size; /* Allocation cluster size */
1234/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1235 __le32 s_clusters_per_group; /* # Clusters per group */
1236 __le32 s_inodes_per_group; /* # Inodes per group */
1237 __le32 s_mtime; /* Mount time */
1238/*30*/ __le32 s_wtime; /* Write time */
1239 __le16 s_mnt_count; /* Mount count */
1240 __le16 s_max_mnt_count; /* Maximal mount count */
1241 __le16 s_magic; /* Magic signature */
1242 __le16 s_state; /* File system state */
1243 __le16 s_errors; /* Behaviour when detecting errors */
1244 __le16 s_minor_rev_level; /* minor revision level */
1245/*40*/ __le32 s_lastcheck; /* time of last check */
1246 __le32 s_checkinterval; /* max. time between checks */
1247 __le32 s_creator_os; /* OS */
1248 __le32 s_rev_level; /* Revision level */
1249/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1250 __le16 s_def_resgid; /* Default gid for reserved blocks */
1251 /*
1252 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1253 *
1254 * Note: the difference between the compatible feature set and
1255 * the incompatible feature set is that if there is a bit set
1256 * in the incompatible feature set that the kernel doesn't
1257 * know about, it should refuse to mount the filesystem.
1258 *
1259 * e2fsck's requirements are more strict; if it doesn't know
1260 * about a feature in either the compatible or incompatible
1261 * feature set, it must abort and not try to meddle with
1262 * things it doesn't understand...
1263 */
1264 __le32 s_first_ino; /* First non-reserved inode */
1265 __le16 s_inode_size; /* size of inode structure */
1266 __le16 s_block_group_nr; /* block group # of this superblock */
1267 __le32 s_feature_compat; /* compatible feature set */
1268/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1269 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1270/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1271/*78*/ char s_volume_name[16]; /* volume name */
1272/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1273/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1274 /*
1275 * Performance hints. Directory preallocation should only
1276 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1277 */
1278 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1279 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1280 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1281 /*
1282 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1283 */
1284/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1285/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1286 __le32 s_journal_dev; /* device number of journal file */
1287 __le32 s_last_orphan; /* start of list of inodes to delete */
1288 __le32 s_hash_seed[4]; /* HTREE hash seed */
1289 __u8 s_def_hash_version; /* Default hash version to use */
1290 __u8 s_jnl_backup_type;
1291 __le16 s_desc_size; /* size of group descriptor */
1292/*100*/ __le32 s_default_mount_opts;
1293 __le32 s_first_meta_bg; /* First metablock block group */
1294 __le32 s_mkfs_time; /* When the filesystem was created */
1295 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1296 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1297/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1298 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1299 __le32 s_free_blocks_count_hi; /* Free blocks count */
1300 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1301 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1302 __le32 s_flags; /* Miscellaneous flags */
1303 __le16 s_raid_stride; /* RAID stride */
1304 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1305 __le64 s_mmp_block; /* Block for multi-mount protection */
1306 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1307 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1308 __u8 s_checksum_type; /* metadata checksum algorithm used */
1309 __u8 s_encryption_level; /* versioning level for encryption */
1310 __u8 s_reserved_pad; /* Padding to next 32bits */
1311 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1312 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1313 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1314 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1315 snapshot's future use */
1316 __le32 s_snapshot_list; /* inode number of the head of the
1317 on-disk snapshot list */
1318#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1319 __le32 s_error_count; /* number of fs errors */
1320 __le32 s_first_error_time; /* first time an error happened */
1321 __le32 s_first_error_ino; /* inode involved in first error */
1322 __le64 s_first_error_block; /* block involved of first error */
1323 __u8 s_first_error_func[32]; /* function where the error happened */
1324 __le32 s_first_error_line; /* line number where error happened */
1325 __le32 s_last_error_time; /* most recent time of an error */
1326 __le32 s_last_error_ino; /* inode involved in last error */
1327 __le32 s_last_error_line; /* line number where error happened */
1328 __le64 s_last_error_block; /* block involved of last error */
1329 __u8 s_last_error_func[32]; /* function where the error happened */
1330#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1331 __u8 s_mount_opts[64];
1332 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1333 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1334 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1335 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1336 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1337 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1338 __le32 s_lpf_ino; /* Location of the lost+found inode */
1339 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1340 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1341 __le32 s_reserved[98]; /* Padding to the end of the block */
1342 __le32 s_checksum; /* crc32c(superblock) */
1343};
1344
1345#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1346
1347#ifdef __KERNEL__
1348
1349/*
1350 * run-time mount flags
1351 */
1352#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1353#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1354#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1355
1356#ifdef CONFIG_EXT4_FS_ENCRYPTION
1357#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1358 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1359#else
1360#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1361#endif
1362
1363/* Number of quota types we support */
1364#define EXT4_MAXQUOTAS 3
1365
1366/*
1367 * fourth extended-fs super-block data in memory
1368 */
1369struct ext4_sb_info {
1370 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1371 unsigned long s_inodes_per_block;/* Number of inodes per block */
1372 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1373 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1374 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1375 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1376 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1377 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1378 ext4_group_t s_groups_count; /* Number of groups in the fs */
1379 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1380 unsigned long s_overhead; /* # of fs overhead clusters */
1381 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1382 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1383 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1384 struct buffer_head * s_sbh; /* Buffer containing the super block */
1385 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1386 struct buffer_head **s_group_desc;
1387 unsigned int s_mount_opt;
1388 unsigned int s_mount_opt2;
1389 unsigned int s_mount_flags;
1390 unsigned int s_def_mount_opt;
1391 ext4_fsblk_t s_sb_block;
1392 atomic64_t s_resv_clusters;
1393 kuid_t s_resuid;
1394 kgid_t s_resgid;
1395 unsigned short s_mount_state;
1396 unsigned short s_pad;
1397 int s_addr_per_block_bits;
1398 int s_desc_per_block_bits;
1399 int s_inode_size;
1400 int s_first_ino;
1401 unsigned int s_inode_readahead_blks;
1402 unsigned int s_inode_goal;
1403 spinlock_t s_next_gen_lock;
1404 u32 s_next_generation;
1405 u32 s_hash_seed[4];
1406 int s_def_hash_version;
1407 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1408 struct percpu_counter s_freeclusters_counter;
1409 struct percpu_counter s_freeinodes_counter;
1410 struct percpu_counter s_dirs_counter;
1411 struct percpu_counter s_dirtyclusters_counter;
1412 struct blockgroup_lock *s_blockgroup_lock;
1413 struct proc_dir_entry *s_proc;
1414 struct kobject s_kobj;
1415 struct completion s_kobj_unregister;
1416 struct super_block *s_sb;
1417
1418 /* Journaling */
1419 struct journal_s *s_journal;
1420 struct list_head s_orphan;
1421 struct mutex s_orphan_lock;
1422 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1423 unsigned long s_commit_interval;
1424 u32 s_max_batch_time;
1425 u32 s_min_batch_time;
1426 struct block_device *journal_bdev;
1427#ifdef CONFIG_QUOTA
1428 /* Names of quota files with journalled quota */
1429 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1430 int s_jquota_fmt; /* Format of quota to use */
1431#endif
1432 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1433 struct rb_root system_blks;
1434
1435#ifdef EXTENTS_STATS
1436 /* ext4 extents stats */
1437 unsigned long s_ext_min;
1438 unsigned long s_ext_max;
1439 unsigned long s_depth_max;
1440 spinlock_t s_ext_stats_lock;
1441 unsigned long s_ext_blocks;
1442 unsigned long s_ext_extents;
1443#endif
1444
1445 /* for buddy allocator */
1446 struct ext4_group_info ***s_group_info;
1447 struct inode *s_buddy_cache;
1448 spinlock_t s_md_lock;
1449 unsigned short *s_mb_offsets;
1450 unsigned int *s_mb_maxs;
1451 unsigned int s_group_info_size;
1452 unsigned int s_mb_free_pending;
1453 struct list_head s_freed_data_list; /* List of blocks to be freed
1454 after commit completed */
1455
1456 /* tunables */
1457 unsigned long s_stripe;
1458 unsigned int s_mb_stream_request;
1459 unsigned int s_mb_max_to_scan;
1460 unsigned int s_mb_min_to_scan;
1461 unsigned int s_mb_stats;
1462 unsigned int s_mb_order2_reqs;
1463 unsigned int s_mb_group_prealloc;
1464 unsigned int s_max_dir_size_kb;
1465 /* where last allocation was done - for stream allocation */
1466 unsigned long s_mb_last_group;
1467 unsigned long s_mb_last_start;
1468
1469 /* stats for buddy allocator */
1470 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1471 atomic_t s_bal_success; /* we found long enough chunks */
1472 atomic_t s_bal_allocated; /* in blocks */
1473 atomic_t s_bal_ex_scanned; /* total extents scanned */
1474 atomic_t s_bal_goals; /* goal hits */
1475 atomic_t s_bal_breaks; /* too long searches */
1476 atomic_t s_bal_2orders; /* 2^order hits */
1477 spinlock_t s_bal_lock;
1478 unsigned long s_mb_buddies_generated;
1479 unsigned long long s_mb_generation_time;
1480 atomic_t s_mb_lost_chunks;
1481 atomic_t s_mb_preallocated;
1482 atomic_t s_mb_discarded;
1483 atomic_t s_lock_busy;
1484
1485 /* locality groups */
1486 struct ext4_locality_group __percpu *s_locality_groups;
1487
1488 /* for write statistics */
1489 unsigned long s_sectors_written_start;
1490 u64 s_kbytes_written;
1491
1492 /* the size of zero-out chunk */
1493 unsigned int s_extent_max_zeroout_kb;
1494
1495 unsigned int s_log_groups_per_flex;
1496 struct flex_groups *s_flex_groups;
1497 ext4_group_t s_flex_groups_allocated;
1498
1499 /* workqueue for reserved extent conversions (buffered io) */
1500 struct workqueue_struct *rsv_conversion_wq;
1501
1502 /* timer for periodic error stats printing */
1503 struct timer_list s_err_report;
1504
1505 /* Lazy inode table initialization info */
1506 struct ext4_li_request *s_li_request;
1507 /* Wait multiplier for lazy initialization thread */
1508 unsigned int s_li_wait_mult;
1509
1510 /* Kernel thread for multiple mount protection */
1511 struct task_struct *s_mmp_tsk;
1512
1513 /* record the last minlen when FITRIM is called. */
1514 atomic_t s_last_trim_minblks;
1515
1516 /* Reference to checksum algorithm driver via cryptoapi */
1517 struct crypto_shash *s_chksum_driver;
1518
1519 /* Precomputed FS UUID checksum for seeding other checksums */
1520 __u32 s_csum_seed;
1521
1522 /* Reclaim extents from extent status tree */
1523 struct shrinker s_es_shrinker;
1524 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1525 long s_es_nr_inode;
1526 struct ext4_es_stats s_es_stats;
1527 struct mb_cache *s_ea_block_cache;
1528 struct mb_cache *s_ea_inode_cache;
1529 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1530
1531 /* Ratelimit ext4 messages. */
1532 struct ratelimit_state s_err_ratelimit_state;
1533 struct ratelimit_state s_warning_ratelimit_state;
1534 struct ratelimit_state s_msg_ratelimit_state;
1535
1536 /* Barrier between changing inodes' journal flags and writepages ops. */
1537 struct percpu_rw_semaphore s_journal_flag_rwsem;
1538 struct dax_device *s_daxdev;
1539};
1540
1541static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1542{
1543 return sb->s_fs_info;
1544}
1545static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1546{
1547 return container_of(inode, struct ext4_inode_info, vfs_inode);
1548}
1549
1550static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1551{
1552 return ino == EXT4_ROOT_INO ||
1553 (ino >= EXT4_FIRST_INO(sb) &&
1554 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1555}
1556
1557/*
1558 * Inode dynamic state flags
1559 */
1560enum {
1561 EXT4_STATE_JDATA, /* journaled data exists */
1562 EXT4_STATE_NEW, /* inode is newly created */
1563 EXT4_STATE_XATTR, /* has in-inode xattrs */
1564 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1565 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1566 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1567 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1568 EXT4_STATE_NEWENTRY, /* File just added to dir */
1569 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1570 nolocking */
1571 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1572 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1573 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1574};
1575
1576#define EXT4_INODE_BIT_FNS(name, field, offset) \
1577static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1578{ \
1579 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1580} \
1581static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1582{ \
1583 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1584} \
1585static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1586{ \
1587 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1588}
1589
1590/* Add these declarations here only so that these functions can be
1591 * found by name. Otherwise, they are very hard to locate. */
1592static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1593static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1594static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1595EXT4_INODE_BIT_FNS(flag, flags, 0)
1596
1597/* Add these declarations here only so that these functions can be
1598 * found by name. Otherwise, they are very hard to locate. */
1599static inline int ext4_test_inode_state(struct inode *inode, int bit);
1600static inline void ext4_set_inode_state(struct inode *inode, int bit);
1601static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1602#if (BITS_PER_LONG < 64)
1603EXT4_INODE_BIT_FNS(state, state_flags, 0)
1604
1605static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1606{
1607 (ei)->i_state_flags = 0;
1608}
1609#else
1610EXT4_INODE_BIT_FNS(state, flags, 32)
1611
1612static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1613{
1614 /* We depend on the fact that callers will set i_flags */
1615}
1616#endif
1617#else
1618/* Assume that user mode programs are passing in an ext4fs superblock, not
1619 * a kernel struct super_block. This will allow us to call the feature-test
1620 * macros from user land. */
1621#define EXT4_SB(sb) (sb)
1622#endif
1623
1624/*
1625 * Returns true if the inode is inode is encrypted
1626 */
1627#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1628
1629/*
1630 * Codes for operating systems
1631 */
1632#define EXT4_OS_LINUX 0
1633#define EXT4_OS_HURD 1
1634#define EXT4_OS_MASIX 2
1635#define EXT4_OS_FREEBSD 3
1636#define EXT4_OS_LITES 4
1637
1638/*
1639 * Revision levels
1640 */
1641#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1642#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1643
1644#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1645#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1646
1647#define EXT4_GOOD_OLD_INODE_SIZE 128
1648
1649/*
1650 * Feature set definitions
1651 */
1652
1653#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1654#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1655#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1656#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1657#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1658#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1659#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1660
1661#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1662#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1663#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1664#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1665#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1666#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1667#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1668#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1669#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1670/*
1671 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1672 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1673 * all other data structures' checksums. However, the METADATA_CSUM and
1674 * GDT_CSUM bits are mutually exclusive.
1675 */
1676#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1677#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1678#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1679
1680#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1681#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1682#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1683#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1684#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1685#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1686#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1687#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1688#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1689#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1690#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1691#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1692#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1693#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1694#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1695
1696#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1697static inline bool ext4_has_feature_##name(struct super_block *sb) \
1698{ \
1699 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1700 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1701} \
1702static inline void ext4_set_feature_##name(struct super_block *sb) \
1703{ \
1704 EXT4_SB(sb)->s_es->s_feature_compat |= \
1705 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1706} \
1707static inline void ext4_clear_feature_##name(struct super_block *sb) \
1708{ \
1709 EXT4_SB(sb)->s_es->s_feature_compat &= \
1710 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1711}
1712
1713#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1714static inline bool ext4_has_feature_##name(struct super_block *sb) \
1715{ \
1716 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1717 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1718} \
1719static inline void ext4_set_feature_##name(struct super_block *sb) \
1720{ \
1721 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1722 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1723} \
1724static inline void ext4_clear_feature_##name(struct super_block *sb) \
1725{ \
1726 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1727 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1728}
1729
1730#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1731static inline bool ext4_has_feature_##name(struct super_block *sb) \
1732{ \
1733 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1734 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1735} \
1736static inline void ext4_set_feature_##name(struct super_block *sb) \
1737{ \
1738 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1739 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1740} \
1741static inline void ext4_clear_feature_##name(struct super_block *sb) \
1742{ \
1743 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1744 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1745}
1746
1747EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1748EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1749EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1750EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1751EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1752EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1753EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1754
1755EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1756EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1757EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1758EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1759EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1760EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1761EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1762EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1763EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1764EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1765EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1766EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1767
1768EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1769EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1770EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1771EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1772EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1773EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1774EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1775EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1776EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1777EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1778EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1779EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1780EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1781EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1782EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1783
1784#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1785#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1786 EXT4_FEATURE_INCOMPAT_META_BG)
1787#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1788 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1789 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1790
1791#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1792#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1793 EXT4_FEATURE_INCOMPAT_RECOVER| \
1794 EXT4_FEATURE_INCOMPAT_META_BG)
1795#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1796 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1797 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1798
1799#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1800#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1801 EXT4_FEATURE_INCOMPAT_RECOVER| \
1802 EXT4_FEATURE_INCOMPAT_META_BG| \
1803 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1804 EXT4_FEATURE_INCOMPAT_64BIT| \
1805 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1806 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1807 EXT4_FEATURE_INCOMPAT_MMP | \
1808 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1809 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1810 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1811 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1812#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1813 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1814 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1815 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1816 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1817 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1818 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1819 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1820 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1821 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1822 EXT4_FEATURE_RO_COMPAT_PROJECT)
1823
1824#define EXTN_FEATURE_FUNCS(ver) \
1825static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1826{ \
1827 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1828 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1829} \
1830static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1831{ \
1832 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1833 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1834} \
1835static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1836{ \
1837 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1838 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1839}
1840
1841EXTN_FEATURE_FUNCS(2)
1842EXTN_FEATURE_FUNCS(3)
1843EXTN_FEATURE_FUNCS(4)
1844
1845static inline bool ext4_has_compat_features(struct super_block *sb)
1846{
1847 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1848}
1849static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1850{
1851 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1852}
1853static inline bool ext4_has_incompat_features(struct super_block *sb)
1854{
1855 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1856}
1857
1858/*
1859 * Superblock flags
1860 */
1861#define EXT4_FLAGS_RESIZING 0
1862#define EXT4_FLAGS_SHUTDOWN 1
1863
1864static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1865{
1866 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1867}
1868
1869
1870/*
1871 * Default values for user and/or group using reserved blocks
1872 */
1873#define EXT4_DEF_RESUID 0
1874#define EXT4_DEF_RESGID 0
1875
1876/*
1877 * Default project ID
1878 */
1879#define EXT4_DEF_PROJID 0
1880
1881#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1882
1883/*
1884 * Default mount options
1885 */
1886#define EXT4_DEFM_DEBUG 0x0001
1887#define EXT4_DEFM_BSDGROUPS 0x0002
1888#define EXT4_DEFM_XATTR_USER 0x0004
1889#define EXT4_DEFM_ACL 0x0008
1890#define EXT4_DEFM_UID16 0x0010
1891#define EXT4_DEFM_JMODE 0x0060
1892#define EXT4_DEFM_JMODE_DATA 0x0020
1893#define EXT4_DEFM_JMODE_ORDERED 0x0040
1894#define EXT4_DEFM_JMODE_WBACK 0x0060
1895#define EXT4_DEFM_NOBARRIER 0x0100
1896#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1897#define EXT4_DEFM_DISCARD 0x0400
1898#define EXT4_DEFM_NODELALLOC 0x0800
1899
1900/*
1901 * Default journal batch times
1902 */
1903#define EXT4_DEF_MIN_BATCH_TIME 0
1904#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1905
1906/*
1907 * Minimum number of groups in a flexgroup before we separate out
1908 * directories into the first block group of a flexgroup
1909 */
1910#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1911
1912/*
1913 * Structure of a directory entry
1914 */
1915#define EXT4_NAME_LEN 255
1916
1917struct ext4_dir_entry {
1918 __le32 inode; /* Inode number */
1919 __le16 rec_len; /* Directory entry length */
1920 __le16 name_len; /* Name length */
1921 char name[EXT4_NAME_LEN]; /* File name */
1922};
1923
1924/*
1925 * The new version of the directory entry. Since EXT4 structures are
1926 * stored in intel byte order, and the name_len field could never be
1927 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1928 * file_type field.
1929 */
1930struct ext4_dir_entry_2 {
1931 __le32 inode; /* Inode number */
1932 __le16 rec_len; /* Directory entry length */
1933 __u8 name_len; /* Name length */
1934 __u8 file_type;
1935 char name[EXT4_NAME_LEN]; /* File name */
1936};
1937
1938/*
1939 * This is a bogus directory entry at the end of each leaf block that
1940 * records checksums.
1941 */
1942struct ext4_dir_entry_tail {
1943 __le32 det_reserved_zero1; /* Pretend to be unused */
1944 __le16 det_rec_len; /* 12 */
1945 __u8 det_reserved_zero2; /* Zero name length */
1946 __u8 det_reserved_ft; /* 0xDE, fake file type */
1947 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1948};
1949
1950#define EXT4_DIRENT_TAIL(block, blocksize) \
1951 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1952 ((blocksize) - \
1953 sizeof(struct ext4_dir_entry_tail))))
1954
1955/*
1956 * Ext4 directory file types. Only the low 3 bits are used. The
1957 * other bits are reserved for now.
1958 */
1959#define EXT4_FT_UNKNOWN 0
1960#define EXT4_FT_REG_FILE 1
1961#define EXT4_FT_DIR 2
1962#define EXT4_FT_CHRDEV 3
1963#define EXT4_FT_BLKDEV 4
1964#define EXT4_FT_FIFO 5
1965#define EXT4_FT_SOCK 6
1966#define EXT4_FT_SYMLINK 7
1967
1968#define EXT4_FT_MAX 8
1969
1970#define EXT4_FT_DIR_CSUM 0xDE
1971
1972/*
1973 * EXT4_DIR_PAD defines the directory entries boundaries
1974 *
1975 * NOTE: It must be a multiple of 4
1976 */
1977#define EXT4_DIR_PAD 4
1978#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1979#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1980 ~EXT4_DIR_ROUND)
1981#define EXT4_MAX_REC_LEN ((1<<16)-1)
1982
1983/*
1984 * If we ever get support for fs block sizes > page_size, we'll need
1985 * to remove the #if statements in the next two functions...
1986 */
1987static inline unsigned int
1988ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1989{
1990 unsigned len = le16_to_cpu(dlen);
1991
1992#if (PAGE_SIZE >= 65536)
1993 if (len == EXT4_MAX_REC_LEN || len == 0)
1994 return blocksize;
1995 return (len & 65532) | ((len & 3) << 16);
1996#else
1997 return len;
1998#endif
1999}
2000
2001static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2002{
2003 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2004 BUG();
2005#if (PAGE_SIZE >= 65536)
2006 if (len < 65536)
2007 return cpu_to_le16(len);
2008 if (len == blocksize) {
2009 if (blocksize == 65536)
2010 return cpu_to_le16(EXT4_MAX_REC_LEN);
2011 else
2012 return cpu_to_le16(0);
2013 }
2014 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2015#else
2016 return cpu_to_le16(len);
2017#endif
2018}
2019
2020/*
2021 * Hash Tree Directory indexing
2022 * (c) Daniel Phillips, 2001
2023 */
2024
2025#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2026 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2027#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2028 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2029#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2030
2031/* Legal values for the dx_root hash_version field: */
2032
2033#define DX_HASH_LEGACY 0
2034#define DX_HASH_HALF_MD4 1
2035#define DX_HASH_TEA 2
2036#define DX_HASH_LEGACY_UNSIGNED 3
2037#define DX_HASH_HALF_MD4_UNSIGNED 4
2038#define DX_HASH_TEA_UNSIGNED 5
2039
2040static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2041 const void *address, unsigned int length)
2042{
2043 struct {
2044 struct shash_desc shash;
2045 char ctx[4];
2046 } desc;
2047 int err;
2048
2049 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2050
2051 desc.shash.tfm = sbi->s_chksum_driver;
2052 desc.shash.flags = 0;
2053 *(u32 *)desc.ctx = crc;
2054
2055 err = crypto_shash_update(&desc.shash, address, length);
2056 BUG_ON(err);
2057
2058 return *(u32 *)desc.ctx;
2059}
2060
2061#ifdef __KERNEL__
2062
2063/* hash info structure used by the directory hash */
2064struct dx_hash_info
2065{
2066 u32 hash;
2067 u32 minor_hash;
2068 int hash_version;
2069 u32 *seed;
2070};
2071
2072
2073/* 32 and 64 bit signed EOF for dx directories */
2074#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2075#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2076
2077
2078/*
2079 * Control parameters used by ext4_htree_next_block
2080 */
2081#define HASH_NB_ALWAYS 1
2082
2083struct ext4_filename {
2084 const struct qstr *usr_fname;
2085 struct fscrypt_str disk_name;
2086 struct dx_hash_info hinfo;
2087#ifdef CONFIG_EXT4_FS_ENCRYPTION
2088 struct fscrypt_str crypto_buf;
2089#endif
2090};
2091
2092#define fname_name(p) ((p)->disk_name.name)
2093#define fname_len(p) ((p)->disk_name.len)
2094
2095/*
2096 * Describe an inode's exact location on disk and in memory
2097 */
2098struct ext4_iloc
2099{
2100 struct buffer_head *bh;
2101 unsigned long offset;
2102 ext4_group_t block_group;
2103};
2104
2105static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2106{
2107 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2108}
2109
2110static inline bool ext4_is_quota_file(struct inode *inode)
2111{
2112 return IS_NOQUOTA(inode) &&
2113 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2114}
2115
2116/*
2117 * This structure is stuffed into the struct file's private_data field
2118 * for directories. It is where we put information so that we can do
2119 * readdir operations in hash tree order.
2120 */
2121struct dir_private_info {
2122 struct rb_root root;
2123 struct rb_node *curr_node;
2124 struct fname *extra_fname;
2125 loff_t last_pos;
2126 __u32 curr_hash;
2127 __u32 curr_minor_hash;
2128 __u32 next_hash;
2129};
2130
2131/* calculate the first block number of the group */
2132static inline ext4_fsblk_t
2133ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2134{
2135 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2136 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2137}
2138
2139/*
2140 * Special error return code only used by dx_probe() and its callers.
2141 */
2142#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2143
2144/* htree levels for ext4 */
2145#define EXT4_HTREE_LEVEL_COMPAT 2
2146#define EXT4_HTREE_LEVEL 3
2147
2148static inline int ext4_dir_htree_level(struct super_block *sb)
2149{
2150 return ext4_has_feature_largedir(sb) ?
2151 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2152}
2153
2154/*
2155 * Timeout and state flag for lazy initialization inode thread.
2156 */
2157#define EXT4_DEF_LI_WAIT_MULT 10
2158#define EXT4_DEF_LI_MAX_START_DELAY 5
2159#define EXT4_LAZYINIT_QUIT 0x0001
2160#define EXT4_LAZYINIT_RUNNING 0x0002
2161
2162/*
2163 * Lazy inode table initialization info
2164 */
2165struct ext4_lazy_init {
2166 unsigned long li_state;
2167 struct list_head li_request_list;
2168 struct mutex li_list_mtx;
2169};
2170
2171struct ext4_li_request {
2172 struct super_block *lr_super;
2173 struct ext4_sb_info *lr_sbi;
2174 ext4_group_t lr_next_group;
2175 struct list_head lr_request;
2176 unsigned long lr_next_sched;
2177 unsigned long lr_timeout;
2178};
2179
2180struct ext4_features {
2181 struct kobject f_kobj;
2182 struct completion f_kobj_unregister;
2183};
2184
2185/*
2186 * This structure will be used for multiple mount protection. It will be
2187 * written into the block number saved in the s_mmp_block field in the
2188 * superblock. Programs that check MMP should assume that if
2189 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2190 * to use the filesystem, regardless of how old the timestamp is.
2191 */
2192#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2193#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2194#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2195#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2196
2197struct mmp_struct {
2198 __le32 mmp_magic; /* Magic number for MMP */
2199 __le32 mmp_seq; /* Sequence no. updated periodically */
2200
2201 /*
2202 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2203 * purposes and do not affect the correctness of the algorithm
2204 */
2205 __le64 mmp_time; /* Time last updated */
2206 char mmp_nodename[64]; /* Node which last updated MMP block */
2207 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2208
2209 /*
2210 * mmp_check_interval is used to verify if the MMP block has been
2211 * updated on the block device. The value is updated based on the
2212 * maximum time to write the MMP block during an update cycle.
2213 */
2214 __le16 mmp_check_interval;
2215
2216 __le16 mmp_pad1;
2217 __le32 mmp_pad2[226];
2218 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2219};
2220
2221/* arguments passed to the mmp thread */
2222struct mmpd_data {
2223 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2224 struct super_block *sb; /* super block of the fs */
2225};
2226
2227/*
2228 * Check interval multiplier
2229 * The MMP block is written every update interval and initially checked every
2230 * update interval x the multiplier (the value is then adapted based on the
2231 * write latency). The reason is that writes can be delayed under load and we
2232 * don't want readers to incorrectly assume that the filesystem is no longer
2233 * in use.
2234 */
2235#define EXT4_MMP_CHECK_MULT 2UL
2236
2237/*
2238 * Minimum interval for MMP checking in seconds.
2239 */
2240#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2241
2242/*
2243 * Maximum interval for MMP checking in seconds.
2244 */
2245#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2246
2247/*
2248 * Function prototypes
2249 */
2250
2251/*
2252 * Ok, these declarations are also in <linux/kernel.h> but none of the
2253 * ext4 source programs needs to include it so they are duplicated here.
2254 */
2255# define NORET_TYPE /**/
2256# define ATTRIB_NORET __attribute__((noreturn))
2257# define NORET_AND noreturn,
2258
2259/* bitmap.c */
2260extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2261void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2262 struct ext4_group_desc *gdp,
2263 struct buffer_head *bh, int sz);
2264int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2265 struct ext4_group_desc *gdp,
2266 struct buffer_head *bh, int sz);
2267void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2268 struct ext4_group_desc *gdp,
2269 struct buffer_head *bh);
2270int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2271 struct ext4_group_desc *gdp,
2272 struct buffer_head *bh);
2273
2274/* balloc.c */
2275extern void ext4_get_group_no_and_offset(struct super_block *sb,
2276 ext4_fsblk_t blocknr,
2277 ext4_group_t *blockgrpp,
2278 ext4_grpblk_t *offsetp);
2279extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2280 ext4_fsblk_t block);
2281
2282extern unsigned int ext4_block_group(struct super_block *sb,
2283 ext4_fsblk_t blocknr);
2284extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2285 ext4_fsblk_t blocknr);
2286extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2287extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2288 ext4_group_t group);
2289extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2290 ext4_fsblk_t goal,
2291 unsigned int flags,
2292 unsigned long *count,
2293 int *errp);
2294extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2295 s64 nclusters, unsigned int flags);
2296extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2297extern void ext4_check_blocks_bitmap(struct super_block *);
2298extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2299 ext4_group_t block_group,
2300 struct buffer_head ** bh);
2301extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2302
2303extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2304 ext4_group_t block_group);
2305extern int ext4_wait_block_bitmap(struct super_block *sb,
2306 ext4_group_t block_group,
2307 struct buffer_head *bh);
2308extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2309 ext4_group_t block_group);
2310extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2311 ext4_group_t block_group,
2312 struct ext4_group_desc *gdp);
2313ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2314
2315static inline bool ext4_encrypted_inode(struct inode *inode)
2316{
2317 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2318}
2319
2320#ifdef CONFIG_EXT4_FS_ENCRYPTION
2321static inline int ext4_fname_setup_filename(struct inode *dir,
2322 const struct qstr *iname,
2323 int lookup, struct ext4_filename *fname)
2324{
2325 struct fscrypt_name name;
2326 int err;
2327
2328 memset(fname, 0, sizeof(struct ext4_filename));
2329
2330 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2331
2332 fname->usr_fname = name.usr_fname;
2333 fname->disk_name = name.disk_name;
2334 fname->hinfo.hash = name.hash;
2335 fname->hinfo.minor_hash = name.minor_hash;
2336 fname->crypto_buf = name.crypto_buf;
2337 return err;
2338}
2339
2340static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2341{
2342 struct fscrypt_name name;
2343
2344 name.crypto_buf = fname->crypto_buf;
2345 fscrypt_free_filename(&name);
2346
2347 fname->crypto_buf.name = NULL;
2348 fname->usr_fname = NULL;
2349 fname->disk_name.name = NULL;
2350}
2351#else
2352static inline int ext4_fname_setup_filename(struct inode *dir,
2353 const struct qstr *iname,
2354 int lookup, struct ext4_filename *fname)
2355{
2356 fname->usr_fname = iname;
2357 fname->disk_name.name = (unsigned char *) iname->name;
2358 fname->disk_name.len = iname->len;
2359 return 0;
2360}
2361static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2362
2363#endif
2364
2365/* dir.c */
2366extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2367 struct file *,
2368 struct ext4_dir_entry_2 *,
2369 struct buffer_head *, char *, int,
2370 unsigned int);
2371#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2372 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2373 (de), (bh), (buf), (size), (offset)))
2374extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2375 __u32 minor_hash,
2376 struct ext4_dir_entry_2 *dirent,
2377 struct fscrypt_str *ent_name);
2378extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2379extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2380 struct buffer_head *bh,
2381 void *buf, int buf_size,
2382 struct ext4_filename *fname,
2383 struct ext4_dir_entry_2 **dest_de);
2384void ext4_insert_dentry(struct inode *inode,
2385 struct ext4_dir_entry_2 *de,
2386 int buf_size,
2387 struct ext4_filename *fname);
2388static inline void ext4_update_dx_flag(struct inode *inode)
2389{
2390 if (!ext4_has_feature_dir_index(inode->i_sb))
2391 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2392}
2393static const unsigned char ext4_filetype_table[] = {
2394 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2395};
2396
2397static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2398{
2399 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2400 return DT_UNKNOWN;
2401
2402 return ext4_filetype_table[filetype];
2403}
2404extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2405 void *buf, int buf_size);
2406
2407/* fsync.c */
2408extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2409
2410/* hash.c */
2411extern int ext4fs_dirhash(const char *name, int len, struct
2412 dx_hash_info *hinfo);
2413
2414/* ialloc.c */
2415extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2416 const struct qstr *qstr, __u32 goal,
2417 uid_t *owner, __u32 i_flags,
2418 int handle_type, unsigned int line_no,
2419 int nblocks);
2420
2421#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2422 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2423 i_flags, 0, 0, 0)
2424#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2425 type, nblocks) \
2426 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2427 0, (type), __LINE__, (nblocks))
2428
2429
2430extern void ext4_free_inode(handle_t *, struct inode *);
2431extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2432extern unsigned long ext4_count_free_inodes(struct super_block *);
2433extern unsigned long ext4_count_dirs(struct super_block *);
2434extern void ext4_check_inodes_bitmap(struct super_block *);
2435extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2436extern int ext4_init_inode_table(struct super_block *sb,
2437 ext4_group_t group, int barrier);
2438extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2439
2440/* mballoc.c */
2441extern const struct file_operations ext4_seq_mb_groups_fops;
2442extern long ext4_mb_stats;
2443extern long ext4_mb_max_to_scan;
2444extern int ext4_mb_init(struct super_block *);
2445extern int ext4_mb_release(struct super_block *);
2446extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2447 struct ext4_allocation_request *, int *);
2448extern int ext4_mb_reserve_blocks(struct super_block *, int);
2449extern void ext4_discard_preallocations(struct inode *);
2450extern int __init ext4_init_mballoc(void);
2451extern void ext4_exit_mballoc(void);
2452extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2453 struct buffer_head *bh, ext4_fsblk_t block,
2454 unsigned long count, int flags);
2455extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2456 ext4_group_t ngroups);
2457extern int ext4_mb_add_groupinfo(struct super_block *sb,
2458 ext4_group_t i, struct ext4_group_desc *desc);
2459extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2460 ext4_fsblk_t block, unsigned long count);
2461extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2462extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2463
2464/* inode.c */
2465int ext4_inode_is_fast_symlink(struct inode *inode);
2466struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2467struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2468int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2469 bool wait, struct buffer_head **bhs);
2470int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2471 struct buffer_head *bh_result, int create);
2472int ext4_get_block(struct inode *inode, sector_t iblock,
2473 struct buffer_head *bh_result, int create);
2474int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2475 struct buffer_head *bh_result, int create);
2476int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2477 struct buffer_head *bh, int create);
2478int ext4_walk_page_buffers(handle_t *handle,
2479 struct buffer_head *head,
2480 unsigned from,
2481 unsigned to,
2482 int *partial,
2483 int (*fn)(handle_t *handle,
2484 struct buffer_head *bh));
2485int do_journal_get_write_access(handle_t *handle,
2486 struct buffer_head *bh);
2487#define FALL_BACK_TO_NONDELALLOC 1
2488#define CONVERT_INLINE_DATA 2
2489
2490extern struct inode *ext4_iget(struct super_block *, unsigned long);
2491extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2492extern int ext4_write_inode(struct inode *, struct writeback_control *);
2493extern int ext4_setattr(struct dentry *, struct iattr *);
2494extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2495extern void ext4_evict_inode(struct inode *);
2496extern void ext4_clear_inode(struct inode *);
2497extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2498extern int ext4_sync_inode(handle_t *, struct inode *);
2499extern void ext4_dirty_inode(struct inode *, int);
2500extern int ext4_change_inode_journal_flag(struct inode *, int);
2501extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2502extern int ext4_inode_attach_jinode(struct inode *inode);
2503extern int ext4_can_truncate(struct inode *inode);
2504extern int ext4_truncate(struct inode *);
2505extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2506extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2507extern void ext4_set_inode_flags(struct inode *);
2508extern int ext4_alloc_da_blocks(struct inode *inode);
2509extern void ext4_set_aops(struct inode *inode);
2510extern int ext4_writepage_trans_blocks(struct inode *);
2511extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2512extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2513 loff_t lstart, loff_t lend);
2514extern int ext4_page_mkwrite(struct vm_fault *vmf);
2515extern int ext4_filemap_fault(struct vm_fault *vmf);
2516extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2517extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2518extern void ext4_da_update_reserve_space(struct inode *inode,
2519 int used, int quota_claim);
2520extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2521 ext4_fsblk_t pblk, ext4_lblk_t len);
2522extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2523 unsigned int map_len,
2524 struct extent_status *result);
2525
2526/* indirect.c */
2527extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2528 struct ext4_map_blocks *map, int flags);
2529extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2530extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2531extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2532extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2533 ext4_lblk_t start, ext4_lblk_t end);
2534
2535/* ioctl.c */
2536extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2537extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2538
2539/* migrate.c */
2540extern int ext4_ext_migrate(struct inode *);
2541extern int ext4_ind_migrate(struct inode *inode);
2542
2543/* namei.c */
2544extern int ext4_dirent_csum_verify(struct inode *inode,
2545 struct ext4_dir_entry *dirent);
2546extern int ext4_orphan_add(handle_t *, struct inode *);
2547extern int ext4_orphan_del(handle_t *, struct inode *);
2548extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2549 __u32 start_minor_hash, __u32 *next_hash);
2550extern int ext4_search_dir(struct buffer_head *bh,
2551 char *search_buf,
2552 int buf_size,
2553 struct inode *dir,
2554 struct ext4_filename *fname,
2555 unsigned int offset,
2556 struct ext4_dir_entry_2 **res_dir);
2557extern int ext4_generic_delete_entry(handle_t *handle,
2558 struct inode *dir,
2559 struct ext4_dir_entry_2 *de_del,
2560 struct buffer_head *bh,
2561 void *entry_buf,
2562 int buf_size,
2563 int csum_size);
2564extern bool ext4_empty_dir(struct inode *inode);
2565
2566/* resize.c */
2567extern int ext4_group_add(struct super_block *sb,
2568 struct ext4_new_group_data *input);
2569extern int ext4_group_extend(struct super_block *sb,
2570 struct ext4_super_block *es,
2571 ext4_fsblk_t n_blocks_count);
2572extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2573
2574/* super.c */
2575extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2576 sector_t block, int op_flags);
2577extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2578extern int ext4_calculate_overhead(struct super_block *sb);
2579extern void ext4_superblock_csum_set(struct super_block *sb);
2580extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2581extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2582extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2583 ext4_group_t ngroup);
2584extern const char *ext4_decode_error(struct super_block *sb, int errno,
2585 char nbuf[16]);
2586
2587extern __printf(4, 5)
2588void __ext4_error(struct super_block *, const char *, unsigned int,
2589 const char *, ...);
2590extern __printf(5, 6)
2591void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2592 const char *, ...);
2593extern __printf(5, 6)
2594void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2595 const char *, ...);
2596extern void __ext4_std_error(struct super_block *, const char *,
2597 unsigned int, int);
2598extern __printf(4, 5)
2599void __ext4_abort(struct super_block *, const char *, unsigned int,
2600 const char *, ...);
2601extern __printf(4, 5)
2602void __ext4_warning(struct super_block *, const char *, unsigned int,
2603 const char *, ...);
2604extern __printf(4, 5)
2605void __ext4_warning_inode(const struct inode *inode, const char *function,
2606 unsigned int line, const char *fmt, ...);
2607extern __printf(3, 4)
2608void __ext4_msg(struct super_block *, const char *, const char *, ...);
2609extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2610 const char *, unsigned int, const char *);
2611extern __printf(7, 8)
2612void __ext4_grp_locked_error(const char *, unsigned int,
2613 struct super_block *, ext4_group_t,
2614 unsigned long, ext4_fsblk_t,
2615 const char *, ...);
2616
2617#define EXT4_ERROR_INODE(inode, fmt, a...) \
2618 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2619
2620#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2621 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2622
2623#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2624 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2625
2626#ifdef CONFIG_PRINTK
2627
2628#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2629 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2630#define ext4_error_file(file, func, line, block, fmt, ...) \
2631 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2632#define ext4_error(sb, fmt, ...) \
2633 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2634#define ext4_abort(sb, fmt, ...) \
2635 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2636#define ext4_warning(sb, fmt, ...) \
2637 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2638#define ext4_warning_inode(inode, fmt, ...) \
2639 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2640#define ext4_msg(sb, level, fmt, ...) \
2641 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2642#define dump_mmp_msg(sb, mmp, msg) \
2643 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2644#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2645 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2646 fmt, ##__VA_ARGS__)
2647
2648#else
2649
2650#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2651do { \
2652 no_printk(fmt, ##__VA_ARGS__); \
2653 __ext4_error_inode(inode, "", 0, block, " "); \
2654} while (0)
2655#define ext4_error_file(file, func, line, block, fmt, ...) \
2656do { \
2657 no_printk(fmt, ##__VA_ARGS__); \
2658 __ext4_error_file(file, "", 0, block, " "); \
2659} while (0)
2660#define ext4_error(sb, fmt, ...) \
2661do { \
2662 no_printk(fmt, ##__VA_ARGS__); \
2663 __ext4_error(sb, "", 0, " "); \
2664} while (0)
2665#define ext4_abort(sb, fmt, ...) \
2666do { \
2667 no_printk(fmt, ##__VA_ARGS__); \
2668 __ext4_abort(sb, "", 0, " "); \
2669} while (0)
2670#define ext4_warning(sb, fmt, ...) \
2671do { \
2672 no_printk(fmt, ##__VA_ARGS__); \
2673 __ext4_warning(sb, "", 0, " "); \
2674} while (0)
2675#define ext4_warning_inode(inode, fmt, ...) \
2676do { \
2677 no_printk(fmt, ##__VA_ARGS__); \
2678 __ext4_warning_inode(inode, "", 0, " "); \
2679} while (0)
2680#define ext4_msg(sb, level, fmt, ...) \
2681do { \
2682 no_printk(fmt, ##__VA_ARGS__); \
2683 __ext4_msg(sb, "", " "); \
2684} while (0)
2685#define dump_mmp_msg(sb, mmp, msg) \
2686 __dump_mmp_msg(sb, mmp, "", 0, "")
2687#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2688do { \
2689 no_printk(fmt, ##__VA_ARGS__); \
2690 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2691} while (0)
2692
2693#endif
2694
2695extern void ext4_update_dynamic_rev(struct super_block *sb);
2696extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2697 __u32 compat);
2698extern int ext4_update_rocompat_feature(handle_t *handle,
2699 struct super_block *sb, __u32 rocompat);
2700extern int ext4_update_incompat_feature(handle_t *handle,
2701 struct super_block *sb, __u32 incompat);
2702extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2703 struct ext4_group_desc *bg);
2704extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2705 struct ext4_group_desc *bg);
2706extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2707 struct ext4_group_desc *bg);
2708extern __u32 ext4_free_group_clusters(struct super_block *sb,
2709 struct ext4_group_desc *bg);
2710extern __u32 ext4_free_inodes_count(struct super_block *sb,
2711 struct ext4_group_desc *bg);
2712extern __u32 ext4_used_dirs_count(struct super_block *sb,
2713 struct ext4_group_desc *bg);
2714extern __u32 ext4_itable_unused_count(struct super_block *sb,
2715 struct ext4_group_desc *bg);
2716extern void ext4_block_bitmap_set(struct super_block *sb,
2717 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2718extern void ext4_inode_bitmap_set(struct super_block *sb,
2719 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2720extern void ext4_inode_table_set(struct super_block *sb,
2721 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2722extern void ext4_free_group_clusters_set(struct super_block *sb,
2723 struct ext4_group_desc *bg,
2724 __u32 count);
2725extern void ext4_free_inodes_set(struct super_block *sb,
2726 struct ext4_group_desc *bg, __u32 count);
2727extern void ext4_used_dirs_set(struct super_block *sb,
2728 struct ext4_group_desc *bg, __u32 count);
2729extern void ext4_itable_unused_set(struct super_block *sb,
2730 struct ext4_group_desc *bg, __u32 count);
2731extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2732 struct ext4_group_desc *gdp);
2733extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2734 struct ext4_group_desc *gdp);
2735extern int ext4_register_li_request(struct super_block *sb,
2736 ext4_group_t first_not_zeroed);
2737
2738static inline int ext4_has_metadata_csum(struct super_block *sb)
2739{
2740 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2741 !EXT4_SB(sb)->s_chksum_driver);
2742
2743 return ext4_has_feature_metadata_csum(sb) &&
2744 (EXT4_SB(sb)->s_chksum_driver != NULL);
2745}
2746
2747static inline int ext4_has_group_desc_csum(struct super_block *sb)
2748{
2749 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2750}
2751
2752static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2753{
2754 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2755 le32_to_cpu(es->s_blocks_count_lo);
2756}
2757
2758static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2759{
2760 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2761 le32_to_cpu(es->s_r_blocks_count_lo);
2762}
2763
2764static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2765{
2766 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2767 le32_to_cpu(es->s_free_blocks_count_lo);
2768}
2769
2770static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2771 ext4_fsblk_t blk)
2772{
2773 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2774 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2775}
2776
2777static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2778 ext4_fsblk_t blk)
2779{
2780 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2781 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2782}
2783
2784static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2785 ext4_fsblk_t blk)
2786{
2787 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2788 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2789}
2790
2791static inline loff_t ext4_isize(struct super_block *sb,
2792 struct ext4_inode *raw_inode)
2793{
2794 if (ext4_has_feature_largedir(sb) ||
2795 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2796 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2797 le32_to_cpu(raw_inode->i_size_lo);
2798
2799 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2800}
2801
2802static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2803{
2804 raw_inode->i_size_lo = cpu_to_le32(i_size);
2805 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2806}
2807
2808static inline
2809struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2810 ext4_group_t group)
2811{
2812 struct ext4_group_info ***grp_info;
2813 long indexv, indexh;
2814 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2815 grp_info = EXT4_SB(sb)->s_group_info;
2816 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2817 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2818 return grp_info[indexv][indexh];
2819}
2820
2821/*
2822 * Reading s_groups_count requires using smp_rmb() afterwards. See
2823 * the locking protocol documented in the comments of ext4_group_add()
2824 * in resize.c
2825 */
2826static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2827{
2828 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2829
2830 smp_rmb();
2831 return ngroups;
2832}
2833
2834static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2835 ext4_group_t block_group)
2836{
2837 return block_group >> sbi->s_log_groups_per_flex;
2838}
2839
2840static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2841{
2842 return 1 << sbi->s_log_groups_per_flex;
2843}
2844
2845#define ext4_std_error(sb, errno) \
2846do { \
2847 if ((errno)) \
2848 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2849} while (0)
2850
2851#ifdef CONFIG_SMP
2852/* Each CPU can accumulate percpu_counter_batch clusters in their local
2853 * counters. So we need to make sure we have free clusters more
2854 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2855 */
2856#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2857#else
2858#define EXT4_FREECLUSTERS_WATERMARK 0
2859#endif
2860
2861/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2862static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2863{
2864 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2865 !inode_is_locked(inode));
2866 down_write(&EXT4_I(inode)->i_data_sem);
2867 if (newsize > EXT4_I(inode)->i_disksize)
2868 EXT4_I(inode)->i_disksize = newsize;
2869 up_write(&EXT4_I(inode)->i_data_sem);
2870}
2871
2872/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2873static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2874{
2875 int changed = 0;
2876
2877 if (newsize > inode->i_size) {
2878 i_size_write(inode, newsize);
2879 changed = 1;
2880 }
2881 if (newsize > EXT4_I(inode)->i_disksize) {
2882 ext4_update_i_disksize(inode, newsize);
2883 changed |= 2;
2884 }
2885 return changed;
2886}
2887
2888int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2889 loff_t len);
2890
2891struct ext4_group_info {
2892 unsigned long bb_state;
2893 struct rb_root bb_free_root;
2894 ext4_grpblk_t bb_first_free; /* first free block */
2895 ext4_grpblk_t bb_free; /* total free blocks */
2896 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2897 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2898 struct list_head bb_prealloc_list;
2899#ifdef DOUBLE_CHECK
2900 void *bb_bitmap;
2901#endif
2902 struct rw_semaphore alloc_sem;
2903 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2904 * regions, index is order.
2905 * bb_counters[3] = 5 means
2906 * 5 free 8-block regions. */
2907};
2908
2909#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2910#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2911#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2912#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2913
2914#define EXT4_MB_GRP_NEED_INIT(grp) \
2915 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2916#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2917 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2918#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2919 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2920
2921#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2922 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2923#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2924 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2925#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2926 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2927
2928#define EXT4_MAX_CONTENTION 8
2929#define EXT4_CONTENTION_THRESHOLD 2
2930
2931static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2932 ext4_group_t group)
2933{
2934 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2935}
2936
2937/*
2938 * Returns true if the filesystem is busy enough that attempts to
2939 * access the block group locks has run into contention.
2940 */
2941static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2942{
2943 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2944}
2945
2946static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2947{
2948 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2949 if (spin_trylock(lock))
2950 /*
2951 * We're able to grab the lock right away, so drop the
2952 * lock contention counter.
2953 */
2954 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2955 else {
2956 /*
2957 * The lock is busy, so bump the contention counter,
2958 * and then wait on the spin lock.
2959 */
2960 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2961 EXT4_MAX_CONTENTION);
2962 spin_lock(lock);
2963 }
2964}
2965
2966static inline void ext4_unlock_group(struct super_block *sb,
2967 ext4_group_t group)
2968{
2969 spin_unlock(ext4_group_lock_ptr(sb, group));
2970}
2971
2972/*
2973 * Block validity checking
2974 */
2975#define ext4_check_indirect_blockref(inode, bh) \
2976 ext4_check_blockref(__func__, __LINE__, inode, \
2977 (__le32 *)(bh)->b_data, \
2978 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2979
2980#define ext4_ind_check_inode(inode) \
2981 ext4_check_blockref(__func__, __LINE__, inode, \
2982 EXT4_I(inode)->i_data, \
2983 EXT4_NDIR_BLOCKS)
2984
2985/*
2986 * Inodes and files operations
2987 */
2988
2989/* dir.c */
2990extern const struct file_operations ext4_dir_operations;
2991
2992/* file.c */
2993extern const struct inode_operations ext4_file_inode_operations;
2994extern const struct file_operations ext4_file_operations;
2995extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2996
2997/* inline.c */
2998extern int ext4_get_max_inline_size(struct inode *inode);
2999extern int ext4_find_inline_data_nolock(struct inode *inode);
3000extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3001 unsigned int len);
3002extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3003
3004extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3005extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3006 struct inode *inode,
3007 loff_t pos, unsigned len,
3008 unsigned flags,
3009 struct page **pagep);
3010extern int ext4_write_inline_data_end(struct inode *inode,
3011 loff_t pos, unsigned len,
3012 unsigned copied,
3013 struct page *page);
3014extern struct buffer_head *
3015ext4_journalled_write_inline_data(struct inode *inode,
3016 unsigned len,
3017 struct page *page);
3018extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3019 struct inode *inode,
3020 loff_t pos, unsigned len,
3021 unsigned flags,
3022 struct page **pagep,
3023 void **fsdata);
3024extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3025 unsigned len, unsigned copied,
3026 struct page *page);
3027extern int ext4_try_add_inline_entry(handle_t *handle,
3028 struct ext4_filename *fname,
3029 struct inode *dir, struct inode *inode);
3030extern int ext4_try_create_inline_dir(handle_t *handle,
3031 struct inode *parent,
3032 struct inode *inode);
3033extern int ext4_read_inline_dir(struct file *filp,
3034 struct dir_context *ctx,
3035 int *has_inline_data);
3036extern int htree_inlinedir_to_tree(struct file *dir_file,
3037 struct inode *dir, ext4_lblk_t block,
3038 struct dx_hash_info *hinfo,
3039 __u32 start_hash, __u32 start_minor_hash,
3040 int *has_inline_data);
3041extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3042 struct ext4_filename *fname,
3043 struct ext4_dir_entry_2 **res_dir,
3044 int *has_inline_data);
3045extern int ext4_delete_inline_entry(handle_t *handle,
3046 struct inode *dir,
3047 struct ext4_dir_entry_2 *de_del,
3048 struct buffer_head *bh,
3049 int *has_inline_data);
3050extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3051extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3052 struct ext4_dir_entry_2 **parent_de,
3053 int *retval);
3054extern int ext4_inline_data_fiemap(struct inode *inode,
3055 struct fiemap_extent_info *fieinfo,
3056 int *has_inline, __u64 start, __u64 len);
3057extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3058
3059extern int ext4_convert_inline_data(struct inode *inode);
3060
3061static inline int ext4_has_inline_data(struct inode *inode)
3062{
3063 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3064 EXT4_I(inode)->i_inline_off;
3065}
3066
3067/* namei.c */
3068extern const struct inode_operations ext4_dir_inode_operations;
3069extern const struct inode_operations ext4_special_inode_operations;
3070extern struct dentry *ext4_get_parent(struct dentry *child);
3071extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3072 struct ext4_dir_entry_2 *de,
3073 int blocksize, int csum_size,
3074 unsigned int parent_ino, int dotdot_real_len);
3075extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3076 unsigned int blocksize);
3077extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3078 struct inode *inode,
3079 struct buffer_head *bh);
3080#define S_SHIFT 12
3081static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3082 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3083 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3084 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3085 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3086 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3087 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3088 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3089};
3090
3091static inline void ext4_set_de_type(struct super_block *sb,
3092 struct ext4_dir_entry_2 *de,
3093 umode_t mode) {
3094 if (ext4_has_feature_filetype(sb))
3095 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3096}
3097
3098/* readpages.c */
3099extern int ext4_mpage_readpages(struct address_space *mapping,
3100 struct list_head *pages, struct page *page,
3101 unsigned nr_pages);
3102
3103/* symlink.c */
3104extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3105extern const struct inode_operations ext4_symlink_inode_operations;
3106extern const struct inode_operations ext4_fast_symlink_inode_operations;
3107
3108/* sysfs.c */
3109extern int ext4_register_sysfs(struct super_block *sb);
3110extern void ext4_unregister_sysfs(struct super_block *sb);
3111extern int __init ext4_init_sysfs(void);
3112extern void ext4_exit_sysfs(void);
3113
3114/* block_validity */
3115extern void ext4_release_system_zone(struct super_block *sb);
3116extern int ext4_setup_system_zone(struct super_block *sb);
3117extern int __init ext4_init_system_zone(void);
3118extern void ext4_exit_system_zone(void);
3119extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3120 ext4_fsblk_t start_blk,
3121 unsigned int count);
3122extern int ext4_check_blockref(const char *, unsigned int,
3123 struct inode *, __le32 *, unsigned int);
3124
3125/* extents.c */
3126struct ext4_ext_path;
3127struct ext4_extent;
3128
3129/*
3130 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3131 * __le32.
3132 */
3133#define EXT_MAX_BLOCKS 0xffffffff
3134
3135extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3136extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3137extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3138extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3139 struct ext4_map_blocks *map, int flags);
3140extern int ext4_ext_truncate(handle_t *, struct inode *);
3141extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3142 ext4_lblk_t end);
3143extern void ext4_ext_init(struct super_block *);
3144extern void ext4_ext_release(struct super_block *);
3145extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3146 loff_t len);
3147extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3148 loff_t offset, ssize_t len);
3149extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3150 struct ext4_map_blocks *map, int flags);
3151extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3152 ext4_lblk_t lblocks);
3153extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3154 int num,
3155 struct ext4_ext_path *path);
3156extern int ext4_can_extents_be_merged(struct inode *inode,
3157 struct ext4_extent *ex1,
3158 struct ext4_extent *ex2);
3159extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3160 struct ext4_ext_path **,
3161 struct ext4_extent *, int);
3162extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3163 struct ext4_ext_path **,
3164 int flags);
3165extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3166extern int ext4_ext_check_inode(struct inode *inode);
3167extern int ext4_find_delalloc_range(struct inode *inode,
3168 ext4_lblk_t lblk_start,
3169 ext4_lblk_t lblk_end);
3170extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3171extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3172extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3173 __u64 start, __u64 len);
3174extern int ext4_ext_precache(struct inode *inode);
3175extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3176extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3177extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3178 struct inode *inode2, ext4_lblk_t lblk1,
3179 ext4_lblk_t lblk2, ext4_lblk_t count,
3180 int mark_unwritten,int *err);
3181
3182/* move_extent.c */
3183extern void ext4_double_down_write_data_sem(struct inode *first,
3184 struct inode *second);
3185extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3186 struct inode *donor_inode);
3187extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3188 __u64 start_orig, __u64 start_donor,
3189 __u64 len, __u64 *moved_len);
3190
3191/* page-io.c */
3192extern int __init ext4_init_pageio(void);
3193extern void ext4_exit_pageio(void);
3194extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3195extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3196extern int ext4_put_io_end(ext4_io_end_t *io_end);
3197extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3198extern void ext4_io_submit_init(struct ext4_io_submit *io,
3199 struct writeback_control *wbc);
3200extern void ext4_end_io_rsv_work(struct work_struct *work);
3201extern void ext4_io_submit(struct ext4_io_submit *io);
3202extern int ext4_bio_write_page(struct ext4_io_submit *io,
3203 struct page *page,
3204 int len,
3205 struct writeback_control *wbc,
3206 bool keep_towrite);
3207
3208/* mmp.c */
3209extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3210
3211/*
3212 * Add new method to test whether block and inode bitmaps are properly
3213 * initialized. With uninit_bg reading the block from disk is not enough
3214 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3215 */
3216#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3217
3218static inline int bitmap_uptodate(struct buffer_head *bh)
3219{
3220 return (buffer_uptodate(bh) &&
3221 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3222}
3223static inline void set_bitmap_uptodate(struct buffer_head *bh)
3224{
3225 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3226}
3227
3228/*
3229 * Disable DIO read nolock optimization, so new dioreaders will be forced
3230 * to grab i_mutex
3231 */
3232static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3233{
3234 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3235 smp_mb();
3236}
3237static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3238{
3239 smp_mb();
3240 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3241}
3242
3243#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3244
3245/* For ioend & aio unwritten conversion wait queues */
3246#define EXT4_WQ_HASH_SZ 37
3247#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3248 EXT4_WQ_HASH_SZ])
3249extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3250
3251extern int ext4_resize_begin(struct super_block *sb);
3252extern void ext4_resize_end(struct super_block *sb);
3253
3254static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3255 struct ext4_io_end *io_end)
3256{
3257 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3258 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3259 atomic_inc(&EXT4_I(inode)->i_unwritten);
3260 }
3261}
3262
3263static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3264{
3265 struct inode *inode = io_end->inode;
3266
3267 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3268 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3269 /* Wake up anyone waiting on unwritten extent conversion */
3270 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3271 wake_up_all(ext4_ioend_wq(inode));
3272 }
3273}
3274
3275extern const struct iomap_ops ext4_iomap_ops;
3276
3277#endif /* __KERNEL__ */
3278
3279#define EFSBADCRC EBADMSG /* Bad CRC detected */
3280#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3281
3282#endif /* _EXT4_H */