blob: 48b558ed00b44b75272745226c8c5f2333e13f85 [file] [log] [blame]
rjw1f884582022-01-06 17:20:42 +08001/* 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
208struct ext4_io_submit {
209 struct writeback_control *io_wbc;
210 struct bio *io_bio;
211 ext4_io_end_t *io_end;
212 sector_t io_next_block;
213};
214
215/*
216 * Special inodes numbers
217 */
218#define EXT4_BAD_INO 1 /* Bad blocks inode */
219#define EXT4_ROOT_INO 2 /* Root inode */
220#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
221#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
222#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
223#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
224#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
225#define EXT4_JOURNAL_INO 8 /* Journal inode */
226
227/* First non-reserved inode for old ext4 filesystems */
228#define EXT4_GOOD_OLD_FIRST_INO 11
229
230/*
231 * Maximal count of links to a file
232 */
233#define EXT4_LINK_MAX 65000
234
235/*
236 * Macro-instructions used to manage several block sizes
237 */
238#define EXT4_MIN_BLOCK_SIZE 1024
239#define EXT4_MAX_BLOCK_SIZE 65536
240#define EXT4_MIN_BLOCK_LOG_SIZE 10
241#define EXT4_MAX_BLOCK_LOG_SIZE 16
242#define EXT4_MAX_CLUSTER_LOG_SIZE 30
243#ifdef __KERNEL__
244# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
245#else
246# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
247#endif
248#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
249#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
250 EXT4_SB(s)->s_cluster_bits)
251#ifdef __KERNEL__
252# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
253# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
254#else
255# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
256#endif
257#ifdef __KERNEL__
258#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
259#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
260#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
261#else
262#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
263 EXT4_GOOD_OLD_INODE_SIZE : \
264 (s)->s_inode_size)
265#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
266 EXT4_GOOD_OLD_FIRST_INO : \
267 (s)->s_first_ino)
268#endif
269#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
270#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
271 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
272 blkbits))
273
274/* Translate a block number to a cluster number */
275#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
276/* Translate a cluster number to a block number */
277#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
278/* Translate # of blks to # of clusters */
279#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
280 (sbi)->s_cluster_bits)
281/* Mask out the low bits to get the starting block of the cluster */
282#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
283 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
284#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
285 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
286/* Get the cluster offset */
287#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
288 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
289#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
290 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
291
292/*
293 * Structure of a blocks group descriptor
294 */
295struct ext4_group_desc
296{
297 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
298 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
299 __le32 bg_inode_table_lo; /* Inodes table block */
300 __le16 bg_free_blocks_count_lo;/* Free blocks count */
301 __le16 bg_free_inodes_count_lo;/* Free inodes count */
302 __le16 bg_used_dirs_count_lo; /* Directories count */
303 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
304 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
305 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
306 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
307 __le16 bg_itable_unused_lo; /* Unused inodes count */
308 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
309 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
310 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
311 __le32 bg_inode_table_hi; /* Inodes table block MSB */
312 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
313 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
314 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
315 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
316 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
317 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
318 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
319 __u32 bg_reserved;
320};
321
322#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
323 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
324 sizeof(__le16))
325#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
326 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
327 sizeof(__le16))
328
329/*
330 * Structure of a flex block group info
331 */
332
333struct flex_groups {
334 atomic64_t free_clusters;
335 atomic_t free_inodes;
336 atomic_t used_dirs;
337};
338
339#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
340#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
341#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
342
343/*
344 * Macro-instructions used to manage group descriptors
345 */
346#define EXT4_MIN_DESC_SIZE 32
347#define EXT4_MIN_DESC_SIZE_64BIT 64
348#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
349#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
350#ifdef __KERNEL__
351# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
352# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
353# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
354# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
355# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
356#else
357# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
358# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
359# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
360#endif
361
362/*
363 * Constants relative to the data blocks
364 */
365#define EXT4_NDIR_BLOCKS 12
366#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
367#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
368#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
369#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
370
371/*
372 * Inode flags
373 */
374#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
375#define EXT4_UNRM_FL 0x00000002 /* Undelete */
376#define EXT4_COMPR_FL 0x00000004 /* Compress file */
377#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
378#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
379#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
380#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
381#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
382/* Reserved for compression usage... */
383#define EXT4_DIRTY_FL 0x00000100
384#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
385#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
386 /* nb: was previously EXT2_ECOMPR_FL */
387#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
388/* End compression flags --- maybe not all used */
389#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
390#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
391#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
392#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
393#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
394#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
395#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
396#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
397#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
398#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
399#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
400#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
401#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
402
403#define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
404#define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
405
406/* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
407#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
408 EXT4_IMMUTABLE_FL | \
409 EXT4_APPEND_FL | \
410 EXT4_NODUMP_FL | \
411 EXT4_NOATIME_FL | \
412 EXT4_PROJINHERIT_FL)
413
414/* Flags that should be inherited by new inodes from their parent. */
415#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
416 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
417 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
418 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
419 EXT4_PROJINHERIT_FL)
420
421/* Flags that are appropriate for regular files (all but dir-specific ones). */
422#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
423
424/* Flags that are appropriate for non-directories/regular files. */
425#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
426
427/* The only flags that should be swapped */
428#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
429
430/* Mask out flags that are inappropriate for the given type of inode. */
431static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
432{
433 if (S_ISDIR(mode))
434 return flags;
435 else if (S_ISREG(mode))
436 return flags & EXT4_REG_FLMASK;
437 else
438 return flags & EXT4_OTHER_FLMASK;
439}
440
441/*
442 * Inode flags used for atomic set/get
443 */
444enum {
445 EXT4_INODE_SECRM = 0, /* Secure deletion */
446 EXT4_INODE_UNRM = 1, /* Undelete */
447 EXT4_INODE_COMPR = 2, /* Compress file */
448 EXT4_INODE_SYNC = 3, /* Synchronous updates */
449 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
450 EXT4_INODE_APPEND = 5, /* writes to file may only append */
451 EXT4_INODE_NODUMP = 6, /* do not dump file */
452 EXT4_INODE_NOATIME = 7, /* do not update atime */
453/* Reserved for compression usage... */
454 EXT4_INODE_DIRTY = 8,
455 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
456 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
457 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
458/* End compression flags --- maybe not all used */
459 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
460 EXT4_INODE_IMAGIC = 13, /* AFS directory */
461 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
462 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
463 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
464 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
465 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
466 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
467 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
468 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
469 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
470 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
471 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
472};
473
474/*
475 * Since it's pretty easy to mix up bit numbers and hex values, we use a
476 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
477 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
478 * any extra space in the compiled kernel image, otherwise, the build will fail.
479 * It's important that these values are the same, since we are using
480 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
481 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
482 * values found in ext2, ext3 and ext4 filesystems, and of course the values
483 * defined in e2fsprogs.
484 *
485 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
486 */
487#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
488#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
489
490static inline void ext4_check_flag_values(void)
491{
492 CHECK_FLAG_VALUE(SECRM);
493 CHECK_FLAG_VALUE(UNRM);
494 CHECK_FLAG_VALUE(COMPR);
495 CHECK_FLAG_VALUE(SYNC);
496 CHECK_FLAG_VALUE(IMMUTABLE);
497 CHECK_FLAG_VALUE(APPEND);
498 CHECK_FLAG_VALUE(NODUMP);
499 CHECK_FLAG_VALUE(NOATIME);
500 CHECK_FLAG_VALUE(DIRTY);
501 CHECK_FLAG_VALUE(COMPRBLK);
502 CHECK_FLAG_VALUE(NOCOMPR);
503 CHECK_FLAG_VALUE(ENCRYPT);
504 CHECK_FLAG_VALUE(INDEX);
505 CHECK_FLAG_VALUE(IMAGIC);
506 CHECK_FLAG_VALUE(JOURNAL_DATA);
507 CHECK_FLAG_VALUE(NOTAIL);
508 CHECK_FLAG_VALUE(DIRSYNC);
509 CHECK_FLAG_VALUE(TOPDIR);
510 CHECK_FLAG_VALUE(HUGE_FILE);
511 CHECK_FLAG_VALUE(EXTENTS);
512 CHECK_FLAG_VALUE(EA_INODE);
513 CHECK_FLAG_VALUE(EOFBLOCKS);
514 CHECK_FLAG_VALUE(INLINE_DATA);
515 CHECK_FLAG_VALUE(PROJINHERIT);
516 CHECK_FLAG_VALUE(RESERVED);
517}
518
519/* Used to pass group descriptor data when online resize is done */
520struct ext4_new_group_input {
521 __u32 group; /* Group number for this data */
522 __u64 block_bitmap; /* Absolute block number of block bitmap */
523 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
524 __u64 inode_table; /* Absolute block number of inode table start */
525 __u32 blocks_count; /* Total number of blocks in this group */
526 __u16 reserved_blocks; /* Number of reserved blocks in this group */
527 __u16 unused;
528};
529
530#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
531struct compat_ext4_new_group_input {
532 u32 group;
533 compat_u64 block_bitmap;
534 compat_u64 inode_bitmap;
535 compat_u64 inode_table;
536 u32 blocks_count;
537 u16 reserved_blocks;
538 u16 unused;
539};
540#endif
541
542/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
543struct ext4_new_group_data {
544 __u32 group;
545 __u64 block_bitmap;
546 __u64 inode_bitmap;
547 __u64 inode_table;
548 __u32 blocks_count;
549 __u16 reserved_blocks;
550 __u16 unused;
551 __u32 free_blocks_count;
552};
553
554/* Indexes used to index group tables in ext4_new_group_data */
555enum {
556 BLOCK_BITMAP = 0, /* block bitmap */
557 INODE_BITMAP, /* inode bitmap */
558 INODE_TABLE, /* inode tables */
559 GROUP_TABLE_COUNT,
560};
561
562/*
563 * Flags used by ext4_map_blocks()
564 */
565 /* Allocate any needed blocks and/or convert an unwritten
566 extent to be an initialized ext4 */
567#define EXT4_GET_BLOCKS_CREATE 0x0001
568 /* Request the creation of an unwritten extent */
569#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
570#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
571 EXT4_GET_BLOCKS_CREATE)
572 /* Caller is from the delayed allocation writeout path
573 * finally doing the actual allocation of delayed blocks */
574#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
575 /* caller is from the direct IO path, request to creation of an
576 unwritten extents if not allocated, split the unwritten
577 extent if blocks has been preallocated already*/
578#define EXT4_GET_BLOCKS_PRE_IO 0x0008
579#define EXT4_GET_BLOCKS_CONVERT 0x0010
580#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
581 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
582 /* Convert extent to initialized after IO complete */
583#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
584 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
585 /* Eventual metadata allocation (due to growing extent tree)
586 * should not fail, so try to use reserved blocks for that.*/
587#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
588 /* Don't normalize allocation size (used for fallocate) */
589#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
590 /* Request will not result in inode size update (user for fallocate) */
591#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
592 /* Convert written extents to unwritten */
593#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
594 /* Write zeros to newly created written extents */
595#define EXT4_GET_BLOCKS_ZERO 0x0200
596#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
597 EXT4_GET_BLOCKS_ZERO)
598 /* Caller will submit data before dropping transaction handle. This
599 * allows jbd2 to avoid submitting data before commit. */
600#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
601
602/*
603 * The bit position of these flags must not overlap with any of the
604 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
605 * read_extent_tree_block(), ext4_split_extent_at(),
606 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
607 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
608 * caching the extents when reading from the extent tree while a
609 * truncate or punch hole operation is in progress.
610 */
611#define EXT4_EX_NOCACHE 0x40000000
612#define EXT4_EX_FORCE_CACHE 0x20000000
613
614/*
615 * Flags used by ext4_free_blocks
616 */
617#define EXT4_FREE_BLOCKS_METADATA 0x0001
618#define EXT4_FREE_BLOCKS_FORGET 0x0002
619#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
620#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
621#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
622#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
623
624/*
625 * ioctl commands
626 */
627#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
628#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
629#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
630#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
631#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
632#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
633#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
634#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
635#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
636#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
637#define EXT4_IOC_MIGRATE _IO('f', 9)
638 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
639 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
640#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
641#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
642#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
643#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
644#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
645#define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
646#define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
647#define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
648
649#ifndef FS_IOC_FSGETXATTR
650/* Until the uapi changes get merged for project quota... */
651
652#define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
653#define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
654
655/*
656 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
657 */
658struct fsxattr {
659 __u32 fsx_xflags; /* xflags field value (get/set) */
660 __u32 fsx_extsize; /* extsize field value (get/set)*/
661 __u32 fsx_nextents; /* nextents field value (get) */
662 __u32 fsx_projid; /* project identifier (get/set) */
663 unsigned char fsx_pad[12];
664};
665
666/*
667 * Flags for the fsx_xflags field
668 */
669#define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
670#define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
671#define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
672#define FS_XFLAG_APPEND 0x00000010 /* all writes append */
673#define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
674#define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
675#define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
676#define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
677#define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
678#define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
679#define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
680#define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
681#define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
682#define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
683#define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
684#endif /* !defined(FS_IOC_FSGETXATTR) */
685
686#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
687#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
688
689#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
690
691/*
692 * Flags for going down operation
693 */
694#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
695#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
696#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
697
698
699#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
700/*
701 * ioctl commands in 32 bit emulation
702 */
703#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
704#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
705#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
706#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
707#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
708#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
709#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
710#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
711#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
712#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
713#endif
714
715/* Max physical block we can address w/o extents */
716#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
717
718/* Max logical block we can support */
719#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
720
721/*
722 * Structure of an inode on the disk
723 */
724struct ext4_inode {
725 __le16 i_mode; /* File mode */
726 __le16 i_uid; /* Low 16 bits of Owner Uid */
727 __le32 i_size_lo; /* Size in bytes */
728 __le32 i_atime; /* Access time */
729 __le32 i_ctime; /* Inode Change time */
730 __le32 i_mtime; /* Modification time */
731 __le32 i_dtime; /* Deletion Time */
732 __le16 i_gid; /* Low 16 bits of Group Id */
733 __le16 i_links_count; /* Links count */
734 __le32 i_blocks_lo; /* Blocks count */
735 __le32 i_flags; /* File flags */
736 union {
737 struct {
738 __le32 l_i_version;
739 } linux1;
740 struct {
741 __u32 h_i_translator;
742 } hurd1;
743 struct {
744 __u32 m_i_reserved1;
745 } masix1;
746 } osd1; /* OS dependent 1 */
747 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
748 __le32 i_generation; /* File version (for NFS) */
749 __le32 i_file_acl_lo; /* File ACL */
750 __le32 i_size_high;
751 __le32 i_obso_faddr; /* Obsoleted fragment address */
752 union {
753 struct {
754 __le16 l_i_blocks_high; /* were l_i_reserved1 */
755 __le16 l_i_file_acl_high;
756 __le16 l_i_uid_high; /* these 2 fields */
757 __le16 l_i_gid_high; /* were reserved2[0] */
758 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
759 __le16 l_i_reserved;
760 } linux2;
761 struct {
762 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
763 __u16 h_i_mode_high;
764 __u16 h_i_uid_high;
765 __u16 h_i_gid_high;
766 __u32 h_i_author;
767 } hurd2;
768 struct {
769 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
770 __le16 m_i_file_acl_high;
771 __u32 m_i_reserved2[2];
772 } masix2;
773 } osd2; /* OS dependent 2 */
774 __le16 i_extra_isize;
775 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
776 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
777 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
778 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
779 __le32 i_crtime; /* File Creation time */
780 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
781 __le32 i_version_hi; /* high 32 bits for 64-bit version */
782 __le32 i_projid; /* Project ID */
783};
784
785struct move_extent {
786 __u32 reserved; /* should be zero */
787 __u32 donor_fd; /* donor file descriptor */
788 __u64 orig_start; /* logical start offset in block for orig */
789 __u64 donor_start; /* logical start offset in block for donor */
790 __u64 len; /* block length to be moved */
791 __u64 moved_len; /* moved block length */
792};
793
794#define EXT4_EPOCH_BITS 2
795#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
796#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
797
798/*
799 * Extended fields will fit into an inode if the filesystem was formatted
800 * with large inodes (-I 256 or larger) and there are not currently any EAs
801 * consuming all of the available space. For new inodes we always reserve
802 * enough space for the kernel's known extended fields, but for inodes
803 * created with an old kernel this might not have been the case. None of
804 * the extended inode fields is critical for correct filesystem operation.
805 * This macro checks if a certain field fits in the inode. Note that
806 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
807 */
808#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
809 ((offsetof(typeof(*ext4_inode), field) + \
810 sizeof((ext4_inode)->field)) \
811 <= (EXT4_GOOD_OLD_INODE_SIZE + \
812 (einode)->i_extra_isize)) \
813
814/*
815 * We use an encoding that preserves the times for extra epoch "00":
816 *
817 * extra msb of adjust for signed
818 * epoch 32-bit 32-bit tv_sec to
819 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
820 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
821 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
822 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
823 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
824 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
825 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
826 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
827 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
828 *
829 * Note that previous versions of the kernel on 64-bit systems would
830 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
831 * 1970. e2fsck will correct this, assuming that it is run on the
832 * affected filesystem before 2242.
833 */
834
835static inline __le32 ext4_encode_extra_time(struct timespec *time)
836{
837 u32 extra = sizeof(time->tv_sec) > 4 ?
838 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
839 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
840}
841
842static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
843{
844 if (unlikely(sizeof(time->tv_sec) > 4 &&
845 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
846
847#if 1
848 /* Handle legacy encoding of pre-1970 dates with epoch
849 * bits 1,1. (This backwards compatibility may be removed
850 * at the discretion of the ext4 developers.)
851 */
852 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
853 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
854 extra_bits = 0;
855 time->tv_sec += extra_bits << 32;
856#else
857 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
858#endif
859 }
860 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
861}
862
863#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
864do { \
865 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
866 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
867 (raw_inode)->xtime ## _extra = \
868 ext4_encode_extra_time(&(inode)->xtime); \
869} while (0)
870
871#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
872do { \
873 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
874 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
875 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
876 (raw_inode)->xtime ## _extra = \
877 ext4_encode_extra_time(&(einode)->xtime); \
878} while (0)
879
880#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
881do { \
882 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
883 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
884 ext4_decode_extra_time(&(inode)->xtime, \
885 raw_inode->xtime ## _extra); \
886 else \
887 (inode)->xtime.tv_nsec = 0; \
888} while (0)
889
890#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
891do { \
892 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
893 (einode)->xtime.tv_sec = \
894 (signed)le32_to_cpu((raw_inode)->xtime); \
895 else \
896 (einode)->xtime.tv_sec = 0; \
897 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
898 ext4_decode_extra_time(&(einode)->xtime, \
899 raw_inode->xtime ## _extra); \
900 else \
901 (einode)->xtime.tv_nsec = 0; \
902} while (0)
903
904#define i_disk_version osd1.linux1.l_i_version
905
906#if defined(__KERNEL__) || defined(__linux__)
907#define i_reserved1 osd1.linux1.l_i_reserved1
908#define i_file_acl_high osd2.linux2.l_i_file_acl_high
909#define i_blocks_high osd2.linux2.l_i_blocks_high
910#define i_uid_low i_uid
911#define i_gid_low i_gid
912#define i_uid_high osd2.linux2.l_i_uid_high
913#define i_gid_high osd2.linux2.l_i_gid_high
914#define i_checksum_lo osd2.linux2.l_i_checksum_lo
915
916#elif defined(__GNU__)
917
918#define i_translator osd1.hurd1.h_i_translator
919#define i_uid_high osd2.hurd2.h_i_uid_high
920#define i_gid_high osd2.hurd2.h_i_gid_high
921#define i_author osd2.hurd2.h_i_author
922
923#elif defined(__masix__)
924
925#define i_reserved1 osd1.masix1.m_i_reserved1
926#define i_file_acl_high osd2.masix2.m_i_file_acl_high
927#define i_reserved2 osd2.masix2.m_i_reserved2
928
929#endif /* defined(__KERNEL__) || defined(__linux__) */
930
931#include "extents_status.h"
932
933/*
934 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
935 *
936 * These are needed to avoid lockdep false positives when we need to
937 * allocate blocks to the quota inode during ext4_map_blocks(), while
938 * holding i_data_sem for a normal (non-quota) inode. Since we don't
939 * do quota tracking for the quota inode, this avoids deadlock (as
940 * well as infinite recursion, since it isn't turtles all the way
941 * down...)
942 *
943 * I_DATA_SEM_NORMAL - Used for most inodes
944 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
945 * where the second inode has larger inode number
946 * than the first
947 * I_DATA_SEM_QUOTA - Used for quota inodes only
948 */
949enum {
950 I_DATA_SEM_NORMAL = 0,
951 I_DATA_SEM_OTHER,
952 I_DATA_SEM_QUOTA,
953};
954
955
956/*
957 * fourth extended file system inode data in memory
958 */
959struct ext4_inode_info {
960 __le32 i_data[15]; /* unconverted */
961 __u32 i_dtime;
962 ext4_fsblk_t i_file_acl;
963
964 /*
965 * i_block_group is the number of the block group which contains
966 * this file's inode. Constant across the lifetime of the inode,
967 * it is used for making block allocation decisions - we try to
968 * place a file's data blocks near its inode block, and new inodes
969 * near to their parent directory's inode.
970 */
971 ext4_group_t i_block_group;
972 ext4_lblk_t i_dir_start_lookup;
973#if (BITS_PER_LONG < 64)
974 unsigned long i_state_flags; /* Dynamic state flags */
975#endif
976 unsigned long i_flags;
977
978 /*
979 * Extended attributes can be read independently of the main file
980 * data. Taking i_mutex even when reading would cause contention
981 * between readers of EAs and writers of regular file data, so
982 * instead we synchronize on xattr_sem when reading or changing
983 * EAs.
984 */
985 struct rw_semaphore xattr_sem;
986
987 struct list_head i_orphan; /* unlinked but open inodes */
988
989 /*
990 * i_disksize keeps track of what the inode size is ON DISK, not
991 * in memory. During truncate, i_size is set to the new size by
992 * the VFS prior to calling ext4_truncate(), but the filesystem won't
993 * set i_disksize to 0 until the truncate is actually under way.
994 *
995 * The intent is that i_disksize always represents the blocks which
996 * are used by this file. This allows recovery to restart truncate
997 * on orphans if we crash during truncate. We actually write i_disksize
998 * into the on-disk inode when writing inodes out, instead of i_size.
999 *
1000 * The only time when i_disksize and i_size may be different is when
1001 * a truncate is in progress. The only things which change i_disksize
1002 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1003 */
1004 loff_t i_disksize;
1005
1006 /*
1007 * i_data_sem is for serialising ext4_truncate() against
1008 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1009 * data tree are chopped off during truncate. We can't do that in
1010 * ext4 because whenever we perform intermediate commits during
1011 * truncate, the inode and all the metadata blocks *must* be in a
1012 * consistent state which allows truncation of the orphans to restart
1013 * during recovery. Hence we must fix the get_block-vs-truncate race
1014 * by other means, so we have i_data_sem.
1015 */
1016 struct rw_semaphore i_data_sem;
1017 /*
1018 * i_mmap_sem is for serializing page faults with truncate / punch hole
1019 * operations. We have to make sure that new page cannot be faulted in
1020 * a section of the inode that is being punched. We cannot easily use
1021 * i_data_sem for this since we need protection for the whole punch
1022 * operation and i_data_sem ranks below transaction start so we have
1023 * to occasionally drop it.
1024 */
1025 struct rw_semaphore i_mmap_sem;
1026 struct inode vfs_inode;
1027 struct jbd2_inode *jinode;
1028
1029 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1030
1031 /*
1032 * File creation time. Its function is same as that of
1033 * struct timespec i_{a,c,m}time in the generic inode.
1034 */
1035 struct timespec i_crtime;
1036
1037 /* mballoc */
1038 struct list_head i_prealloc_list;
1039 spinlock_t i_prealloc_lock;
1040
1041 /* extents status tree */
1042 struct ext4_es_tree i_es_tree;
1043 rwlock_t i_es_lock;
1044 struct list_head i_es_list;
1045 unsigned int i_es_all_nr; /* protected by i_es_lock */
1046 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1047 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1048 extents to shrink. Protected by
1049 i_es_lock */
1050
1051 /* ialloc */
1052 ext4_group_t i_last_alloc_group;
1053
1054 /* allocation reservation info for delalloc */
1055 /* In case of bigalloc, this refer to clusters rather than blocks */
1056 unsigned int i_reserved_data_blocks;
1057 ext4_lblk_t i_da_metadata_calc_last_lblock;
1058 int i_da_metadata_calc_len;
1059
1060 /* on-disk additional length */
1061 __u16 i_extra_isize;
1062
1063 /* Indicate the inline data space. */
1064 u16 i_inline_off;
1065 u16 i_inline_size;
1066
1067#ifdef CONFIG_QUOTA
1068 /* quota space reservation, managed internally by quota code */
1069 qsize_t i_reserved_quota;
1070#endif
1071
1072 /* Lock protecting lists below */
1073 spinlock_t i_completed_io_lock;
1074 /*
1075 * Completed IOs that need unwritten extents handling and have
1076 * transaction reserved
1077 */
1078 struct list_head i_rsv_conversion_list;
1079 struct work_struct i_rsv_conversion_work;
1080 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1081
1082 spinlock_t i_block_reservation_lock;
1083
1084 /*
1085 * Transactions that contain inode's metadata needed to complete
1086 * fsync and fdatasync, respectively.
1087 */
1088 tid_t i_sync_tid;
1089 tid_t i_datasync_tid;
1090
1091#ifdef CONFIG_QUOTA
1092 struct dquot *i_dquot[MAXQUOTAS];
1093#endif
1094
1095 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1096 __u32 i_csum_seed;
1097
1098 kprojid_t i_projid;
1099};
1100
1101/*
1102 * File system states
1103 */
1104#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1105#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1106#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1107
1108/*
1109 * Misc. filesystem flags
1110 */
1111#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1112#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1113#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1114
1115/*
1116 * Mount flags set via mount options or defaults
1117 */
1118#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1119#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1120#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1121#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1122#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1123#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1124#define EXT4_MOUNT_ERRORS_MASK 0x00070
1125#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1126#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1127#ifdef CONFIG_FS_DAX
1128#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1129#else
1130#define EXT4_MOUNT_DAX 0
1131#endif
1132#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1133#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1134#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1135#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1136#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1137#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1138#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1139#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1140#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1141#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1142#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1143#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1144 * enable enforcement for hidden
1145 * quota files */
1146#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1147 * enforcement for hidden quota
1148 * files */
1149#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1150 * enforcement */
1151#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1152#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1153#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1154#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1155#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1156#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1157#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1158#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1159
1160/*
1161 * Mount flags set either automatically (could not be set by mount option)
1162 * based on per file system feature or property or in special cases such as
1163 * distinguishing between explicit mount option definition and default.
1164 */
1165#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1166 specified delalloc */
1167#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1168 size of blocksize * 8
1169 blocks */
1170#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1171 file systems */
1172
1173#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1174 specified journal checksum */
1175
1176#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1177 ~EXT4_MOUNT_##opt
1178#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1179 EXT4_MOUNT_##opt
1180#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1181 EXT4_MOUNT_##opt)
1182
1183#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1184 ~EXT4_MOUNT2_##opt
1185#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1186 EXT4_MOUNT2_##opt
1187#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1188 EXT4_MOUNT2_##opt)
1189
1190#define ext4_test_and_set_bit __test_and_set_bit_le
1191#define ext4_set_bit __set_bit_le
1192#define ext4_set_bit_atomic ext2_set_bit_atomic
1193#define ext4_test_and_clear_bit __test_and_clear_bit_le
1194#define ext4_clear_bit __clear_bit_le
1195#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1196#define ext4_test_bit test_bit_le
1197#define ext4_find_next_zero_bit find_next_zero_bit_le
1198#define ext4_find_next_bit find_next_bit_le
1199
1200extern void ext4_set_bits(void *bm, int cur, int len);
1201
1202/*
1203 * Maximal mount counts between two filesystem checks
1204 */
1205#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1206#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1207
1208/*
1209 * Behaviour when detecting errors
1210 */
1211#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1212#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1213#define EXT4_ERRORS_PANIC 3 /* Panic */
1214#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1215
1216/* Metadata checksum algorithm codes */
1217#define EXT4_CRC32C_CHKSUM 1
1218
1219/*
1220 * Structure of the super block
1221 */
1222struct ext4_super_block {
1223/*00*/ __le32 s_inodes_count; /* Inodes count */
1224 __le32 s_blocks_count_lo; /* Blocks count */
1225 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1226 __le32 s_free_blocks_count_lo; /* Free blocks count */
1227/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1228 __le32 s_first_data_block; /* First Data Block */
1229 __le32 s_log_block_size; /* Block size */
1230 __le32 s_log_cluster_size; /* Allocation cluster size */
1231/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1232 __le32 s_clusters_per_group; /* # Clusters per group */
1233 __le32 s_inodes_per_group; /* # Inodes per group */
1234 __le32 s_mtime; /* Mount time */
1235/*30*/ __le32 s_wtime; /* Write time */
1236 __le16 s_mnt_count; /* Mount count */
1237 __le16 s_max_mnt_count; /* Maximal mount count */
1238 __le16 s_magic; /* Magic signature */
1239 __le16 s_state; /* File system state */
1240 __le16 s_errors; /* Behaviour when detecting errors */
1241 __le16 s_minor_rev_level; /* minor revision level */
1242/*40*/ __le32 s_lastcheck; /* time of last check */
1243 __le32 s_checkinterval; /* max. time between checks */
1244 __le32 s_creator_os; /* OS */
1245 __le32 s_rev_level; /* Revision level */
1246/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1247 __le16 s_def_resgid; /* Default gid for reserved blocks */
1248 /*
1249 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1250 *
1251 * Note: the difference between the compatible feature set and
1252 * the incompatible feature set is that if there is a bit set
1253 * in the incompatible feature set that the kernel doesn't
1254 * know about, it should refuse to mount the filesystem.
1255 *
1256 * e2fsck's requirements are more strict; if it doesn't know
1257 * about a feature in either the compatible or incompatible
1258 * feature set, it must abort and not try to meddle with
1259 * things it doesn't understand...
1260 */
1261 __le32 s_first_ino; /* First non-reserved inode */
1262 __le16 s_inode_size; /* size of inode structure */
1263 __le16 s_block_group_nr; /* block group # of this superblock */
1264 __le32 s_feature_compat; /* compatible feature set */
1265/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1266 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1267/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1268/*78*/ char s_volume_name[16]; /* volume name */
1269/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1270/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1271 /*
1272 * Performance hints. Directory preallocation should only
1273 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1274 */
1275 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1276 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1277 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1278 /*
1279 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1280 */
1281/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1282/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1283 __le32 s_journal_dev; /* device number of journal file */
1284 __le32 s_last_orphan; /* start of list of inodes to delete */
1285 __le32 s_hash_seed[4]; /* HTREE hash seed */
1286 __u8 s_def_hash_version; /* Default hash version to use */
1287 __u8 s_jnl_backup_type;
1288 __le16 s_desc_size; /* size of group descriptor */
1289/*100*/ __le32 s_default_mount_opts;
1290 __le32 s_first_meta_bg; /* First metablock block group */
1291 __le32 s_mkfs_time; /* When the filesystem was created */
1292 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1293 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1294/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1295 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1296 __le32 s_free_blocks_count_hi; /* Free blocks count */
1297 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1298 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1299 __le32 s_flags; /* Miscellaneous flags */
1300 __le16 s_raid_stride; /* RAID stride */
1301 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1302 __le64 s_mmp_block; /* Block for multi-mount protection */
1303 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1304 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1305 __u8 s_checksum_type; /* metadata checksum algorithm used */
1306 __u8 s_encryption_level; /* versioning level for encryption */
1307 __u8 s_reserved_pad; /* Padding to next 32bits */
1308 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1309 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1310 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1311 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1312 snapshot's future use */
1313 __le32 s_snapshot_list; /* inode number of the head of the
1314 on-disk snapshot list */
1315#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1316 __le32 s_error_count; /* number of fs errors */
1317 __le32 s_first_error_time; /* first time an error happened */
1318 __le32 s_first_error_ino; /* inode involved in first error */
1319 __le64 s_first_error_block; /* block involved of first error */
1320 __u8 s_first_error_func[32]; /* function where the error happened */
1321 __le32 s_first_error_line; /* line number where error happened */
1322 __le32 s_last_error_time; /* most recent time of an error */
1323 __le32 s_last_error_ino; /* inode involved in last error */
1324 __le32 s_last_error_line; /* line number where error happened */
1325 __le64 s_last_error_block; /* block involved of last error */
1326 __u8 s_last_error_func[32]; /* function where the error happened */
1327#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1328 __u8 s_mount_opts[64];
1329 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1330 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1331 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1332 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1333 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1334 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1335 __le32 s_lpf_ino; /* Location of the lost+found inode */
1336 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1337 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1338 __le32 s_reserved[98]; /* Padding to the end of the block */
1339 __le32 s_checksum; /* crc32c(superblock) */
1340};
1341
1342#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1343
1344#ifdef __KERNEL__
1345
1346/*
1347 * run-time mount flags
1348 */
1349#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1350#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1351#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1352
1353#ifdef CONFIG_EXT4_FS_ENCRYPTION
1354#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1355 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1356#else
1357#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1358#endif
1359
1360/* Number of quota types we support */
1361#define EXT4_MAXQUOTAS 3
1362
1363/*
1364 * fourth extended-fs super-block data in memory
1365 */
1366struct ext4_sb_info {
1367 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1368 unsigned long s_inodes_per_block;/* Number of inodes per block */
1369 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1370 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1371 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1372 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1373 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1374 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1375 ext4_group_t s_groups_count; /* Number of groups in the fs */
1376 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1377 unsigned long s_overhead; /* # of fs overhead clusters */
1378 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1379 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1380 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1381 struct buffer_head * s_sbh; /* Buffer containing the super block */
1382 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1383 struct buffer_head * __rcu *s_group_desc;
1384 unsigned int s_mount_opt;
1385 unsigned int s_mount_opt2;
1386 unsigned int s_mount_flags;
1387 unsigned int s_def_mount_opt;
1388 ext4_fsblk_t s_sb_block;
1389 atomic64_t s_resv_clusters;
1390 kuid_t s_resuid;
1391 kgid_t s_resgid;
1392 unsigned short s_mount_state;
1393 unsigned short s_pad;
1394 int s_addr_per_block_bits;
1395 int s_desc_per_block_bits;
1396 int s_inode_size;
1397 int s_first_ino;
1398 unsigned int s_inode_readahead_blks;
1399 unsigned int s_inode_goal;
1400 spinlock_t s_next_gen_lock;
1401 u32 s_next_generation;
1402 u32 s_hash_seed[4];
1403 int s_def_hash_version;
1404 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1405 struct percpu_counter s_freeclusters_counter;
1406 struct percpu_counter s_freeinodes_counter;
1407 struct percpu_counter s_dirs_counter;
1408 struct percpu_counter s_dirtyclusters_counter;
1409 struct blockgroup_lock *s_blockgroup_lock;
1410 struct proc_dir_entry *s_proc;
1411 struct kobject s_kobj;
1412 struct completion s_kobj_unregister;
1413 struct super_block *s_sb;
1414
1415 /* Journaling */
1416 struct journal_s *s_journal;
1417 struct list_head s_orphan;
1418 struct mutex s_orphan_lock;
1419 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1420 unsigned long s_commit_interval;
1421 u32 s_max_batch_time;
1422 u32 s_min_batch_time;
1423 struct block_device *journal_bdev;
1424#ifdef CONFIG_QUOTA
1425 /* Names of quota files with journalled quota */
1426 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1427 int s_jquota_fmt; /* Format of quota to use */
1428#endif
1429 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1430 struct rb_root system_blks;
1431
1432#ifdef EXTENTS_STATS
1433 /* ext4 extents stats */
1434 unsigned long s_ext_min;
1435 unsigned long s_ext_max;
1436 unsigned long s_depth_max;
1437 spinlock_t s_ext_stats_lock;
1438 unsigned long s_ext_blocks;
1439 unsigned long s_ext_extents;
1440#endif
1441
1442 /* for buddy allocator */
1443 struct ext4_group_info ** __rcu *s_group_info;
1444 struct inode *s_buddy_cache;
1445 spinlock_t s_md_lock;
1446 unsigned short *s_mb_offsets;
1447 unsigned int *s_mb_maxs;
1448 unsigned int s_group_info_size;
1449 unsigned int s_mb_free_pending;
1450 struct list_head s_freed_data_list; /* List of blocks to be freed
1451 after commit completed */
1452
1453 /* tunables */
1454 unsigned long s_stripe;
1455 unsigned int s_mb_stream_request;
1456 unsigned int s_mb_max_to_scan;
1457 unsigned int s_mb_min_to_scan;
1458 unsigned int s_mb_stats;
1459 unsigned int s_mb_order2_reqs;
1460 unsigned int s_mb_group_prealloc;
1461 unsigned int s_max_dir_size_kb;
1462 /* where last allocation was done - for stream allocation */
1463 unsigned long s_mb_last_group;
1464 unsigned long s_mb_last_start;
1465
1466 /* stats for buddy allocator */
1467 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1468 atomic_t s_bal_success; /* we found long enough chunks */
1469 atomic_t s_bal_allocated; /* in blocks */
1470 atomic_t s_bal_ex_scanned; /* total extents scanned */
1471 atomic_t s_bal_goals; /* goal hits */
1472 atomic_t s_bal_breaks; /* too long searches */
1473 atomic_t s_bal_2orders; /* 2^order hits */
1474 spinlock_t s_bal_lock;
1475 unsigned long s_mb_buddies_generated;
1476 unsigned long long s_mb_generation_time;
1477 atomic_t s_mb_lost_chunks;
1478 atomic_t s_mb_preallocated;
1479 atomic_t s_mb_discarded;
1480 atomic_t s_lock_busy;
1481
1482 /* locality groups */
1483 struct ext4_locality_group __percpu *s_locality_groups;
1484
1485 /* for write statistics */
1486 unsigned long s_sectors_written_start;
1487 u64 s_kbytes_written;
1488
1489 /* the size of zero-out chunk */
1490 unsigned int s_extent_max_zeroout_kb;
1491
1492 unsigned int s_log_groups_per_flex;
1493 struct flex_groups * __rcu *s_flex_groups;
1494 ext4_group_t s_flex_groups_allocated;
1495
1496 /* workqueue for reserved extent conversions (buffered io) */
1497 struct workqueue_struct *rsv_conversion_wq;
1498
1499 /* timer for periodic error stats printing */
1500 struct timer_list s_err_report;
1501
1502 /* Lazy inode table initialization info */
1503 struct ext4_li_request *s_li_request;
1504 /* Wait multiplier for lazy initialization thread */
1505 unsigned int s_li_wait_mult;
1506
1507 /* Kernel thread for multiple mount protection */
1508 struct task_struct *s_mmp_tsk;
1509
1510 /* record the last minlen when FITRIM is called. */
1511 atomic_t s_last_trim_minblks;
1512
1513 /* Reference to checksum algorithm driver via cryptoapi */
1514 struct crypto_shash *s_chksum_driver;
1515
1516 /* Precomputed FS UUID checksum for seeding other checksums */
1517 __u32 s_csum_seed;
1518
1519 /* Reclaim extents from extent status tree */
1520 struct shrinker s_es_shrinker;
1521 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1522 long s_es_nr_inode;
1523 struct ext4_es_stats s_es_stats;
1524 struct mb_cache *s_ea_block_cache;
1525 struct mb_cache *s_ea_inode_cache;
1526 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1527
1528 /* Ratelimit ext4 messages. */
1529 struct ratelimit_state s_err_ratelimit_state;
1530 struct ratelimit_state s_warning_ratelimit_state;
1531 struct ratelimit_state s_msg_ratelimit_state;
1532
1533 /*
1534 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1535 * or EXTENTS flag.
1536 */
1537 struct percpu_rw_semaphore s_writepages_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 * Returns: sbi->field[index]
1559 * Used to access an array element from the following sbi fields which require
1560 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1561 * - s_group_desc
1562 * - s_group_info
1563 * - s_flex_group
1564 */
1565#define sbi_array_rcu_deref(sbi, field, index) \
1566({ \
1567 typeof(*((sbi)->field)) _v; \
1568 rcu_read_lock(); \
1569 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1570 rcu_read_unlock(); \
1571 _v; \
1572})
1573
1574/*
1575 * Inode dynamic state flags
1576 */
1577enum {
1578 EXT4_STATE_JDATA, /* journaled data exists */
1579 EXT4_STATE_NEW, /* inode is newly created */
1580 EXT4_STATE_XATTR, /* has in-inode xattrs */
1581 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1582 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1583 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1584 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1585 EXT4_STATE_NEWENTRY, /* File just added to dir */
1586 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1587 nolocking */
1588 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1589 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1590 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1591};
1592
1593#define EXT4_INODE_BIT_FNS(name, field, offset) \
1594static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1595{ \
1596 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1597} \
1598static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1599{ \
1600 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1601} \
1602static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1603{ \
1604 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1605}
1606
1607/* Add these declarations here only so that these functions can be
1608 * found by name. Otherwise, they are very hard to locate. */
1609static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1610static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1611static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1612EXT4_INODE_BIT_FNS(flag, flags, 0)
1613
1614/* Add these declarations here only so that these functions can be
1615 * found by name. Otherwise, they are very hard to locate. */
1616static inline int ext4_test_inode_state(struct inode *inode, int bit);
1617static inline void ext4_set_inode_state(struct inode *inode, int bit);
1618static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1619#if (BITS_PER_LONG < 64)
1620EXT4_INODE_BIT_FNS(state, state_flags, 0)
1621
1622static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1623{
1624 (ei)->i_state_flags = 0;
1625}
1626#else
1627EXT4_INODE_BIT_FNS(state, flags, 32)
1628
1629static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1630{
1631 /* We depend on the fact that callers will set i_flags */
1632}
1633#endif
1634#else
1635/* Assume that user mode programs are passing in an ext4fs superblock, not
1636 * a kernel struct super_block. This will allow us to call the feature-test
1637 * macros from user land. */
1638#define EXT4_SB(sb) (sb)
1639#endif
1640
1641/*
1642 * Returns true if the inode is inode is encrypted
1643 */
1644#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1645
1646/*
1647 * Codes for operating systems
1648 */
1649#define EXT4_OS_LINUX 0
1650#define EXT4_OS_HURD 1
1651#define EXT4_OS_MASIX 2
1652#define EXT4_OS_FREEBSD 3
1653#define EXT4_OS_LITES 4
1654
1655/*
1656 * Revision levels
1657 */
1658#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1659#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1660
1661#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1662#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1663
1664#define EXT4_GOOD_OLD_INODE_SIZE 128
1665
1666/*
1667 * Feature set definitions
1668 */
1669
1670#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1671#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1672#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1673#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1674#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1675#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1676#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1677
1678#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1679#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1680#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1681#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1682#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1683#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1684#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1685#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1686#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1687/*
1688 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1689 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1690 * all other data structures' checksums. However, the METADATA_CSUM and
1691 * GDT_CSUM bits are mutually exclusive.
1692 */
1693#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1694#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1695#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1696
1697#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1698#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1699#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1700#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1701#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1702#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1703#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1704#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1705#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1706#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1707#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1708#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1709#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1710#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1711#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1712
1713#define EXT4_FEATURE_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_compat & \
1717 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1718} \
1719static inline void ext4_set_feature_##name(struct super_block *sb) \
1720{ \
1721 EXT4_SB(sb)->s_es->s_feature_compat |= \
1722 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1723} \
1724static inline void ext4_clear_feature_##name(struct super_block *sb) \
1725{ \
1726 EXT4_SB(sb)->s_es->s_feature_compat &= \
1727 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1728}
1729
1730#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1731static inline bool ext4_has_feature_##name(struct super_block *sb) \
1732{ \
1733 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1734 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1735} \
1736static inline void ext4_set_feature_##name(struct super_block *sb) \
1737{ \
1738 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1739 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1740} \
1741static inline void ext4_clear_feature_##name(struct super_block *sb) \
1742{ \
1743 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1744 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1745}
1746
1747#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1748static inline bool ext4_has_feature_##name(struct super_block *sb) \
1749{ \
1750 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1751 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1752} \
1753static inline void ext4_set_feature_##name(struct super_block *sb) \
1754{ \
1755 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1756 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1757} \
1758static inline void ext4_clear_feature_##name(struct super_block *sb) \
1759{ \
1760 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1761 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1762}
1763
1764EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1765EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1766EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1767EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1768EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1769EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1770EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1771
1772EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1773EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1774EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1775EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1776EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1777EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1778EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1779EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1780EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1781EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1782EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1783EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1784
1785EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1786EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1787EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1788EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1789EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1790EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1791EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1792EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1793EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1794EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1795EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1796EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1797EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1798EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1799EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1800
1801#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1802#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1803 EXT4_FEATURE_INCOMPAT_META_BG)
1804#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1805 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1806 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1807
1808#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1809#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1810 EXT4_FEATURE_INCOMPAT_RECOVER| \
1811 EXT4_FEATURE_INCOMPAT_META_BG)
1812#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1813 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1814 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1815
1816#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1817#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1818 EXT4_FEATURE_INCOMPAT_RECOVER| \
1819 EXT4_FEATURE_INCOMPAT_META_BG| \
1820 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1821 EXT4_FEATURE_INCOMPAT_64BIT| \
1822 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1823 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1824 EXT4_FEATURE_INCOMPAT_MMP | \
1825 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1826 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1827 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1828 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1829#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1830 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1831 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1832 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1833 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1834 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1835 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1836 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1837 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1838 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1839 EXT4_FEATURE_RO_COMPAT_PROJECT)
1840
1841#define EXTN_FEATURE_FUNCS(ver) \
1842static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1843{ \
1844 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1845 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1846} \
1847static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1848{ \
1849 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1850 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1851} \
1852static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1853{ \
1854 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1855 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1856}
1857
1858EXTN_FEATURE_FUNCS(2)
1859EXTN_FEATURE_FUNCS(3)
1860EXTN_FEATURE_FUNCS(4)
1861
1862static inline bool ext4_has_compat_features(struct super_block *sb)
1863{
1864 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1865}
1866static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1867{
1868 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1869}
1870static inline bool ext4_has_incompat_features(struct super_block *sb)
1871{
1872 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1873}
1874
1875/*
1876 * Superblock flags
1877 */
1878#define EXT4_FLAGS_RESIZING 0
1879#define EXT4_FLAGS_SHUTDOWN 1
1880
1881static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1882{
1883 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1884}
1885
1886
1887/*
1888 * Default values for user and/or group using reserved blocks
1889 */
1890#define EXT4_DEF_RESUID 0
1891#define EXT4_DEF_RESGID 0
1892
1893/*
1894 * Default project ID
1895 */
1896#define EXT4_DEF_PROJID 0
1897
1898#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1899
1900/*
1901 * Default mount options
1902 */
1903#define EXT4_DEFM_DEBUG 0x0001
1904#define EXT4_DEFM_BSDGROUPS 0x0002
1905#define EXT4_DEFM_XATTR_USER 0x0004
1906#define EXT4_DEFM_ACL 0x0008
1907#define EXT4_DEFM_UID16 0x0010
1908#define EXT4_DEFM_JMODE 0x0060
1909#define EXT4_DEFM_JMODE_DATA 0x0020
1910#define EXT4_DEFM_JMODE_ORDERED 0x0040
1911#define EXT4_DEFM_JMODE_WBACK 0x0060
1912#define EXT4_DEFM_NOBARRIER 0x0100
1913#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1914#define EXT4_DEFM_DISCARD 0x0400
1915#define EXT4_DEFM_NODELALLOC 0x0800
1916
1917/*
1918 * Default journal batch times
1919 */
1920#define EXT4_DEF_MIN_BATCH_TIME 0
1921#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1922
1923/*
1924 * Minimum number of groups in a flexgroup before we separate out
1925 * directories into the first block group of a flexgroup
1926 */
1927#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1928
1929/*
1930 * Structure of a directory entry
1931 */
1932#define EXT4_NAME_LEN 255
1933
1934struct ext4_dir_entry {
1935 __le32 inode; /* Inode number */
1936 __le16 rec_len; /* Directory entry length */
1937 __le16 name_len; /* Name length */
1938 char name[EXT4_NAME_LEN]; /* File name */
1939};
1940
1941/*
1942 * The new version of the directory entry. Since EXT4 structures are
1943 * stored in intel byte order, and the name_len field could never be
1944 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1945 * file_type field.
1946 */
1947struct ext4_dir_entry_2 {
1948 __le32 inode; /* Inode number */
1949 __le16 rec_len; /* Directory entry length */
1950 __u8 name_len; /* Name length */
1951 __u8 file_type;
1952 char name[EXT4_NAME_LEN]; /* File name */
1953};
1954
1955/*
1956 * This is a bogus directory entry at the end of each leaf block that
1957 * records checksums.
1958 */
1959struct ext4_dir_entry_tail {
1960 __le32 det_reserved_zero1; /* Pretend to be unused */
1961 __le16 det_rec_len; /* 12 */
1962 __u8 det_reserved_zero2; /* Zero name length */
1963 __u8 det_reserved_ft; /* 0xDE, fake file type */
1964 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1965};
1966
1967#define EXT4_DIRENT_TAIL(block, blocksize) \
1968 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1969 ((blocksize) - \
1970 sizeof(struct ext4_dir_entry_tail))))
1971
1972/*
1973 * Ext4 directory file types. Only the low 3 bits are used. The
1974 * other bits are reserved for now.
1975 */
1976#define EXT4_FT_UNKNOWN 0
1977#define EXT4_FT_REG_FILE 1
1978#define EXT4_FT_DIR 2
1979#define EXT4_FT_CHRDEV 3
1980#define EXT4_FT_BLKDEV 4
1981#define EXT4_FT_FIFO 5
1982#define EXT4_FT_SOCK 6
1983#define EXT4_FT_SYMLINK 7
1984
1985#define EXT4_FT_MAX 8
1986
1987#define EXT4_FT_DIR_CSUM 0xDE
1988
1989/*
1990 * EXT4_DIR_PAD defines the directory entries boundaries
1991 *
1992 * NOTE: It must be a multiple of 4
1993 */
1994#define EXT4_DIR_PAD 4
1995#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1996#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1997 ~EXT4_DIR_ROUND)
1998#define EXT4_MAX_REC_LEN ((1<<16)-1)
1999
2000/*
2001 * If we ever get support for fs block sizes > page_size, we'll need
2002 * to remove the #if statements in the next two functions...
2003 */
2004static inline unsigned int
2005ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2006{
2007 unsigned len = le16_to_cpu(dlen);
2008
2009#if (PAGE_SIZE >= 65536)
2010 if (len == EXT4_MAX_REC_LEN || len == 0)
2011 return blocksize;
2012 return (len & 65532) | ((len & 3) << 16);
2013#else
2014 return len;
2015#endif
2016}
2017
2018static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2019{
2020 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2021 BUG();
2022#if (PAGE_SIZE >= 65536)
2023 if (len < 65536)
2024 return cpu_to_le16(len);
2025 if (len == blocksize) {
2026 if (blocksize == 65536)
2027 return cpu_to_le16(EXT4_MAX_REC_LEN);
2028 else
2029 return cpu_to_le16(0);
2030 }
2031 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2032#else
2033 return cpu_to_le16(len);
2034#endif
2035}
2036
2037/*
2038 * Hash Tree Directory indexing
2039 * (c) Daniel Phillips, 2001
2040 */
2041
2042#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2043 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2044#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2045 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2046#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2047
2048/* Legal values for the dx_root hash_version field: */
2049
2050#define DX_HASH_LEGACY 0
2051#define DX_HASH_HALF_MD4 1
2052#define DX_HASH_TEA 2
2053#define DX_HASH_LEGACY_UNSIGNED 3
2054#define DX_HASH_HALF_MD4_UNSIGNED 4
2055#define DX_HASH_TEA_UNSIGNED 5
2056
2057static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2058 const void *address, unsigned int length)
2059{
2060 struct {
2061 struct shash_desc shash;
2062 char ctx[4];
2063 } desc;
2064 int err;
2065
2066 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2067
2068 desc.shash.tfm = sbi->s_chksum_driver;
2069 desc.shash.flags = 0;
2070 *(u32 *)desc.ctx = crc;
2071
2072 err = crypto_shash_update(&desc.shash, address, length);
2073 BUG_ON(err);
2074
2075 return *(u32 *)desc.ctx;
2076}
2077
2078#ifdef __KERNEL__
2079
2080/* hash info structure used by the directory hash */
2081struct dx_hash_info
2082{
2083 u32 hash;
2084 u32 minor_hash;
2085 int hash_version;
2086 u32 *seed;
2087};
2088
2089
2090/* 32 and 64 bit signed EOF for dx directories */
2091#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2092#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2093
2094
2095/*
2096 * Control parameters used by ext4_htree_next_block
2097 */
2098#define HASH_NB_ALWAYS 1
2099
2100struct ext4_filename {
2101 const struct qstr *usr_fname;
2102 struct fscrypt_str disk_name;
2103 struct dx_hash_info hinfo;
2104#ifdef CONFIG_EXT4_FS_ENCRYPTION
2105 struct fscrypt_str crypto_buf;
2106#endif
2107};
2108
2109#define fname_name(p) ((p)->disk_name.name)
2110#define fname_len(p) ((p)->disk_name.len)
2111
2112/*
2113 * Describe an inode's exact location on disk and in memory
2114 */
2115struct ext4_iloc
2116{
2117 struct buffer_head *bh;
2118 unsigned long offset;
2119 ext4_group_t block_group;
2120};
2121
2122static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2123{
2124 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2125}
2126
2127static inline bool ext4_is_quota_file(struct inode *inode)
2128{
2129 return IS_NOQUOTA(inode) &&
2130 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2131}
2132
2133/*
2134 * This structure is stuffed into the struct file's private_data field
2135 * for directories. It is where we put information so that we can do
2136 * readdir operations in hash tree order.
2137 */
2138struct dir_private_info {
2139 struct rb_root root;
2140 struct rb_node *curr_node;
2141 struct fname *extra_fname;
2142 loff_t last_pos;
2143 __u32 curr_hash;
2144 __u32 curr_minor_hash;
2145 __u32 next_hash;
2146};
2147
2148/* calculate the first block number of the group */
2149static inline ext4_fsblk_t
2150ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2151{
2152 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2153 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2154}
2155
2156/*
2157 * Special error return code only used by dx_probe() and its callers.
2158 */
2159#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2160
2161/* htree levels for ext4 */
2162#define EXT4_HTREE_LEVEL_COMPAT 2
2163#define EXT4_HTREE_LEVEL 3
2164
2165static inline int ext4_dir_htree_level(struct super_block *sb)
2166{
2167 return ext4_has_feature_largedir(sb) ?
2168 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2169}
2170
2171/*
2172 * Timeout and state flag for lazy initialization inode thread.
2173 */
2174#define EXT4_DEF_LI_WAIT_MULT 10
2175#define EXT4_DEF_LI_MAX_START_DELAY 5
2176#define EXT4_LAZYINIT_QUIT 0x0001
2177#define EXT4_LAZYINIT_RUNNING 0x0002
2178
2179/*
2180 * Lazy inode table initialization info
2181 */
2182struct ext4_lazy_init {
2183 unsigned long li_state;
2184 struct list_head li_request_list;
2185 struct mutex li_list_mtx;
2186};
2187
2188struct ext4_li_request {
2189 struct super_block *lr_super;
2190 struct ext4_sb_info *lr_sbi;
2191 ext4_group_t lr_next_group;
2192 struct list_head lr_request;
2193 unsigned long lr_next_sched;
2194 unsigned long lr_timeout;
2195};
2196
2197struct ext4_features {
2198 struct kobject f_kobj;
2199 struct completion f_kobj_unregister;
2200};
2201
2202/*
2203 * This structure will be used for multiple mount protection. It will be
2204 * written into the block number saved in the s_mmp_block field in the
2205 * superblock. Programs that check MMP should assume that if
2206 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2207 * to use the filesystem, regardless of how old the timestamp is.
2208 */
2209#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2210#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2211#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2212#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2213
2214struct mmp_struct {
2215 __le32 mmp_magic; /* Magic number for MMP */
2216 __le32 mmp_seq; /* Sequence no. updated periodically */
2217
2218 /*
2219 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2220 * purposes and do not affect the correctness of the algorithm
2221 */
2222 __le64 mmp_time; /* Time last updated */
2223 char mmp_nodename[64]; /* Node which last updated MMP block */
2224 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2225
2226 /*
2227 * mmp_check_interval is used to verify if the MMP block has been
2228 * updated on the block device. The value is updated based on the
2229 * maximum time to write the MMP block during an update cycle.
2230 */
2231 __le16 mmp_check_interval;
2232
2233 __le16 mmp_pad1;
2234 __le32 mmp_pad2[226];
2235 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2236};
2237
2238/* arguments passed to the mmp thread */
2239struct mmpd_data {
2240 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2241 struct super_block *sb; /* super block of the fs */
2242};
2243
2244/*
2245 * Check interval multiplier
2246 * The MMP block is written every update interval and initially checked every
2247 * update interval x the multiplier (the value is then adapted based on the
2248 * write latency). The reason is that writes can be delayed under load and we
2249 * don't want readers to incorrectly assume that the filesystem is no longer
2250 * in use.
2251 */
2252#define EXT4_MMP_CHECK_MULT 2UL
2253
2254/*
2255 * Minimum interval for MMP checking in seconds.
2256 */
2257#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2258
2259/*
2260 * Maximum interval for MMP checking in seconds.
2261 */
2262#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2263
2264/*
2265 * Function prototypes
2266 */
2267
2268/*
2269 * Ok, these declarations are also in <linux/kernel.h> but none of the
2270 * ext4 source programs needs to include it so they are duplicated here.
2271 */
2272# define NORET_TYPE /**/
2273# define ATTRIB_NORET __attribute__((noreturn))
2274# define NORET_AND noreturn,
2275
2276/* bitmap.c */
2277extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2278void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2279 struct ext4_group_desc *gdp,
2280 struct buffer_head *bh, int sz);
2281int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2282 struct ext4_group_desc *gdp,
2283 struct buffer_head *bh, int sz);
2284void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2285 struct ext4_group_desc *gdp,
2286 struct buffer_head *bh);
2287int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2288 struct ext4_group_desc *gdp,
2289 struct buffer_head *bh);
2290
2291/* balloc.c */
2292extern void ext4_get_group_no_and_offset(struct super_block *sb,
2293 ext4_fsblk_t blocknr,
2294 ext4_group_t *blockgrpp,
2295 ext4_grpblk_t *offsetp);
2296extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2297 ext4_fsblk_t block);
2298
2299extern unsigned int ext4_block_group(struct super_block *sb,
2300 ext4_fsblk_t blocknr);
2301extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2302 ext4_fsblk_t blocknr);
2303extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2304extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2305 ext4_group_t group);
2306extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2307 ext4_fsblk_t goal,
2308 unsigned int flags,
2309 unsigned long *count,
2310 int *errp);
2311extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2312 s64 nclusters, unsigned int flags);
2313extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2314extern void ext4_check_blocks_bitmap(struct super_block *);
2315extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2316 ext4_group_t block_group,
2317 struct buffer_head ** bh);
2318extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2319
2320extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2321 ext4_group_t block_group);
2322extern int ext4_wait_block_bitmap(struct super_block *sb,
2323 ext4_group_t block_group,
2324 struct buffer_head *bh);
2325extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2326 ext4_group_t block_group);
2327extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2328 ext4_group_t block_group,
2329 struct ext4_group_desc *gdp);
2330ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2331
2332static inline bool ext4_encrypted_inode(struct inode *inode)
2333{
2334 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2335}
2336
2337#ifdef CONFIG_EXT4_FS_ENCRYPTION
2338static inline int ext4_fname_setup_filename(struct inode *dir,
2339 const struct qstr *iname,
2340 int lookup, struct ext4_filename *fname)
2341{
2342 struct fscrypt_name name;
2343 int err;
2344
2345 memset(fname, 0, sizeof(struct ext4_filename));
2346
2347 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2348
2349 fname->usr_fname = name.usr_fname;
2350 fname->disk_name = name.disk_name;
2351 fname->hinfo.hash = name.hash;
2352 fname->hinfo.minor_hash = name.minor_hash;
2353 fname->crypto_buf = name.crypto_buf;
2354 return err;
2355}
2356
2357static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2358{
2359 struct fscrypt_name name;
2360
2361 name.crypto_buf = fname->crypto_buf;
2362 fscrypt_free_filename(&name);
2363
2364 fname->crypto_buf.name = NULL;
2365 fname->usr_fname = NULL;
2366 fname->disk_name.name = NULL;
2367}
2368#else
2369static inline int ext4_fname_setup_filename(struct inode *dir,
2370 const struct qstr *iname,
2371 int lookup, struct ext4_filename *fname)
2372{
2373 fname->usr_fname = iname;
2374 fname->disk_name.name = (unsigned char *) iname->name;
2375 fname->disk_name.len = iname->len;
2376 return 0;
2377}
2378static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2379
2380#endif
2381
2382/* dir.c */
2383extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2384 struct file *,
2385 struct ext4_dir_entry_2 *,
2386 struct buffer_head *, char *, int,
2387 unsigned int);
2388#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2389 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2390 (de), (bh), (buf), (size), (offset)))
2391extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2392 __u32 minor_hash,
2393 struct ext4_dir_entry_2 *dirent,
2394 struct fscrypt_str *ent_name);
2395extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2396extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2397 struct buffer_head *bh,
2398 void *buf, int buf_size,
2399 struct ext4_filename *fname,
2400 struct ext4_dir_entry_2 **dest_de);
2401void ext4_insert_dentry(struct inode *inode,
2402 struct ext4_dir_entry_2 *de,
2403 int buf_size,
2404 struct ext4_filename *fname);
2405static inline void ext4_update_dx_flag(struct inode *inode)
2406{
2407 if (!ext4_has_feature_dir_index(inode->i_sb)) {
2408 /* ext4_iget() should have caught this... */
2409 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2410 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2411 }
2412}
2413static const unsigned char ext4_filetype_table[] = {
2414 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2415};
2416
2417static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2418{
2419 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2420 return DT_UNKNOWN;
2421
2422 return ext4_filetype_table[filetype];
2423}
2424extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2425 void *buf, int buf_size);
2426
2427/* fsync.c */
2428extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2429
2430/* hash.c */
2431extern int ext4fs_dirhash(const char *name, int len, struct
2432 dx_hash_info *hinfo);
2433
2434/* ialloc.c */
2435extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2436 const struct qstr *qstr, __u32 goal,
2437 uid_t *owner, __u32 i_flags,
2438 int handle_type, unsigned int line_no,
2439 int nblocks);
2440
2441#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2442 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2443 i_flags, 0, 0, 0)
2444#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2445 type, nblocks) \
2446 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2447 0, (type), __LINE__, (nblocks))
2448
2449
2450extern void ext4_free_inode(handle_t *, struct inode *);
2451extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2452extern unsigned long ext4_count_free_inodes(struct super_block *);
2453extern unsigned long ext4_count_dirs(struct super_block *);
2454extern void ext4_check_inodes_bitmap(struct super_block *);
2455extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2456extern int ext4_init_inode_table(struct super_block *sb,
2457 ext4_group_t group, int barrier);
2458extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2459
2460/* mballoc.c */
2461extern const struct file_operations ext4_seq_mb_groups_fops;
2462extern long ext4_mb_stats;
2463extern long ext4_mb_max_to_scan;
2464extern int ext4_mb_init(struct super_block *);
2465extern int ext4_mb_release(struct super_block *);
2466extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2467 struct ext4_allocation_request *, int *);
2468extern int ext4_mb_reserve_blocks(struct super_block *, int);
2469extern void ext4_discard_preallocations(struct inode *);
2470extern int __init ext4_init_mballoc(void);
2471extern void ext4_exit_mballoc(void);
2472extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2473 struct buffer_head *bh, ext4_fsblk_t block,
2474 unsigned long count, int flags);
2475extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2476 ext4_group_t ngroups);
2477extern int ext4_mb_add_groupinfo(struct super_block *sb,
2478 ext4_group_t i, struct ext4_group_desc *desc);
2479extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2480 ext4_fsblk_t block, unsigned long count);
2481extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2482extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2483
2484/* inode.c */
2485int ext4_inode_is_fast_symlink(struct inode *inode);
2486struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2487struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2488int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2489 bool wait, struct buffer_head **bhs);
2490int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2491 struct buffer_head *bh_result, int create);
2492int ext4_get_block(struct inode *inode, sector_t iblock,
2493 struct buffer_head *bh_result, int create);
2494int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2495 struct buffer_head *bh_result, int create);
2496int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2497 struct buffer_head *bh, int create);
2498int ext4_walk_page_buffers(handle_t *handle,
2499 struct buffer_head *head,
2500 unsigned from,
2501 unsigned to,
2502 int *partial,
2503 int (*fn)(handle_t *handle,
2504 struct buffer_head *bh));
2505int do_journal_get_write_access(handle_t *handle,
2506 struct buffer_head *bh);
2507#define FALL_BACK_TO_NONDELALLOC 1
2508#define CONVERT_INLINE_DATA 2
2509
2510typedef enum {
2511 EXT4_IGET_NORMAL = 0,
2512 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2513 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2514} ext4_iget_flags;
2515
2516extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2517 ext4_iget_flags flags, const char *function,
2518 unsigned int line);
2519
2520#define ext4_iget(sb, ino, flags) \
2521 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2522
2523extern int ext4_write_inode(struct inode *, struct writeback_control *);
2524extern int ext4_setattr(struct dentry *, struct iattr *);
2525extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2526extern void ext4_evict_inode(struct inode *);
2527extern void ext4_clear_inode(struct inode *);
2528extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2529extern int ext4_sync_inode(handle_t *, struct inode *);
2530extern void ext4_dirty_inode(struct inode *, int);
2531extern int ext4_change_inode_journal_flag(struct inode *, int);
2532extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2533extern int ext4_inode_attach_jinode(struct inode *inode);
2534extern int ext4_can_truncate(struct inode *inode);
2535extern int ext4_truncate(struct inode *);
2536extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2537extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2538extern void ext4_set_inode_flags(struct inode *);
2539extern int ext4_alloc_da_blocks(struct inode *inode);
2540extern void ext4_set_aops(struct inode *inode);
2541extern int ext4_writepage_trans_blocks(struct inode *);
2542extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2543extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2544 loff_t lstart, loff_t lend);
2545extern int ext4_page_mkwrite(struct vm_fault *vmf);
2546extern int ext4_filemap_fault(struct vm_fault *vmf);
2547extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2548extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2549extern void ext4_da_update_reserve_space(struct inode *inode,
2550 int used, int quota_claim);
2551extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2552 ext4_fsblk_t pblk, ext4_lblk_t len);
2553extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2554 unsigned int map_len,
2555 struct extent_status *result);
2556
2557/* indirect.c */
2558extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2559 struct ext4_map_blocks *map, int flags);
2560extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2561extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2562extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2563extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2564 ext4_lblk_t start, ext4_lblk_t end);
2565
2566/* ioctl.c */
2567extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2568extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2569
2570/* migrate.c */
2571extern int ext4_ext_migrate(struct inode *);
2572extern int ext4_ind_migrate(struct inode *inode);
2573
2574/* namei.c */
2575extern int ext4_dirent_csum_verify(struct inode *inode,
2576 struct ext4_dir_entry *dirent);
2577extern int ext4_orphan_add(handle_t *, struct inode *);
2578extern int ext4_orphan_del(handle_t *, struct inode *);
2579extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2580 __u32 start_minor_hash, __u32 *next_hash);
2581extern int ext4_search_dir(struct buffer_head *bh,
2582 char *search_buf,
2583 int buf_size,
2584 struct inode *dir,
2585 struct ext4_filename *fname,
2586 unsigned int offset,
2587 struct ext4_dir_entry_2 **res_dir);
2588extern int ext4_generic_delete_entry(handle_t *handle,
2589 struct inode *dir,
2590 struct ext4_dir_entry_2 *de_del,
2591 struct buffer_head *bh,
2592 void *entry_buf,
2593 int buf_size,
2594 int csum_size);
2595extern bool ext4_empty_dir(struct inode *inode);
2596
2597/* resize.c */
2598extern void ext4_kvfree_array_rcu(void *to_free);
2599extern int ext4_group_add(struct super_block *sb,
2600 struct ext4_new_group_data *input);
2601extern int ext4_group_extend(struct super_block *sb,
2602 struct ext4_super_block *es,
2603 ext4_fsblk_t n_blocks_count);
2604extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2605
2606/* super.c */
2607extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2608 sector_t block, int op_flags);
2609extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2610extern int ext4_calculate_overhead(struct super_block *sb);
2611extern void ext4_superblock_csum_set(struct super_block *sb);
2612extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2613extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2614extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2615 ext4_group_t ngroup);
2616extern const char *ext4_decode_error(struct super_block *sb, int errno,
2617 char nbuf[16]);
2618
2619extern __printf(4, 5)
2620void __ext4_error(struct super_block *, const char *, unsigned int,
2621 const char *, ...);
2622extern __printf(5, 6)
2623void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2624 const char *, ...);
2625extern __printf(5, 6)
2626void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2627 const char *, ...);
2628extern void __ext4_std_error(struct super_block *, const char *,
2629 unsigned int, int);
2630extern __printf(4, 5)
2631void __ext4_abort(struct super_block *, const char *, unsigned int,
2632 const char *, ...);
2633extern __printf(4, 5)
2634void __ext4_warning(struct super_block *, const char *, unsigned int,
2635 const char *, ...);
2636extern __printf(4, 5)
2637void __ext4_warning_inode(const struct inode *inode, const char *function,
2638 unsigned int line, const char *fmt, ...);
2639extern __printf(3, 4)
2640void __ext4_msg(struct super_block *, const char *, const char *, ...);
2641extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2642 const char *, unsigned int, const char *);
2643extern __printf(7, 8)
2644void __ext4_grp_locked_error(const char *, unsigned int,
2645 struct super_block *, ext4_group_t,
2646 unsigned long, ext4_fsblk_t,
2647 const char *, ...);
2648
2649#define EXT4_ERROR_INODE(inode, fmt, a...) \
2650 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2651
2652#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2653 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2654
2655#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2656 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2657
2658#ifdef CONFIG_PRINTK
2659
2660#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2661 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2662#define ext4_error_file(file, func, line, block, fmt, ...) \
2663 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2664#define ext4_error(sb, fmt, ...) \
2665 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2666#define ext4_abort(sb, fmt, ...) \
2667 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2668#define ext4_warning(sb, fmt, ...) \
2669 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2670#define ext4_warning_inode(inode, fmt, ...) \
2671 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2672#define ext4_msg(sb, level, fmt, ...) \
2673 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2674#define dump_mmp_msg(sb, mmp, msg) \
2675 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2676#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2677 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2678 fmt, ##__VA_ARGS__)
2679
2680#else
2681
2682#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2683do { \
2684 no_printk(fmt, ##__VA_ARGS__); \
2685 __ext4_error_inode(inode, "", 0, block, " "); \
2686} while (0)
2687#define ext4_error_file(file, func, line, block, fmt, ...) \
2688do { \
2689 no_printk(fmt, ##__VA_ARGS__); \
2690 __ext4_error_file(file, "", 0, block, " "); \
2691} while (0)
2692#define ext4_error(sb, fmt, ...) \
2693do { \
2694 no_printk(fmt, ##__VA_ARGS__); \
2695 __ext4_error(sb, "", 0, " "); \
2696} while (0)
2697#define ext4_abort(sb, fmt, ...) \
2698do { \
2699 no_printk(fmt, ##__VA_ARGS__); \
2700 __ext4_abort(sb, "", 0, " "); \
2701} while (0)
2702#define ext4_warning(sb, fmt, ...) \
2703do { \
2704 no_printk(fmt, ##__VA_ARGS__); \
2705 __ext4_warning(sb, "", 0, " "); \
2706} while (0)
2707#define ext4_warning_inode(inode, fmt, ...) \
2708do { \
2709 no_printk(fmt, ##__VA_ARGS__); \
2710 __ext4_warning_inode(inode, "", 0, " "); \
2711} while (0)
2712#define ext4_msg(sb, level, fmt, ...) \
2713do { \
2714 no_printk(fmt, ##__VA_ARGS__); \
2715 __ext4_msg(sb, "", " "); \
2716} while (0)
2717#define dump_mmp_msg(sb, mmp, msg) \
2718 __dump_mmp_msg(sb, mmp, "", 0, "")
2719#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2720do { \
2721 no_printk(fmt, ##__VA_ARGS__); \
2722 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2723} while (0)
2724
2725#endif
2726
2727extern void ext4_update_dynamic_rev(struct super_block *sb);
2728extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2729 __u32 compat);
2730extern int ext4_update_rocompat_feature(handle_t *handle,
2731 struct super_block *sb, __u32 rocompat);
2732extern int ext4_update_incompat_feature(handle_t *handle,
2733 struct super_block *sb, __u32 incompat);
2734extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2735 struct ext4_group_desc *bg);
2736extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2737 struct ext4_group_desc *bg);
2738extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2739 struct ext4_group_desc *bg);
2740extern __u32 ext4_free_group_clusters(struct super_block *sb,
2741 struct ext4_group_desc *bg);
2742extern __u32 ext4_free_inodes_count(struct super_block *sb,
2743 struct ext4_group_desc *bg);
2744extern __u32 ext4_used_dirs_count(struct super_block *sb,
2745 struct ext4_group_desc *bg);
2746extern __u32 ext4_itable_unused_count(struct super_block *sb,
2747 struct ext4_group_desc *bg);
2748extern void ext4_block_bitmap_set(struct super_block *sb,
2749 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2750extern void ext4_inode_bitmap_set(struct super_block *sb,
2751 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2752extern void ext4_inode_table_set(struct super_block *sb,
2753 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2754extern void ext4_free_group_clusters_set(struct super_block *sb,
2755 struct ext4_group_desc *bg,
2756 __u32 count);
2757extern void ext4_free_inodes_set(struct super_block *sb,
2758 struct ext4_group_desc *bg, __u32 count);
2759extern void ext4_used_dirs_set(struct super_block *sb,
2760 struct ext4_group_desc *bg, __u32 count);
2761extern void ext4_itable_unused_set(struct super_block *sb,
2762 struct ext4_group_desc *bg, __u32 count);
2763extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2764 struct ext4_group_desc *gdp);
2765extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2766 struct ext4_group_desc *gdp);
2767extern int ext4_register_li_request(struct super_block *sb,
2768 ext4_group_t first_not_zeroed);
2769
2770static inline int ext4_has_metadata_csum(struct super_block *sb)
2771{
2772 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2773 !EXT4_SB(sb)->s_chksum_driver);
2774
2775 return ext4_has_feature_metadata_csum(sb) &&
2776 (EXT4_SB(sb)->s_chksum_driver != NULL);
2777}
2778
2779static inline int ext4_has_group_desc_csum(struct super_block *sb)
2780{
2781 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2782}
2783
2784static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2785{
2786 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2787 le32_to_cpu(es->s_blocks_count_lo);
2788}
2789
2790static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2791{
2792 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2793 le32_to_cpu(es->s_r_blocks_count_lo);
2794}
2795
2796static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2797{
2798 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2799 le32_to_cpu(es->s_free_blocks_count_lo);
2800}
2801
2802static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2803 ext4_fsblk_t blk)
2804{
2805 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2806 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2807}
2808
2809static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2810 ext4_fsblk_t blk)
2811{
2812 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2813 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2814}
2815
2816static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2817 ext4_fsblk_t blk)
2818{
2819 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2820 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2821}
2822
2823static inline loff_t ext4_isize(struct super_block *sb,
2824 struct ext4_inode *raw_inode)
2825{
2826 if (ext4_has_feature_largedir(sb) ||
2827 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2828 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2829 le32_to_cpu(raw_inode->i_size_lo);
2830
2831 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2832}
2833
2834static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2835{
2836 raw_inode->i_size_lo = cpu_to_le32(i_size);
2837 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2838}
2839
2840static inline
2841struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2842 ext4_group_t group)
2843{
2844 struct ext4_group_info **grp_info;
2845 long indexv, indexh;
2846 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2847 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2848 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2849 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
2850 return grp_info[indexh];
2851}
2852
2853/*
2854 * Reading s_groups_count requires using smp_rmb() afterwards. See
2855 * the locking protocol documented in the comments of ext4_group_add()
2856 * in resize.c
2857 */
2858static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2859{
2860 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2861
2862 smp_rmb();
2863 return ngroups;
2864}
2865
2866static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2867 ext4_group_t block_group)
2868{
2869 return block_group >> sbi->s_log_groups_per_flex;
2870}
2871
2872static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2873{
2874 return 1 << sbi->s_log_groups_per_flex;
2875}
2876
2877#define ext4_std_error(sb, errno) \
2878do { \
2879 if ((errno)) \
2880 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2881} while (0)
2882
2883#ifdef CONFIG_SMP
2884/* Each CPU can accumulate percpu_counter_batch clusters in their local
2885 * counters. So we need to make sure we have free clusters more
2886 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2887 */
2888#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2889#else
2890#define EXT4_FREECLUSTERS_WATERMARK 0
2891#endif
2892
2893/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2894static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2895{
2896 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2897 !inode_is_locked(inode));
2898 down_write(&EXT4_I(inode)->i_data_sem);
2899 if (newsize > EXT4_I(inode)->i_disksize)
2900 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
2901 up_write(&EXT4_I(inode)->i_data_sem);
2902}
2903
2904/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2905static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2906{
2907 int changed = 0;
2908
2909 if (newsize > inode->i_size) {
2910 i_size_write(inode, newsize);
2911 changed = 1;
2912 }
2913 if (newsize > EXT4_I(inode)->i_disksize) {
2914 ext4_update_i_disksize(inode, newsize);
2915 changed |= 2;
2916 }
2917 return changed;
2918}
2919
2920int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2921 loff_t len);
2922
2923struct ext4_group_info {
2924 unsigned long bb_state;
2925 struct rb_root bb_free_root;
2926 ext4_grpblk_t bb_first_free; /* first free block */
2927 ext4_grpblk_t bb_free; /* total free blocks */
2928 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2929 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2930 struct list_head bb_prealloc_list;
2931#ifdef DOUBLE_CHECK
2932 void *bb_bitmap;
2933#endif
2934 struct rw_semaphore alloc_sem;
2935 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2936 * regions, index is order.
2937 * bb_counters[3] = 5 means
2938 * 5 free 8-block regions. */
2939};
2940
2941#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2942#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2943#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2944#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2945
2946#define EXT4_MB_GRP_NEED_INIT(grp) \
2947 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2948#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2949 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2950#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2951 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2952
2953#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2954 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2955#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2956 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2957#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2958 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2959
2960#define EXT4_MAX_CONTENTION 8
2961#define EXT4_CONTENTION_THRESHOLD 2
2962
2963static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2964 ext4_group_t group)
2965{
2966 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2967}
2968
2969/*
2970 * Returns true if the filesystem is busy enough that attempts to
2971 * access the block group locks has run into contention.
2972 */
2973static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2974{
2975 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2976}
2977
2978static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2979{
2980 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2981 if (spin_trylock(lock))
2982 /*
2983 * We're able to grab the lock right away, so drop the
2984 * lock contention counter.
2985 */
2986 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2987 else {
2988 /*
2989 * The lock is busy, so bump the contention counter,
2990 * and then wait on the spin lock.
2991 */
2992 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2993 EXT4_MAX_CONTENTION);
2994 spin_lock(lock);
2995 }
2996}
2997
2998static inline void ext4_unlock_group(struct super_block *sb,
2999 ext4_group_t group)
3000{
3001 spin_unlock(ext4_group_lock_ptr(sb, group));
3002}
3003
3004/*
3005 * Block validity checking
3006 */
3007#define ext4_check_indirect_blockref(inode, bh) \
3008 ext4_check_blockref(__func__, __LINE__, inode, \
3009 (__le32 *)(bh)->b_data, \
3010 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3011
3012#define ext4_ind_check_inode(inode) \
3013 ext4_check_blockref(__func__, __LINE__, inode, \
3014 EXT4_I(inode)->i_data, \
3015 EXT4_NDIR_BLOCKS)
3016
3017/*
3018 * Inodes and files operations
3019 */
3020
3021/* dir.c */
3022extern const struct file_operations ext4_dir_operations;
3023
3024/* file.c */
3025extern const struct inode_operations ext4_file_inode_operations;
3026extern const struct file_operations ext4_file_operations;
3027extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3028
3029/* inline.c */
3030extern int ext4_get_max_inline_size(struct inode *inode);
3031extern int ext4_find_inline_data_nolock(struct inode *inode);
3032extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3033 unsigned int len);
3034extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3035
3036extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3037extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3038 struct inode *inode,
3039 loff_t pos, unsigned len,
3040 unsigned flags,
3041 struct page **pagep);
3042extern int ext4_write_inline_data_end(struct inode *inode,
3043 loff_t pos, unsigned len,
3044 unsigned copied,
3045 struct page *page);
3046extern struct buffer_head *
3047ext4_journalled_write_inline_data(struct inode *inode,
3048 unsigned len,
3049 struct page *page);
3050extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3051 struct inode *inode,
3052 loff_t pos, unsigned len,
3053 unsigned flags,
3054 struct page **pagep,
3055 void **fsdata);
3056extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3057 unsigned len, unsigned copied,
3058 struct page *page);
3059extern int ext4_try_add_inline_entry(handle_t *handle,
3060 struct ext4_filename *fname,
3061 struct inode *dir, struct inode *inode);
3062extern int ext4_try_create_inline_dir(handle_t *handle,
3063 struct inode *parent,
3064 struct inode *inode);
3065extern int ext4_read_inline_dir(struct file *filp,
3066 struct dir_context *ctx,
3067 int *has_inline_data);
3068extern int htree_inlinedir_to_tree(struct file *dir_file,
3069 struct inode *dir, ext4_lblk_t block,
3070 struct dx_hash_info *hinfo,
3071 __u32 start_hash, __u32 start_minor_hash,
3072 int *has_inline_data);
3073extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3074 struct ext4_filename *fname,
3075 struct ext4_dir_entry_2 **res_dir,
3076 int *has_inline_data);
3077extern int ext4_delete_inline_entry(handle_t *handle,
3078 struct inode *dir,
3079 struct ext4_dir_entry_2 *de_del,
3080 struct buffer_head *bh,
3081 int *has_inline_data);
3082extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3083extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3084 struct ext4_dir_entry_2 **parent_de,
3085 int *retval);
3086extern int ext4_inline_data_fiemap(struct inode *inode,
3087 struct fiemap_extent_info *fieinfo,
3088 int *has_inline, __u64 start, __u64 len);
3089extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3090
3091extern int ext4_convert_inline_data(struct inode *inode);
3092
3093static inline int ext4_has_inline_data(struct inode *inode)
3094{
3095 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3096 EXT4_I(inode)->i_inline_off;
3097}
3098
3099/* namei.c */
3100extern const struct inode_operations ext4_dir_inode_operations;
3101extern const struct inode_operations ext4_special_inode_operations;
3102extern struct dentry *ext4_get_parent(struct dentry *child);
3103extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3104 struct ext4_dir_entry_2 *de,
3105 int blocksize, int csum_size,
3106 unsigned int parent_ino, int dotdot_real_len);
3107extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3108 unsigned int blocksize);
3109extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3110 struct inode *inode,
3111 struct buffer_head *bh);
3112#define S_SHIFT 12
3113static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3114 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3115 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3116 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3117 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3118 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3119 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3120 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3121};
3122
3123static inline void ext4_set_de_type(struct super_block *sb,
3124 struct ext4_dir_entry_2 *de,
3125 umode_t mode) {
3126 if (ext4_has_feature_filetype(sb))
3127 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3128}
3129
3130/* readpages.c */
3131extern int ext4_mpage_readpages(struct address_space *mapping,
3132 struct list_head *pages, struct page *page,
3133 unsigned nr_pages);
3134
3135/* symlink.c */
3136extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3137extern const struct inode_operations ext4_symlink_inode_operations;
3138extern const struct inode_operations ext4_fast_symlink_inode_operations;
3139
3140/* sysfs.c */
3141extern int ext4_register_sysfs(struct super_block *sb);
3142extern void ext4_unregister_sysfs(struct super_block *sb);
3143extern int __init ext4_init_sysfs(void);
3144extern void ext4_exit_sysfs(void);
3145
3146/* block_validity */
3147extern void ext4_release_system_zone(struct super_block *sb);
3148extern int ext4_setup_system_zone(struct super_block *sb);
3149extern int __init ext4_init_system_zone(void);
3150extern void ext4_exit_system_zone(void);
3151extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3152 ext4_fsblk_t start_blk,
3153 unsigned int count);
3154extern int ext4_check_blockref(const char *, unsigned int,
3155 struct inode *, __le32 *, unsigned int);
3156
3157/* extents.c */
3158struct ext4_ext_path;
3159struct ext4_extent;
3160
3161/*
3162 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3163 * __le32.
3164 */
3165#define EXT_MAX_BLOCKS 0xffffffff
3166
3167extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3168extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3169extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3170extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3171 struct ext4_map_blocks *map, int flags);
3172extern int ext4_ext_truncate(handle_t *, struct inode *);
3173extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3174 ext4_lblk_t end);
3175extern void ext4_ext_init(struct super_block *);
3176extern void ext4_ext_release(struct super_block *);
3177extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3178 loff_t len);
3179extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3180 loff_t offset, ssize_t len);
3181extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3182 struct ext4_map_blocks *map, int flags);
3183extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3184 ext4_lblk_t lblocks);
3185extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3186 int num,
3187 struct ext4_ext_path *path);
3188extern int ext4_can_extents_be_merged(struct inode *inode,
3189 struct ext4_extent *ex1,
3190 struct ext4_extent *ex2);
3191extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3192 struct ext4_ext_path **,
3193 struct ext4_extent *, int);
3194extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3195 struct ext4_ext_path **,
3196 int flags);
3197extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3198extern int ext4_ext_check_inode(struct inode *inode);
3199extern int ext4_find_delalloc_range(struct inode *inode,
3200 ext4_lblk_t lblk_start,
3201 ext4_lblk_t lblk_end);
3202extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3203extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3204extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3205 __u64 start, __u64 len);
3206extern int ext4_ext_precache(struct inode *inode);
3207extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3208extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3209extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3210 struct inode *inode2, ext4_lblk_t lblk1,
3211 ext4_lblk_t lblk2, ext4_lblk_t count,
3212 int mark_unwritten,int *err);
3213
3214/* move_extent.c */
3215extern void ext4_double_down_write_data_sem(struct inode *first,
3216 struct inode *second);
3217extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3218 struct inode *donor_inode);
3219extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3220 __u64 start_orig, __u64 start_donor,
3221 __u64 len, __u64 *moved_len);
3222
3223/* page-io.c */
3224extern int __init ext4_init_pageio(void);
3225extern void ext4_exit_pageio(void);
3226extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3227extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3228extern int ext4_put_io_end(ext4_io_end_t *io_end);
3229extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3230extern void ext4_io_submit_init(struct ext4_io_submit *io,
3231 struct writeback_control *wbc);
3232extern void ext4_end_io_rsv_work(struct work_struct *work);
3233extern void ext4_io_submit(struct ext4_io_submit *io);
3234extern int ext4_bio_write_page(struct ext4_io_submit *io,
3235 struct page *page,
3236 int len,
3237 struct writeback_control *wbc,
3238 bool keep_towrite);
3239
3240/* mmp.c */
3241extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3242
3243/*
3244 * Add new method to test whether block and inode bitmaps are properly
3245 * initialized. With uninit_bg reading the block from disk is not enough
3246 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3247 */
3248#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3249
3250static inline int bitmap_uptodate(struct buffer_head *bh)
3251{
3252 return (buffer_uptodate(bh) &&
3253 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3254}
3255static inline void set_bitmap_uptodate(struct buffer_head *bh)
3256{
3257 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3258}
3259
3260/*
3261 * Disable DIO read nolock optimization, so new dioreaders will be forced
3262 * to grab i_mutex
3263 */
3264static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3265{
3266 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3267 smp_mb();
3268}
3269static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3270{
3271 smp_mb();
3272 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3273}
3274
3275#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3276
3277/* For ioend & aio unwritten conversion wait queues */
3278#define EXT4_WQ_HASH_SZ 37
3279#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3280 EXT4_WQ_HASH_SZ])
3281extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3282
3283extern int ext4_resize_begin(struct super_block *sb);
3284extern void ext4_resize_end(struct super_block *sb);
3285
3286static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3287 struct ext4_io_end *io_end)
3288{
3289 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3290 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3291 atomic_inc(&EXT4_I(inode)->i_unwritten);
3292 }
3293}
3294
3295static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3296{
3297 struct inode *inode = io_end->inode;
3298
3299 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3300 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3301 /* Wake up anyone waiting on unwritten extent conversion */
3302 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3303 wake_up_all(ext4_ioend_wq(inode));
3304 }
3305}
3306
3307extern const struct iomap_ops ext4_iomap_ops;
3308
3309#endif /* __KERNEL__ */
3310
3311#define EFSBADCRC EBADMSG /* Bad CRC detected */
3312#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3313
3314#endif /* _EXT4_H */