Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1/*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16#ifndef _EXT4_H
17#define _EXT4_H
18
19#include <linux/types.h>
20#include <linux/blkdev.h>
21#include <linux/magic.h>
22#include <linux/jbd2.h>
23#include <linux/quota.h>
24#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
28#include <linux/timer.h>
29#include <linux/version.h>
30#include <linux/wait.h>
31#include <linux/blockgroup_lock.h>
32#include <linux/percpu_counter.h>
33#include <linux/ratelimit.h>
34#include <crypto/hash.h>
35#include <linux/fscrypto.h>
36#include <linux/falloc.h>
37#include <linux/percpu-rwsem.h>
38#ifdef __KERNEL__
39#include <linux/compat.h>
40#endif
41
42/*
43 * The fourth extended filesystem constants/structures
44 */
45
46/*
47 * with AGGRESSIVE_CHECK allocator runs consistency checks over
48 * structures. these checks slow things down a lot
49 */
50#define AGGRESSIVE_CHECK__
51
52/*
53 * with DOUBLE_CHECK defined mballoc creates persistent in-core
54 * bitmaps, maintains and uses them to check for double allocations
55 */
56#define DOUBLE_CHECK__
57
58/*
59 * Define EXT4FS_DEBUG to produce debug messages
60 */
61#undef EXT4FS_DEBUG
62
63/*
64 * Debug code
65 */
66#ifdef EXT4FS_DEBUG
67#define ext4_debug(f, a...) \
68 do { \
69 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
70 __FILE__, __LINE__, __func__); \
71 printk(KERN_DEBUG f, ## a); \
72 } while (0)
73#else
74#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
75#endif
76
77/*
78 * Turn on EXT_DEBUG to get lots of info about extents operations.
79 */
80#define EXT_DEBUG__
81#ifdef EXT_DEBUG
82#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
83#else
84#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
85#endif
86
87/* data type for block offset of block group */
88typedef int ext4_grpblk_t;
89
90/* data type for filesystem-wide blocks number */
91typedef unsigned long long ext4_fsblk_t;
92
93/* data type for file logical block number */
94typedef __u32 ext4_lblk_t;
95
96/* data type for block group number */
97typedef unsigned int ext4_group_t;
98
99enum SHIFT_DIRECTION {
100 SHIFT_LEFT = 0,
101 SHIFT_RIGHT,
102};
103
104/*
105 * Flags used in mballoc's allocation_context flags field.
106 *
107 * Also used to show what's going on for debugging purposes when the
108 * flag field is exported via the traceport interface
109 */
110
111/* prefer goal again. length */
112#define EXT4_MB_HINT_MERGE 0x0001
113/* blocks already reserved */
114#define EXT4_MB_HINT_RESERVED 0x0002
115/* metadata is being allocated */
116#define EXT4_MB_HINT_METADATA 0x0004
117/* first blocks in the file */
118#define EXT4_MB_HINT_FIRST 0x0008
119/* search for the best chunk */
120#define EXT4_MB_HINT_BEST 0x0010
121/* data is being allocated */
122#define EXT4_MB_HINT_DATA 0x0020
123/* don't preallocate (for tails) */
124#define EXT4_MB_HINT_NOPREALLOC 0x0040
125/* allocate for locality group */
126#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
127/* allocate goal blocks or none */
128#define EXT4_MB_HINT_GOAL_ONLY 0x0100
129/* goal is meaningful */
130#define EXT4_MB_HINT_TRY_GOAL 0x0200
131/* blocks already pre-reserved by delayed allocation */
132#define EXT4_MB_DELALLOC_RESERVED 0x0400
133/* We are doing stream allocation */
134#define EXT4_MB_STREAM_ALLOC 0x0800
135/* Use reserved root blocks if needed */
136#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
137/* Use blocks from reserved pool */
138#define EXT4_MB_USE_RESERVED 0x2000
139
140struct ext4_allocation_request {
141 /* target inode for block we're allocating */
142 struct inode *inode;
143 /* how many blocks we want to allocate */
144 unsigned int len;
145 /* logical block in target inode */
146 ext4_lblk_t logical;
147 /* the closest logical allocated block to the left */
148 ext4_lblk_t lleft;
149 /* the closest logical allocated block to the right */
150 ext4_lblk_t lright;
151 /* phys. target (a hint) */
152 ext4_fsblk_t goal;
153 /* phys. block for the closest logical allocated block to the left */
154 ext4_fsblk_t pleft;
155 /* phys. block for the closest logical allocated block to the right */
156 ext4_fsblk_t pright;
157 /* flags. see above EXT4_MB_HINT_* */
158 unsigned int flags;
159};
160
161/*
162 * Logical to physical block mapping, used by ext4_map_blocks()
163 *
164 * This structure is used to pass requests into ext4_map_blocks() as
165 * well as to store the information returned by ext4_map_blocks(). It
166 * takes less room on the stack than a struct buffer_head.
167 */
168#define EXT4_MAP_NEW (1 << BH_New)
169#define EXT4_MAP_MAPPED (1 << BH_Mapped)
170#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
171#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
172#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
173 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
174
175struct ext4_map_blocks {
176 ext4_fsblk_t m_pblk;
177 ext4_lblk_t m_lblk;
178 unsigned int m_len;
179 unsigned int m_flags;
180};
181
182/*
183 * Flags for ext4_io_end->flags
184 */
185#define EXT4_IO_END_UNWRITTEN 0x0001
186
187/*
188 * For converting unwritten extents on a work queue. 'handle' is used for
189 * buffered writeback.
190 */
191typedef struct ext4_io_end {
192 struct list_head list; /* per-file finished IO list */
193 handle_t *handle; /* handle reserved for extent
194 * conversion */
195 struct inode *inode; /* file being written to */
196 struct bio *bio; /* Linked list of completed
197 * bios covering the extent */
198 unsigned int flag; /* unwritten or not */
199 atomic_t count; /* reference counter */
200 loff_t offset; /* offset in the file */
201 ssize_t size; /* size of the extent */
202} ext4_io_end_t;
203
204struct ext4_io_submit {
205 struct writeback_control *io_wbc;
206 struct bio *io_bio;
207 ext4_io_end_t *io_end;
208 sector_t io_next_block;
209};
210
211/*
212 * Special inodes numbers
213 */
214#define EXT4_BAD_INO 1 /* Bad blocks inode */
215#define EXT4_ROOT_INO 2 /* Root inode */
216#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
217#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
218#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
219#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
220#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
221#define EXT4_JOURNAL_INO 8 /* Journal inode */
222
223/* First non-reserved inode for old ext4 filesystems */
224#define EXT4_GOOD_OLD_FIRST_INO 11
225
226/*
227 * Maximal count of links to a file
228 */
229#define EXT4_LINK_MAX 65000
230
231/*
232 * Macro-instructions used to manage several block sizes
233 */
234#define EXT4_MIN_BLOCK_SIZE 1024
235#define EXT4_MAX_BLOCK_SIZE 65536
236#define EXT4_MIN_BLOCK_LOG_SIZE 10
237#define EXT4_MAX_BLOCK_LOG_SIZE 16
238#ifdef __KERNEL__
239# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
240#else
241# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
242#endif
243#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
244#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
245 EXT4_SB(s)->s_cluster_bits)
246#ifdef __KERNEL__
247# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
248# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
249#else
250# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
251#endif
252#ifdef __KERNEL__
253#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
254#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
255#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
256#else
257#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
258 EXT4_GOOD_OLD_INODE_SIZE : \
259 (s)->s_inode_size)
260#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
261 EXT4_GOOD_OLD_FIRST_INO : \
262 (s)->s_first_ino)
263#endif
264#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
265
266/* Translate a block number to a cluster number */
267#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
268/* Translate a cluster number to a block number */
269#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
270/* Translate # of blks to # of clusters */
271#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
272 (sbi)->s_cluster_bits)
273/* Mask out the low bits to get the starting block of the cluster */
274#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
275 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
276#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
277 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
278/* Get the cluster offset */
279#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
280 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
281#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
282 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
283
284/*
285 * Structure of a blocks group descriptor
286 */
287struct ext4_group_desc
288{
289 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
290 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
291 __le32 bg_inode_table_lo; /* Inodes table block */
292 __le16 bg_free_blocks_count_lo;/* Free blocks count */
293 __le16 bg_free_inodes_count_lo;/* Free inodes count */
294 __le16 bg_used_dirs_count_lo; /* Directories count */
295 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
296 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
297 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
298 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
299 __le16 bg_itable_unused_lo; /* Unused inodes count */
300 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
301 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
302 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
303 __le32 bg_inode_table_hi; /* Inodes table block MSB */
304 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
305 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
306 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
307 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
308 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
309 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
310 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
311 __u32 bg_reserved;
312};
313
314#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
315 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
316 sizeof(__le16))
317#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
318 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
319 sizeof(__le16))
320
321/*
322 * Structure of a flex block group info
323 */
324
325struct flex_groups {
326 atomic64_t free_clusters;
327 atomic_t free_inodes;
328 atomic_t used_dirs;
329};
330
331#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
332#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
333#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
334
335/*
336 * Macro-instructions used to manage group descriptors
337 */
338#define EXT4_MIN_DESC_SIZE 32
339#define EXT4_MIN_DESC_SIZE_64BIT 64
340#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
341#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
342#ifdef __KERNEL__
343# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
344# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
345# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
346# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
347# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
348#else
349# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
350# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
351# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
352#endif
353
354/*
355 * Constants relative to the data blocks
356 */
357#define EXT4_NDIR_BLOCKS 12
358#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
359#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
360#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
361#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
362
363/*
364 * Inode flags
365 */
366#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
367#define EXT4_UNRM_FL 0x00000002 /* Undelete */
368#define EXT4_COMPR_FL 0x00000004 /* Compress file */
369#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
370#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
371#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
372#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
373#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
374/* Reserved for compression usage... */
375#define EXT4_DIRTY_FL 0x00000100
376#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
377#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
378 /* nb: was previously EXT2_ECOMPR_FL */
379#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
380/* End compression flags --- maybe not all used */
381#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
382#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
383#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
384#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
385#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
386#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
387#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
388#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
389#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
390#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
391#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
392#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
393#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
394
395#define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
396#define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
397
398#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
399 EXT4_IMMUTABLE_FL | \
400 EXT4_APPEND_FL | \
401 EXT4_NODUMP_FL | \
402 EXT4_NOATIME_FL | \
403 EXT4_PROJINHERIT_FL)
404
405/* Flags that should be inherited by new inodes from their parent. */
406#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
407 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
408 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
409 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
410 EXT4_PROJINHERIT_FL)
411
412/* Flags that are appropriate for regular files (all but dir-specific ones). */
413#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
414
415/* Flags that are appropriate for non-directories/regular files. */
416#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
417
418/* Mask out flags that are inappropriate for the given type of inode. */
419static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
420{
421 if (S_ISDIR(mode))
422 return flags;
423 else if (S_ISREG(mode))
424 return flags & EXT4_REG_FLMASK;
425 else
426 return flags & EXT4_OTHER_FLMASK;
427}
428
429/*
430 * Inode flags used for atomic set/get
431 */
432enum {
433 EXT4_INODE_SECRM = 0, /* Secure deletion */
434 EXT4_INODE_UNRM = 1, /* Undelete */
435 EXT4_INODE_COMPR = 2, /* Compress file */
436 EXT4_INODE_SYNC = 3, /* Synchronous updates */
437 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
438 EXT4_INODE_APPEND = 5, /* writes to file may only append */
439 EXT4_INODE_NODUMP = 6, /* do not dump file */
440 EXT4_INODE_NOATIME = 7, /* do not update atime */
441/* Reserved for compression usage... */
442 EXT4_INODE_DIRTY = 8,
443 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
444 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
445 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
446/* End compression flags --- maybe not all used */
447 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
448 EXT4_INODE_IMAGIC = 13, /* AFS directory */
449 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
450 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
451 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
452 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
453 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
454 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
455 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
456 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
457 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
458 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
459 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
460};
461
462/*
463 * Since it's pretty easy to mix up bit numbers and hex values, we use a
464 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
465 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
466 * any extra space in the compiled kernel image, otherwise, the build will fail.
467 * It's important that these values are the same, since we are using
468 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
469 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
470 * values found in ext2, ext3 and ext4 filesystems, and of course the values
471 * defined in e2fsprogs.
472 *
473 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
474 */
475#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
476#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
477
478static inline void ext4_check_flag_values(void)
479{
480 CHECK_FLAG_VALUE(SECRM);
481 CHECK_FLAG_VALUE(UNRM);
482 CHECK_FLAG_VALUE(COMPR);
483 CHECK_FLAG_VALUE(SYNC);
484 CHECK_FLAG_VALUE(IMMUTABLE);
485 CHECK_FLAG_VALUE(APPEND);
486 CHECK_FLAG_VALUE(NODUMP);
487 CHECK_FLAG_VALUE(NOATIME);
488 CHECK_FLAG_VALUE(DIRTY);
489 CHECK_FLAG_VALUE(COMPRBLK);
490 CHECK_FLAG_VALUE(NOCOMPR);
491 CHECK_FLAG_VALUE(ENCRYPT);
492 CHECK_FLAG_VALUE(INDEX);
493 CHECK_FLAG_VALUE(IMAGIC);
494 CHECK_FLAG_VALUE(JOURNAL_DATA);
495 CHECK_FLAG_VALUE(NOTAIL);
496 CHECK_FLAG_VALUE(DIRSYNC);
497 CHECK_FLAG_VALUE(TOPDIR);
498 CHECK_FLAG_VALUE(HUGE_FILE);
499 CHECK_FLAG_VALUE(EXTENTS);
500 CHECK_FLAG_VALUE(EA_INODE);
501 CHECK_FLAG_VALUE(EOFBLOCKS);
502 CHECK_FLAG_VALUE(INLINE_DATA);
503 CHECK_FLAG_VALUE(PROJINHERIT);
504 CHECK_FLAG_VALUE(RESERVED);
505}
506
507/* Used to pass group descriptor data when online resize is done */
508struct ext4_new_group_input {
509 __u32 group; /* Group number for this data */
510 __u64 block_bitmap; /* Absolute block number of block bitmap */
511 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
512 __u64 inode_table; /* Absolute block number of inode table start */
513 __u32 blocks_count; /* Total number of blocks in this group */
514 __u16 reserved_blocks; /* Number of reserved blocks in this group */
515 __u16 unused;
516};
517
518#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
519struct compat_ext4_new_group_input {
520 u32 group;
521 compat_u64 block_bitmap;
522 compat_u64 inode_bitmap;
523 compat_u64 inode_table;
524 u32 blocks_count;
525 u16 reserved_blocks;
526 u16 unused;
527};
528#endif
529
530/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
531struct ext4_new_group_data {
532 __u32 group;
533 __u64 block_bitmap;
534 __u64 inode_bitmap;
535 __u64 inode_table;
536 __u32 blocks_count;
537 __u16 reserved_blocks;
538 __u16 unused;
539 __u32 free_blocks_count;
540};
541
542/* Indexes used to index group tables in ext4_new_group_data */
543enum {
544 BLOCK_BITMAP = 0, /* block bitmap */
545 INODE_BITMAP, /* inode bitmap */
546 INODE_TABLE, /* inode tables */
547 GROUP_TABLE_COUNT,
548};
549
550/*
551 * Flags used by ext4_map_blocks()
552 */
553 /* Allocate any needed blocks and/or convert an unwritten
554 extent to be an initialized ext4 */
555#define EXT4_GET_BLOCKS_CREATE 0x0001
556 /* Request the creation of an unwritten extent */
557#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
558#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
559 EXT4_GET_BLOCKS_CREATE)
560 /* Caller is from the delayed allocation writeout path
561 * finally doing the actual allocation of delayed blocks */
562#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
563 /* caller is from the direct IO path, request to creation of an
564 unwritten extents if not allocated, split the unwritten
565 extent if blocks has been preallocated already*/
566#define EXT4_GET_BLOCKS_PRE_IO 0x0008
567#define EXT4_GET_BLOCKS_CONVERT 0x0010
568#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
569 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
570 /* Convert extent to initialized after IO complete */
571#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
572 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
573 /* Eventual metadata allocation (due to growing extent tree)
574 * should not fail, so try to use reserved blocks for that.*/
575#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
576 /* Don't normalize allocation size (used for fallocate) */
577#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
578 /* Request will not result in inode size update (user for fallocate) */
579#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
580 /* Convert written extents to unwritten */
581#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
582 /* Write zeros to newly created written extents */
583#define EXT4_GET_BLOCKS_ZERO 0x0200
584#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
585 EXT4_GET_BLOCKS_ZERO)
586 /* Caller will submit data before dropping transaction handle. This
587 * allows jbd2 to avoid submitting data before commit. */
588#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
589
590/*
591 * The bit position of these flags must not overlap with any of the
592 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
593 * read_extent_tree_block(), ext4_split_extent_at(),
594 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
595 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
596 * caching the extents when reading from the extent tree while a
597 * truncate or punch hole operation is in progress.
598 */
599#define EXT4_EX_NOCACHE 0x40000000
600#define EXT4_EX_FORCE_CACHE 0x20000000
601
602/*
603 * Flags used by ext4_free_blocks
604 */
605#define EXT4_FREE_BLOCKS_METADATA 0x0001
606#define EXT4_FREE_BLOCKS_FORGET 0x0002
607#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
608#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
609#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
610#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
611
612/*
613 * ioctl commands
614 */
615#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
616#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
617#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
618#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
619#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
620#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
621#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
622#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
623#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
624#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
625#define EXT4_IOC_MIGRATE _IO('f', 9)
626 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
627 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
628#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
629#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
630#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
631#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
632#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
633#define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
634#define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
635#define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
636
637#ifndef FS_IOC_FSGETXATTR
638/* Until the uapi changes get merged for project quota... */
639
640#define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
641#define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
642
643/*
644 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
645 */
646struct fsxattr {
647 __u32 fsx_xflags; /* xflags field value (get/set) */
648 __u32 fsx_extsize; /* extsize field value (get/set)*/
649 __u32 fsx_nextents; /* nextents field value (get) */
650 __u32 fsx_projid; /* project identifier (get/set) */
651 unsigned char fsx_pad[12];
652};
653
654/*
655 * Flags for the fsx_xflags field
656 */
657#define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
658#define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
659#define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
660#define FS_XFLAG_APPEND 0x00000010 /* all writes append */
661#define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
662#define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
663#define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
664#define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
665#define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
666#define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
667#define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
668#define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
669#define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
670#define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
671#define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
672#endif /* !defined(FS_IOC_FSGETXATTR) */
673
674#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
675#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
676
677#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
678/*
679 * ioctl commands in 32 bit emulation
680 */
681#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
682#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
683#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
684#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
685#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
686#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
687#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
688#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
689#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
690#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
691#endif
692
693/* Max physical block we can address w/o extents */
694#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
695
696/*
697 * Structure of an inode on the disk
698 */
699struct ext4_inode {
700 __le16 i_mode; /* File mode */
701 __le16 i_uid; /* Low 16 bits of Owner Uid */
702 __le32 i_size_lo; /* Size in bytes */
703 __le32 i_atime; /* Access time */
704 __le32 i_ctime; /* Inode Change time */
705 __le32 i_mtime; /* Modification time */
706 __le32 i_dtime; /* Deletion Time */
707 __le16 i_gid; /* Low 16 bits of Group Id */
708 __le16 i_links_count; /* Links count */
709 __le32 i_blocks_lo; /* Blocks count */
710 __le32 i_flags; /* File flags */
711 union {
712 struct {
713 __le32 l_i_version;
714 } linux1;
715 struct {
716 __u32 h_i_translator;
717 } hurd1;
718 struct {
719 __u32 m_i_reserved1;
720 } masix1;
721 } osd1; /* OS dependent 1 */
722 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
723 __le32 i_generation; /* File version (for NFS) */
724 __le32 i_file_acl_lo; /* File ACL */
725 __le32 i_size_high;
726 __le32 i_obso_faddr; /* Obsoleted fragment address */
727 union {
728 struct {
729 __le16 l_i_blocks_high; /* were l_i_reserved1 */
730 __le16 l_i_file_acl_high;
731 __le16 l_i_uid_high; /* these 2 fields */
732 __le16 l_i_gid_high; /* were reserved2[0] */
733 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
734 __le16 l_i_reserved;
735 } linux2;
736 struct {
737 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
738 __u16 h_i_mode_high;
739 __u16 h_i_uid_high;
740 __u16 h_i_gid_high;
741 __u32 h_i_author;
742 } hurd2;
743 struct {
744 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
745 __le16 m_i_file_acl_high;
746 __u32 m_i_reserved2[2];
747 } masix2;
748 } osd2; /* OS dependent 2 */
749 __le16 i_extra_isize;
750 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
751 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
752 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
753 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
754 __le32 i_crtime; /* File Creation time */
755 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
756 __le32 i_version_hi; /* high 32 bits for 64-bit version */
757 __le32 i_projid; /* Project ID */
758};
759
760struct move_extent {
761 __u32 reserved; /* should be zero */
762 __u32 donor_fd; /* donor file descriptor */
763 __u64 orig_start; /* logical start offset in block for orig */
764 __u64 donor_start; /* logical start offset in block for donor */
765 __u64 len; /* block length to be moved */
766 __u64 moved_len; /* moved block length */
767};
768
769#define EXT4_EPOCH_BITS 2
770#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
771#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
772
773/*
774 * Extended fields will fit into an inode if the filesystem was formatted
775 * with large inodes (-I 256 or larger) and there are not currently any EAs
776 * consuming all of the available space. For new inodes we always reserve
777 * enough space for the kernel's known extended fields, but for inodes
778 * created with an old kernel this might not have been the case. None of
779 * the extended inode fields is critical for correct filesystem operation.
780 * This macro checks if a certain field fits in the inode. Note that
781 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
782 */
783#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
784 ((offsetof(typeof(*ext4_inode), field) + \
785 sizeof((ext4_inode)->field)) \
786 <= (EXT4_GOOD_OLD_INODE_SIZE + \
787 (einode)->i_extra_isize)) \
788
789/*
790 * We use an encoding that preserves the times for extra epoch "00":
791 *
792 * extra msb of adjust for signed
793 * epoch 32-bit 32-bit tv_sec to
794 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
795 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
796 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
797 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
798 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
799 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
800 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
801 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
802 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
803 *
804 * Note that previous versions of the kernel on 64-bit systems would
805 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
806 * 1970. e2fsck will correct this, assuming that it is run on the
807 * affected filesystem before 2242.
808 */
809
810static inline __le32 ext4_encode_extra_time(struct timespec *time)
811{
812 u32 extra = sizeof(time->tv_sec) > 4 ?
813 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
814 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
815}
816
817static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
818{
819 if (unlikely(sizeof(time->tv_sec) > 4 &&
820 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
821#if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
822 /* Handle legacy encoding of pre-1970 dates with epoch
823 * bits 1,1. We assume that by kernel version 4.20,
824 * everyone will have run fsck over the affected
825 * filesystems to correct the problem. (This
826 * backwards compatibility may be removed before this
827 * time, at the discretion of the ext4 developers.)
828 */
829 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
830 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
831 extra_bits = 0;
832 time->tv_sec += extra_bits << 32;
833#else
834 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
835#endif
836 }
837 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
838}
839
840#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
841do { \
842 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
843 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
844 (raw_inode)->xtime ## _extra = \
845 ext4_encode_extra_time(&(inode)->xtime); \
846} while (0)
847
848#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
849do { \
850 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
851 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
852 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
853 (raw_inode)->xtime ## _extra = \
854 ext4_encode_extra_time(&(einode)->xtime); \
855} while (0)
856
857#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
858do { \
859 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
860 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
861 ext4_decode_extra_time(&(inode)->xtime, \
862 raw_inode->xtime ## _extra); \
863 else \
864 (inode)->xtime.tv_nsec = 0; \
865} while (0)
866
867#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
868do { \
869 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
870 (einode)->xtime.tv_sec = \
871 (signed)le32_to_cpu((raw_inode)->xtime); \
872 else \
873 (einode)->xtime.tv_sec = 0; \
874 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
875 ext4_decode_extra_time(&(einode)->xtime, \
876 raw_inode->xtime ## _extra); \
877 else \
878 (einode)->xtime.tv_nsec = 0; \
879} while (0)
880
881#define i_disk_version osd1.linux1.l_i_version
882
883#if defined(__KERNEL__) || defined(__linux__)
884#define i_reserved1 osd1.linux1.l_i_reserved1
885#define i_file_acl_high osd2.linux2.l_i_file_acl_high
886#define i_blocks_high osd2.linux2.l_i_blocks_high
887#define i_uid_low i_uid
888#define i_gid_low i_gid
889#define i_uid_high osd2.linux2.l_i_uid_high
890#define i_gid_high osd2.linux2.l_i_gid_high
891#define i_checksum_lo osd2.linux2.l_i_checksum_lo
892
893#elif defined(__GNU__)
894
895#define i_translator osd1.hurd1.h_i_translator
896#define i_uid_high osd2.hurd2.h_i_uid_high
897#define i_gid_high osd2.hurd2.h_i_gid_high
898#define i_author osd2.hurd2.h_i_author
899
900#elif defined(__masix__)
901
902#define i_reserved1 osd1.masix1.m_i_reserved1
903#define i_file_acl_high osd2.masix2.m_i_file_acl_high
904#define i_reserved2 osd2.masix2.m_i_reserved2
905
906#endif /* defined(__KERNEL__) || defined(__linux__) */
907
908#include "extents_status.h"
909
910/*
911 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
912 *
913 * These are needed to avoid lockdep false positives when we need to
914 * allocate blocks to the quota inode during ext4_map_blocks(), while
915 * holding i_data_sem for a normal (non-quota) inode. Since we don't
916 * do quota tracking for the quota inode, this avoids deadlock (as
917 * well as infinite recursion, since it isn't turtles all the way
918 * down...)
919 *
920 * I_DATA_SEM_NORMAL - Used for most inodes
921 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
922 * where the second inode has larger inode number
923 * than the first
924 * I_DATA_SEM_QUOTA - Used for quota inodes only
925 */
926enum {
927 I_DATA_SEM_NORMAL = 0,
928 I_DATA_SEM_OTHER,
929 I_DATA_SEM_QUOTA,
930};
931
932
933/*
934 * fourth extended file system inode data in memory
935 */
936struct ext4_inode_info {
937 __le32 i_data[15]; /* unconverted */
938 __u32 i_dtime;
939 ext4_fsblk_t i_file_acl;
940
941 /*
942 * i_block_group is the number of the block group which contains
943 * this file's inode. Constant across the lifetime of the inode,
944 * it is ued for making block allocation decisions - we try to
945 * place a file's data blocks near its inode block, and new inodes
946 * near to their parent directory's inode.
947 */
948 ext4_group_t i_block_group;
949 ext4_lblk_t i_dir_start_lookup;
950#if (BITS_PER_LONG < 64)
951 unsigned long i_state_flags; /* Dynamic state flags */
952#endif
953 unsigned long i_flags;
954
955 /*
956 * Extended attributes can be read independently of the main file
957 * data. Taking i_mutex even when reading would cause contention
958 * between readers of EAs and writers of regular file data, so
959 * instead we synchronize on xattr_sem when reading or changing
960 * EAs.
961 */
962 struct rw_semaphore xattr_sem;
963
964 struct list_head i_orphan; /* unlinked but open inodes */
965
966 /*
967 * i_disksize keeps track of what the inode size is ON DISK, not
968 * in memory. During truncate, i_size is set to the new size by
969 * the VFS prior to calling ext4_truncate(), but the filesystem won't
970 * set i_disksize to 0 until the truncate is actually under way.
971 *
972 * The intent is that i_disksize always represents the blocks which
973 * are used by this file. This allows recovery to restart truncate
974 * on orphans if we crash during truncate. We actually write i_disksize
975 * into the on-disk inode when writing inodes out, instead of i_size.
976 *
977 * The only time when i_disksize and i_size may be different is when
978 * a truncate is in progress. The only things which change i_disksize
979 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
980 */
981 loff_t i_disksize;
982
983 /*
984 * i_data_sem is for serialising ext4_truncate() against
985 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
986 * data tree are chopped off during truncate. We can't do that in
987 * ext4 because whenever we perform intermediate commits during
988 * truncate, the inode and all the metadata blocks *must* be in a
989 * consistent state which allows truncation of the orphans to restart
990 * during recovery. Hence we must fix the get_block-vs-truncate race
991 * by other means, so we have i_data_sem.
992 */
993 struct rw_semaphore i_data_sem;
994 /*
995 * i_mmap_sem is for serializing page faults with truncate / punch hole
996 * operations. We have to make sure that new page cannot be faulted in
997 * a section of the inode that is being punched. We cannot easily use
998 * i_data_sem for this since we need protection for the whole punch
999 * operation and i_data_sem ranks below transaction start so we have
1000 * to occasionally drop it.
1001 */
1002 struct rw_semaphore i_mmap_sem;
1003 struct inode vfs_inode;
1004 struct jbd2_inode *jinode;
1005
1006 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1007
1008 /*
1009 * File creation time. Its function is same as that of
1010 * struct timespec i_{a,c,m}time in the generic inode.
1011 */
1012 struct timespec i_crtime;
1013
1014 /* mballoc */
1015 struct list_head i_prealloc_list;
1016 spinlock_t i_prealloc_lock;
1017
1018 /* extents status tree */
1019 struct ext4_es_tree i_es_tree;
1020 rwlock_t i_es_lock;
1021 struct list_head i_es_list;
1022 unsigned int i_es_all_nr; /* protected by i_es_lock */
1023 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1024 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1025 extents to shrink. Protected by
1026 i_es_lock */
1027
1028 /* ialloc */
1029 ext4_group_t i_last_alloc_group;
1030
1031 /* allocation reservation info for delalloc */
1032 /* In case of bigalloc, these refer to clusters rather than blocks */
1033 unsigned int i_reserved_data_blocks;
1034 unsigned int i_reserved_meta_blocks;
1035 unsigned int i_allocated_meta_blocks;
1036 ext4_lblk_t i_da_metadata_calc_last_lblock;
1037 int i_da_metadata_calc_len;
1038
1039 /* on-disk additional length */
1040 __u16 i_extra_isize;
1041
1042 /* Indicate the inline data space. */
1043 u16 i_inline_off;
1044 u16 i_inline_size;
1045
1046#ifdef CONFIG_QUOTA
1047 /* quota space reservation, managed internally by quota code */
1048 qsize_t i_reserved_quota;
1049#endif
1050
1051 /* Lock protecting lists below */
1052 spinlock_t i_completed_io_lock;
1053 /*
1054 * Completed IOs that need unwritten extents handling and have
1055 * transaction reserved
1056 */
1057 struct list_head i_rsv_conversion_list;
1058 struct work_struct i_rsv_conversion_work;
1059 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1060
1061 spinlock_t i_block_reservation_lock;
1062
1063 /*
1064 * Transactions that contain inode's metadata needed to complete
1065 * fsync and fdatasync, respectively.
1066 */
1067 tid_t i_sync_tid;
1068 tid_t i_datasync_tid;
1069
1070#ifdef CONFIG_QUOTA
1071 struct dquot *i_dquot[MAXQUOTAS];
1072#endif
1073
1074 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1075 __u32 i_csum_seed;
1076
1077 kprojid_t i_projid;
1078};
1079
1080/*
1081 * File system states
1082 */
1083#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1084#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1085#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1086
1087/*
1088 * Misc. filesystem flags
1089 */
1090#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1091#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1092#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1093
1094/*
1095 * Mount flags set via mount options or defaults
1096 */
1097#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1098#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1099#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1100#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1101#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1102#define EXT4_MOUNT_ERRORS_MASK 0x00070
1103#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1104#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1105#ifdef CONFIG_FS_DAX
1106#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1107#else
1108#define EXT4_MOUNT_DAX 0
1109#endif
1110#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1111#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1112#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1113#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1114#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1115#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1116#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1117#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1118#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1119#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1120#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1121#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1122#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1123#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1124#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1125#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1126#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1127#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1128#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1129#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1130#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1131
1132/*
1133 * Mount flags set either automatically (could not be set by mount option)
1134 * based on per file system feature or property or in special cases such as
1135 * distinguishing between explicit mount option definition and default.
1136 */
1137#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1138 specified delalloc */
1139#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1140 size of blocksize * 8
1141 blocks */
1142#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1143 file systems */
1144
1145#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1146 specified journal checksum */
1147
1148#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1149 ~EXT4_MOUNT_##opt
1150#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1151 EXT4_MOUNT_##opt
1152#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1153 EXT4_MOUNT_##opt)
1154
1155#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1156 ~EXT4_MOUNT2_##opt
1157#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1158 EXT4_MOUNT2_##opt
1159#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1160 EXT4_MOUNT2_##opt)
1161
1162#define ext4_test_and_set_bit __test_and_set_bit_le
1163#define ext4_set_bit __set_bit_le
1164#define ext4_set_bit_atomic ext2_set_bit_atomic
1165#define ext4_test_and_clear_bit __test_and_clear_bit_le
1166#define ext4_clear_bit __clear_bit_le
1167#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1168#define ext4_test_bit test_bit_le
1169#define ext4_find_next_zero_bit find_next_zero_bit_le
1170#define ext4_find_next_bit find_next_bit_le
1171
1172extern void ext4_set_bits(void *bm, int cur, int len);
1173
1174/*
1175 * Maximal mount counts between two filesystem checks
1176 */
1177#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1178#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1179
1180/*
1181 * Behaviour when detecting errors
1182 */
1183#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1184#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1185#define EXT4_ERRORS_PANIC 3 /* Panic */
1186#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1187
1188/* Metadata checksum algorithm codes */
1189#define EXT4_CRC32C_CHKSUM 1
1190
1191/*
1192 * Structure of the super block
1193 */
1194struct ext4_super_block {
1195/*00*/ __le32 s_inodes_count; /* Inodes count */
1196 __le32 s_blocks_count_lo; /* Blocks count */
1197 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1198 __le32 s_free_blocks_count_lo; /* Free blocks count */
1199/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1200 __le32 s_first_data_block; /* First Data Block */
1201 __le32 s_log_block_size; /* Block size */
1202 __le32 s_log_cluster_size; /* Allocation cluster size */
1203/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1204 __le32 s_clusters_per_group; /* # Clusters per group */
1205 __le32 s_inodes_per_group; /* # Inodes per group */
1206 __le32 s_mtime; /* Mount time */
1207/*30*/ __le32 s_wtime; /* Write time */
1208 __le16 s_mnt_count; /* Mount count */
1209 __le16 s_max_mnt_count; /* Maximal mount count */
1210 __le16 s_magic; /* Magic signature */
1211 __le16 s_state; /* File system state */
1212 __le16 s_errors; /* Behaviour when detecting errors */
1213 __le16 s_minor_rev_level; /* minor revision level */
1214/*40*/ __le32 s_lastcheck; /* time of last check */
1215 __le32 s_checkinterval; /* max. time between checks */
1216 __le32 s_creator_os; /* OS */
1217 __le32 s_rev_level; /* Revision level */
1218/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1219 __le16 s_def_resgid; /* Default gid for reserved blocks */
1220 /*
1221 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1222 *
1223 * Note: the difference between the compatible feature set and
1224 * the incompatible feature set is that if there is a bit set
1225 * in the incompatible feature set that the kernel doesn't
1226 * know about, it should refuse to mount the filesystem.
1227 *
1228 * e2fsck's requirements are more strict; if it doesn't know
1229 * about a feature in either the compatible or incompatible
1230 * feature set, it must abort and not try to meddle with
1231 * things it doesn't understand...
1232 */
1233 __le32 s_first_ino; /* First non-reserved inode */
1234 __le16 s_inode_size; /* size of inode structure */
1235 __le16 s_block_group_nr; /* block group # of this superblock */
1236 __le32 s_feature_compat; /* compatible feature set */
1237/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1238 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1239/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1240/*78*/ char s_volume_name[16]; /* volume name */
1241/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1242/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1243 /*
1244 * Performance hints. Directory preallocation should only
1245 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1246 */
1247 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1248 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1249 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1250 /*
1251 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1252 */
1253/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1254/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1255 __le32 s_journal_dev; /* device number of journal file */
1256 __le32 s_last_orphan; /* start of list of inodes to delete */
1257 __le32 s_hash_seed[4]; /* HTREE hash seed */
1258 __u8 s_def_hash_version; /* Default hash version to use */
1259 __u8 s_jnl_backup_type;
1260 __le16 s_desc_size; /* size of group descriptor */
1261/*100*/ __le32 s_default_mount_opts;
1262 __le32 s_first_meta_bg; /* First metablock block group */
1263 __le32 s_mkfs_time; /* When the filesystem was created */
1264 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1265 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1266/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1267 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1268 __le32 s_free_blocks_count_hi; /* Free blocks count */
1269 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1270 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1271 __le32 s_flags; /* Miscellaneous flags */
1272 __le16 s_raid_stride; /* RAID stride */
1273 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1274 __le64 s_mmp_block; /* Block for multi-mount protection */
1275 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1276 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1277 __u8 s_checksum_type; /* metadata checksum algorithm used */
1278 __u8 s_encryption_level; /* versioning level for encryption */
1279 __u8 s_reserved_pad; /* Padding to next 32bits */
1280 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1281 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1282 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1283 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1284 snapshot's future use */
1285 __le32 s_snapshot_list; /* inode number of the head of the
1286 on-disk snapshot list */
1287#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1288 __le32 s_error_count; /* number of fs errors */
1289 __le32 s_first_error_time; /* first time an error happened */
1290 __le32 s_first_error_ino; /* inode involved in first error */
1291 __le64 s_first_error_block; /* block involved of first error */
1292 __u8 s_first_error_func[32]; /* function where the error happened */
1293 __le32 s_first_error_line; /* line number where error happened */
1294 __le32 s_last_error_time; /* most recent time of an error */
1295 __le32 s_last_error_ino; /* inode involved in last error */
1296 __le32 s_last_error_line; /* line number where error happened */
1297 __le64 s_last_error_block; /* block involved of last error */
1298 __u8 s_last_error_func[32]; /* function where the error happened */
1299#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1300 __u8 s_mount_opts[64];
1301 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1302 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1303 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1304 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1305 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1306 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1307 __le32 s_lpf_ino; /* Location of the lost+found inode */
1308 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1309 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1310 __le32 s_reserved[98]; /* Padding to the end of the block */
1311 __le32 s_checksum; /* crc32c(superblock) */
1312};
1313
1314#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1315
1316#ifdef __KERNEL__
1317
1318/*
1319 * run-time mount flags
1320 */
1321#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1322#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1323#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1324
1325#ifdef CONFIG_EXT4_FS_ENCRYPTION
1326#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1327 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1328#else
1329#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1330#endif
1331
1332/* Number of quota types we support */
1333#define EXT4_MAXQUOTAS 3
1334
1335#ifdef CONFIG_EXT4_FS_ENCRYPTION
1336#define EXT4_KEY_DESC_PREFIX "ext4:"
1337#define EXT4_KEY_DESC_PREFIX_SIZE 5
1338#endif
1339
1340/*
1341 * fourth extended-fs super-block data in memory
1342 */
1343struct ext4_sb_info {
1344 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1345 unsigned long s_inodes_per_block;/* Number of inodes per block */
1346 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1347 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1348 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1349 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1350 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1351 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1352 ext4_group_t s_groups_count; /* Number of groups in the fs */
1353 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1354 unsigned long s_overhead; /* # of fs overhead clusters */
1355 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1356 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1357 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1358 struct buffer_head * s_sbh; /* Buffer containing the super block */
1359 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1360 struct buffer_head **s_group_desc;
1361 unsigned int s_mount_opt;
1362 unsigned int s_mount_opt2;
1363 unsigned int s_mount_flags;
1364 unsigned int s_def_mount_opt;
1365 ext4_fsblk_t s_sb_block;
1366 atomic64_t s_resv_clusters;
1367 kuid_t s_resuid;
1368 kgid_t s_resgid;
1369 unsigned short s_mount_state;
1370 unsigned short s_pad;
1371 int s_addr_per_block_bits;
1372 int s_desc_per_block_bits;
1373 int s_inode_size;
1374 int s_first_ino;
1375 unsigned int s_inode_readahead_blks;
1376 unsigned int s_inode_goal;
1377 spinlock_t s_next_gen_lock;
1378 u32 s_next_generation;
1379 u32 s_hash_seed[4];
1380 int s_def_hash_version;
1381 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1382 struct percpu_counter s_freeclusters_counter;
1383 struct percpu_counter s_freeinodes_counter;
1384 struct percpu_counter s_dirs_counter;
1385 struct percpu_counter s_dirtyclusters_counter;
1386 struct blockgroup_lock *s_blockgroup_lock;
1387 struct proc_dir_entry *s_proc;
1388 struct kobject s_kobj;
1389 struct completion s_kobj_unregister;
1390 struct super_block *s_sb;
1391
1392 /* Journaling */
1393 struct journal_s *s_journal;
1394 struct list_head s_orphan;
1395 struct mutex s_orphan_lock;
1396 unsigned long s_resize_flags; /* Flags indicating if there
1397 is a resizer */
1398 unsigned long s_commit_interval;
1399 u32 s_max_batch_time;
1400 u32 s_min_batch_time;
1401 struct block_device *journal_bdev;
1402#ifdef CONFIG_QUOTA
1403 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1404 int s_jquota_fmt; /* Format of quota to use */
1405#endif
1406 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1407 struct rb_root system_blks;
1408
1409#ifdef EXTENTS_STATS
1410 /* ext4 extents stats */
1411 unsigned long s_ext_min;
1412 unsigned long s_ext_max;
1413 unsigned long s_depth_max;
1414 spinlock_t s_ext_stats_lock;
1415 unsigned long s_ext_blocks;
1416 unsigned long s_ext_extents;
1417#endif
1418
1419 /* for buddy allocator */
1420 struct ext4_group_info ***s_group_info;
1421 struct inode *s_buddy_cache;
1422 spinlock_t s_md_lock;
1423 unsigned short *s_mb_offsets;
1424 unsigned int *s_mb_maxs;
1425 unsigned int s_group_info_size;
1426 unsigned int s_mb_free_pending;
1427
1428 /* tunables */
1429 unsigned long s_stripe;
1430 unsigned int s_mb_stream_request;
1431 unsigned int s_mb_max_to_scan;
1432 unsigned int s_mb_min_to_scan;
1433 unsigned int s_mb_stats;
1434 unsigned int s_mb_order2_reqs;
1435 unsigned int s_mb_group_prealloc;
1436 unsigned int s_max_dir_size_kb;
1437 /* where last allocation was done - for stream allocation */
1438 unsigned long s_mb_last_group;
1439 unsigned long s_mb_last_start;
1440
1441 /* stats for buddy allocator */
1442 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1443 atomic_t s_bal_success; /* we found long enough chunks */
1444 atomic_t s_bal_allocated; /* in blocks */
1445 atomic_t s_bal_ex_scanned; /* total extents scanned */
1446 atomic_t s_bal_goals; /* goal hits */
1447 atomic_t s_bal_breaks; /* too long searches */
1448 atomic_t s_bal_2orders; /* 2^order hits */
1449 spinlock_t s_bal_lock;
1450 unsigned long s_mb_buddies_generated;
1451 unsigned long long s_mb_generation_time;
1452 atomic_t s_mb_lost_chunks;
1453 atomic_t s_mb_preallocated;
1454 atomic_t s_mb_discarded;
1455 atomic_t s_lock_busy;
1456
1457 /* locality groups */
1458 struct ext4_locality_group __percpu *s_locality_groups;
1459
1460 /* for write statistics */
1461 unsigned long s_sectors_written_start;
1462 u64 s_kbytes_written;
1463
1464 /* the size of zero-out chunk */
1465 unsigned int s_extent_max_zeroout_kb;
1466
1467 unsigned int s_log_groups_per_flex;
1468 struct flex_groups *s_flex_groups;
1469 ext4_group_t s_flex_groups_allocated;
1470
1471 /* workqueue for reserved extent conversions (buffered io) */
1472 struct workqueue_struct *rsv_conversion_wq;
1473
1474 /* timer for periodic error stats printing */
1475 struct timer_list s_err_report;
1476
1477 /* Lazy inode table initialization info */
1478 struct ext4_li_request *s_li_request;
1479 /* Wait multiplier for lazy initialization thread */
1480 unsigned int s_li_wait_mult;
1481
1482 /* Kernel thread for multiple mount protection */
1483 struct task_struct *s_mmp_tsk;
1484
1485 /* record the last minlen when FITRIM is called. */
1486 atomic_t s_last_trim_minblks;
1487
1488 /* Reference to checksum algorithm driver via cryptoapi */
1489 struct crypto_shash *s_chksum_driver;
1490
1491 /* Precomputed FS UUID checksum for seeding other checksums */
1492 __u32 s_csum_seed;
1493
1494 /* Reclaim extents from extent status tree */
1495 struct shrinker s_es_shrinker;
1496 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1497 long s_es_nr_inode;
1498 struct ext4_es_stats s_es_stats;
1499 struct mb_cache *s_mb_cache;
1500 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1501
1502 /* Ratelimit ext4 messages. */
1503 struct ratelimit_state s_err_ratelimit_state;
1504 struct ratelimit_state s_warning_ratelimit_state;
1505 struct ratelimit_state s_msg_ratelimit_state;
1506
1507 /* Barrier between changing inodes' journal flags and writepages ops. */
1508 struct percpu_rw_semaphore s_journal_flag_rwsem;
1509
1510 /* Encryption support */
1511#ifdef CONFIG_EXT4_FS_ENCRYPTION
1512 u8 key_prefix[EXT4_KEY_DESC_PREFIX_SIZE];
1513 u8 key_prefix_size;
1514#endif
1515};
1516
1517static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1518{
1519 return sb->s_fs_info;
1520}
1521static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1522{
1523 return container_of(inode, struct ext4_inode_info, vfs_inode);
1524}
1525
1526static inline struct timespec ext4_current_time(struct inode *inode)
1527{
1528 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1529 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1530}
1531
1532static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1533{
1534 return ino == EXT4_ROOT_INO ||
1535 ino == EXT4_USR_QUOTA_INO ||
1536 ino == EXT4_GRP_QUOTA_INO ||
1537 ino == EXT4_BOOT_LOADER_INO ||
1538 ino == EXT4_JOURNAL_INO ||
1539 ino == EXT4_RESIZE_INO ||
1540 (ino >= EXT4_FIRST_INO(sb) &&
1541 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1542}
1543
1544/*
1545 * Inode dynamic state flags
1546 */
1547enum {
1548 EXT4_STATE_JDATA, /* journaled data exists */
1549 EXT4_STATE_NEW, /* inode is newly created */
1550 EXT4_STATE_XATTR, /* has in-inode xattrs */
1551 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1552 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1553 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1554 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1555 EXT4_STATE_NEWENTRY, /* File just added to dir */
1556 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1557 nolocking */
1558 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1559 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1560};
1561
1562#define EXT4_INODE_BIT_FNS(name, field, offset) \
1563static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1564{ \
1565 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1566} \
1567static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1568{ \
1569 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1570} \
1571static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1572{ \
1573 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1574}
1575
1576/* Add these declarations here only so that these functions can be
1577 * found by name. Otherwise, they are very hard to locate. */
1578static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1579static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1580static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1581EXT4_INODE_BIT_FNS(flag, flags, 0)
1582
1583/* Add these declarations here only so that these functions can be
1584 * found by name. Otherwise, they are very hard to locate. */
1585static inline int ext4_test_inode_state(struct inode *inode, int bit);
1586static inline void ext4_set_inode_state(struct inode *inode, int bit);
1587static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1588#if (BITS_PER_LONG < 64)
1589EXT4_INODE_BIT_FNS(state, state_flags, 0)
1590
1591static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1592{
1593 (ei)->i_state_flags = 0;
1594}
1595#else
1596EXT4_INODE_BIT_FNS(state, flags, 32)
1597
1598static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1599{
1600 /* We depend on the fact that callers will set i_flags */
1601}
1602#endif
1603#else
1604/* Assume that user mode programs are passing in an ext4fs superblock, not
1605 * a kernel struct super_block. This will allow us to call the feature-test
1606 * macros from user land. */
1607#define EXT4_SB(sb) (sb)
1608#endif
1609
1610/*
1611 * Returns true if the inode is inode is encrypted
1612 */
1613#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1614
1615/*
1616 * Codes for operating systems
1617 */
1618#define EXT4_OS_LINUX 0
1619#define EXT4_OS_HURD 1
1620#define EXT4_OS_MASIX 2
1621#define EXT4_OS_FREEBSD 3
1622#define EXT4_OS_LITES 4
1623
1624/*
1625 * Revision levels
1626 */
1627#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1628#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1629
1630#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1631#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1632
1633#define EXT4_GOOD_OLD_INODE_SIZE 128
1634
1635/*
1636 * Feature set definitions
1637 */
1638
1639/* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1640#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1641 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1642#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1643 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1644#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1645 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1646#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1647 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1648#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1649 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1650#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1651 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1652#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1653 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1654#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1655 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1656#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1657 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1658
1659#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1660#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1661#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1662#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1663#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1664#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1665#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1666
1667#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1668#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1669#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1670#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1671#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1672#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1673#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1674#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1675#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1676/*
1677 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1678 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1679 * all other data structures' checksums. However, the METADATA_CSUM and
1680 * GDT_CSUM bits are mutually exclusive.
1681 */
1682#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1683#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1684#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1685
1686#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1687#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1688#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1689#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1690#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1691#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1692#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1693#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1694#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1695#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1696#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1697#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1698#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1699#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1700#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1701
1702#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1703static inline bool ext4_has_feature_##name(struct super_block *sb) \
1704{ \
1705 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1706 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1707} \
1708static inline void ext4_set_feature_##name(struct super_block *sb) \
1709{ \
1710 EXT4_SB(sb)->s_es->s_feature_compat |= \
1711 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1712} \
1713static inline void ext4_clear_feature_##name(struct super_block *sb) \
1714{ \
1715 EXT4_SB(sb)->s_es->s_feature_compat &= \
1716 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1717}
1718
1719#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1720static inline bool ext4_has_feature_##name(struct super_block *sb) \
1721{ \
1722 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1723 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1724} \
1725static inline void ext4_set_feature_##name(struct super_block *sb) \
1726{ \
1727 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1728 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1729} \
1730static inline void ext4_clear_feature_##name(struct super_block *sb) \
1731{ \
1732 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1733 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1734}
1735
1736#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1737static inline bool ext4_has_feature_##name(struct super_block *sb) \
1738{ \
1739 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1740 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1741} \
1742static inline void ext4_set_feature_##name(struct super_block *sb) \
1743{ \
1744 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1745 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1746} \
1747static inline void ext4_clear_feature_##name(struct super_block *sb) \
1748{ \
1749 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1750 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1751}
1752
1753EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1754EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1755EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1756EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1757EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1758EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1759EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1760
1761EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1762EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1763EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1764EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1765EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1766EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1767EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1768EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1769EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1770EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1771EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1772EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1773
1774EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1775EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1776EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1777EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1778EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1779EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1780EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1781EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1782EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1783EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1784EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1785EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1786EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1787EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1788EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1789
1790#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1791#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1792 EXT4_FEATURE_INCOMPAT_META_BG)
1793#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1794 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1795 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1796
1797#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1798#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1799 EXT4_FEATURE_INCOMPAT_RECOVER| \
1800 EXT4_FEATURE_INCOMPAT_META_BG)
1801#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1802 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1803 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1804
1805#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1806#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1807 EXT4_FEATURE_INCOMPAT_RECOVER| \
1808 EXT4_FEATURE_INCOMPAT_META_BG| \
1809 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1810 EXT4_FEATURE_INCOMPAT_64BIT| \
1811 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1812 EXT4_FEATURE_INCOMPAT_MMP | \
1813 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1814 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1815 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1816#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1817 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1818 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1819 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1820 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1821 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1822 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1823 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1824 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1825 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1826 EXT4_FEATURE_RO_COMPAT_PROJECT)
1827
1828#define EXTN_FEATURE_FUNCS(ver) \
1829static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1830{ \
1831 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1832 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1833} \
1834static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1835{ \
1836 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1837 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1838} \
1839static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1840{ \
1841 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1842 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1843}
1844
1845EXTN_FEATURE_FUNCS(2)
1846EXTN_FEATURE_FUNCS(3)
1847EXTN_FEATURE_FUNCS(4)
1848
1849static inline bool ext4_has_compat_features(struct super_block *sb)
1850{
1851 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1852}
1853static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1854{
1855 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1856}
1857static inline bool ext4_has_incompat_features(struct super_block *sb)
1858{
1859 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1860}
1861
1862/*
1863 * Default values for user and/or group using reserved blocks
1864 */
1865#define EXT4_DEF_RESUID 0
1866#define EXT4_DEF_RESGID 0
1867
1868/*
1869 * Default project ID
1870 */
1871#define EXT4_DEF_PROJID 0
1872
1873#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1874
1875/*
1876 * Default mount options
1877 */
1878#define EXT4_DEFM_DEBUG 0x0001
1879#define EXT4_DEFM_BSDGROUPS 0x0002
1880#define EXT4_DEFM_XATTR_USER 0x0004
1881#define EXT4_DEFM_ACL 0x0008
1882#define EXT4_DEFM_UID16 0x0010
1883#define EXT4_DEFM_JMODE 0x0060
1884#define EXT4_DEFM_JMODE_DATA 0x0020
1885#define EXT4_DEFM_JMODE_ORDERED 0x0040
1886#define EXT4_DEFM_JMODE_WBACK 0x0060
1887#define EXT4_DEFM_NOBARRIER 0x0100
1888#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1889#define EXT4_DEFM_DISCARD 0x0400
1890#define EXT4_DEFM_NODELALLOC 0x0800
1891
1892/*
1893 * Default journal batch times
1894 */
1895#define EXT4_DEF_MIN_BATCH_TIME 0
1896#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1897
1898/*
1899 * Minimum number of groups in a flexgroup before we separate out
1900 * directories into the first block group of a flexgroup
1901 */
1902#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1903
1904/*
1905 * Structure of a directory entry
1906 */
1907#define EXT4_NAME_LEN 255
1908
1909struct ext4_dir_entry {
1910 __le32 inode; /* Inode number */
1911 __le16 rec_len; /* Directory entry length */
1912 __le16 name_len; /* Name length */
1913 char name[EXT4_NAME_LEN]; /* File name */
1914};
1915
1916/*
1917 * The new version of the directory entry. Since EXT4 structures are
1918 * stored in intel byte order, and the name_len field could never be
1919 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1920 * file_type field.
1921 */
1922struct ext4_dir_entry_2 {
1923 __le32 inode; /* Inode number */
1924 __le16 rec_len; /* Directory entry length */
1925 __u8 name_len; /* Name length */
1926 __u8 file_type;
1927 char name[EXT4_NAME_LEN]; /* File name */
1928};
1929
1930/*
1931 * This is a bogus directory entry at the end of each leaf block that
1932 * records checksums.
1933 */
1934struct ext4_dir_entry_tail {
1935 __le32 det_reserved_zero1; /* Pretend to be unused */
1936 __le16 det_rec_len; /* 12 */
1937 __u8 det_reserved_zero2; /* Zero name length */
1938 __u8 det_reserved_ft; /* 0xDE, fake file type */
1939 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1940};
1941
1942#define EXT4_DIRENT_TAIL(block, blocksize) \
1943 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1944 ((blocksize) - \
1945 sizeof(struct ext4_dir_entry_tail))))
1946
1947/*
1948 * Ext4 directory file types. Only the low 3 bits are used. The
1949 * other bits are reserved for now.
1950 */
1951#define EXT4_FT_UNKNOWN 0
1952#define EXT4_FT_REG_FILE 1
1953#define EXT4_FT_DIR 2
1954#define EXT4_FT_CHRDEV 3
1955#define EXT4_FT_BLKDEV 4
1956#define EXT4_FT_FIFO 5
1957#define EXT4_FT_SOCK 6
1958#define EXT4_FT_SYMLINK 7
1959
1960#define EXT4_FT_MAX 8
1961
1962#define EXT4_FT_DIR_CSUM 0xDE
1963
1964/*
1965 * EXT4_DIR_PAD defines the directory entries boundaries
1966 *
1967 * NOTE: It must be a multiple of 4
1968 */
1969#define EXT4_DIR_PAD 4
1970#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1971#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1972 ~EXT4_DIR_ROUND)
1973#define EXT4_MAX_REC_LEN ((1<<16)-1)
1974
1975/*
1976 * If we ever get support for fs block sizes > page_size, we'll need
1977 * to remove the #if statements in the next two functions...
1978 */
1979static inline unsigned int
1980ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1981{
1982 unsigned len = le16_to_cpu(dlen);
1983
1984#if (PAGE_SIZE >= 65536)
1985 if (len == EXT4_MAX_REC_LEN || len == 0)
1986 return blocksize;
1987 return (len & 65532) | ((len & 3) << 16);
1988#else
1989 return len;
1990#endif
1991}
1992
1993static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1994{
1995 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1996 BUG();
1997#if (PAGE_SIZE >= 65536)
1998 if (len < 65536)
1999 return cpu_to_le16(len);
2000 if (len == blocksize) {
2001 if (blocksize == 65536)
2002 return cpu_to_le16(EXT4_MAX_REC_LEN);
2003 else
2004 return cpu_to_le16(0);
2005 }
2006 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2007#else
2008 return cpu_to_le16(len);
2009#endif
2010}
2011
2012/*
2013 * Hash Tree Directory indexing
2014 * (c) Daniel Phillips, 2001
2015 */
2016
2017#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2018 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2019#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2020#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2021
2022/* Legal values for the dx_root hash_version field: */
2023
2024#define DX_HASH_LEGACY 0
2025#define DX_HASH_HALF_MD4 1
2026#define DX_HASH_TEA 2
2027#define DX_HASH_LEGACY_UNSIGNED 3
2028#define DX_HASH_HALF_MD4_UNSIGNED 4
2029#define DX_HASH_TEA_UNSIGNED 5
2030
2031static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2032 const void *address, unsigned int length)
2033{
2034 struct {
2035 struct shash_desc shash;
2036 char ctx[4];
2037 } desc;
2038 int err;
2039
2040 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2041
2042 desc.shash.tfm = sbi->s_chksum_driver;
2043 desc.shash.flags = 0;
2044 *(u32 *)desc.ctx = crc;
2045
2046 err = crypto_shash_update(&desc.shash, address, length);
2047 BUG_ON(err);
2048
2049 return *(u32 *)desc.ctx;
2050}
2051
2052#ifdef __KERNEL__
2053
2054/* hash info structure used by the directory hash */
2055struct dx_hash_info
2056{
2057 u32 hash;
2058 u32 minor_hash;
2059 int hash_version;
2060 u32 *seed;
2061};
2062
2063
2064/* 32 and 64 bit signed EOF for dx directories */
2065#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2066#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2067
2068
2069/*
2070 * Control parameters used by ext4_htree_next_block
2071 */
2072#define HASH_NB_ALWAYS 1
2073
2074struct ext4_filename {
2075 const struct qstr *usr_fname;
2076 struct fscrypt_str disk_name;
2077 struct dx_hash_info hinfo;
2078#ifdef CONFIG_EXT4_FS_ENCRYPTION
2079 struct fscrypt_str crypto_buf;
2080#endif
2081};
2082
2083#define fname_name(p) ((p)->disk_name.name)
2084#define fname_len(p) ((p)->disk_name.len)
2085
2086/*
2087 * Describe an inode's exact location on disk and in memory
2088 */
2089struct ext4_iloc
2090{
2091 struct buffer_head *bh;
2092 unsigned long offset;
2093 ext4_group_t block_group;
2094};
2095
2096static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2097{
2098 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2099}
2100
2101/*
2102 * This structure is stuffed into the struct file's private_data field
2103 * for directories. It is where we put information so that we can do
2104 * readdir operations in hash tree order.
2105 */
2106struct dir_private_info {
2107 struct rb_root root;
2108 struct rb_node *curr_node;
2109 struct fname *extra_fname;
2110 loff_t last_pos;
2111 __u32 curr_hash;
2112 __u32 curr_minor_hash;
2113 __u32 next_hash;
2114};
2115
2116/* calculate the first block number of the group */
2117static inline ext4_fsblk_t
2118ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2119{
2120 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2121 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2122}
2123
2124/*
2125 * Special error return code only used by dx_probe() and its callers.
2126 */
2127#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2128
2129/*
2130 * Timeout and state flag for lazy initialization inode thread.
2131 */
2132#define EXT4_DEF_LI_WAIT_MULT 10
2133#define EXT4_DEF_LI_MAX_START_DELAY 5
2134#define EXT4_LAZYINIT_QUIT 0x0001
2135#define EXT4_LAZYINIT_RUNNING 0x0002
2136
2137/*
2138 * Lazy inode table initialization info
2139 */
2140struct ext4_lazy_init {
2141 unsigned long li_state;
2142 struct list_head li_request_list;
2143 struct mutex li_list_mtx;
2144};
2145
2146struct ext4_li_request {
2147 struct super_block *lr_super;
2148 struct ext4_sb_info *lr_sbi;
2149 ext4_group_t lr_next_group;
2150 struct list_head lr_request;
2151 unsigned long lr_next_sched;
2152 unsigned long lr_timeout;
2153};
2154
2155struct ext4_features {
2156 struct kobject f_kobj;
2157 struct completion f_kobj_unregister;
2158};
2159
2160/*
2161 * This structure will be used for multiple mount protection. It will be
2162 * written into the block number saved in the s_mmp_block field in the
2163 * superblock. Programs that check MMP should assume that if
2164 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2165 * to use the filesystem, regardless of how old the timestamp is.
2166 */
2167#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2168#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2169#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2170#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2171
2172struct mmp_struct {
2173 __le32 mmp_magic; /* Magic number for MMP */
2174 __le32 mmp_seq; /* Sequence no. updated periodically */
2175
2176 /*
2177 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2178 * purposes and do not affect the correctness of the algorithm
2179 */
2180 __le64 mmp_time; /* Time last updated */
2181 char mmp_nodename[64]; /* Node which last updated MMP block */
2182 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2183
2184 /*
2185 * mmp_check_interval is used to verify if the MMP block has been
2186 * updated on the block device. The value is updated based on the
2187 * maximum time to write the MMP block during an update cycle.
2188 */
2189 __le16 mmp_check_interval;
2190
2191 __le16 mmp_pad1;
2192 __le32 mmp_pad2[226];
2193 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2194};
2195
2196/* arguments passed to the mmp thread */
2197struct mmpd_data {
2198 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2199 struct super_block *sb; /* super block of the fs */
2200};
2201
2202/*
2203 * Check interval multiplier
2204 * The MMP block is written every update interval and initially checked every
2205 * update interval x the multiplier (the value is then adapted based on the
2206 * write latency). The reason is that writes can be delayed under load and we
2207 * don't want readers to incorrectly assume that the filesystem is no longer
2208 * in use.
2209 */
2210#define EXT4_MMP_CHECK_MULT 2UL
2211
2212/*
2213 * Minimum interval for MMP checking in seconds.
2214 */
2215#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2216
2217/*
2218 * Maximum interval for MMP checking in seconds.
2219 */
2220#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2221
2222/*
2223 * Function prototypes
2224 */
2225
2226/*
2227 * Ok, these declarations are also in <linux/kernel.h> but none of the
2228 * ext4 source programs needs to include it so they are duplicated here.
2229 */
2230# define NORET_TYPE /**/
2231# define ATTRIB_NORET __attribute__((noreturn))
2232# define NORET_AND noreturn,
2233
2234/* bitmap.c */
2235extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2236void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2237 struct ext4_group_desc *gdp,
2238 struct buffer_head *bh, int sz);
2239int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2240 struct ext4_group_desc *gdp,
2241 struct buffer_head *bh, int sz);
2242void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2243 struct ext4_group_desc *gdp,
2244 struct buffer_head *bh);
2245int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2246 struct ext4_group_desc *gdp,
2247 struct buffer_head *bh);
2248
2249/* balloc.c */
2250extern void ext4_get_group_no_and_offset(struct super_block *sb,
2251 ext4_fsblk_t blocknr,
2252 ext4_group_t *blockgrpp,
2253 ext4_grpblk_t *offsetp);
2254extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2255 ext4_fsblk_t block);
2256
2257extern unsigned int ext4_block_group(struct super_block *sb,
2258 ext4_fsblk_t blocknr);
2259extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2260 ext4_fsblk_t blocknr);
2261extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2262extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2263 ext4_group_t group);
2264extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2265 ext4_fsblk_t goal,
2266 unsigned int flags,
2267 unsigned long *count,
2268 int *errp);
2269extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2270 s64 nclusters, unsigned int flags);
2271extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2272extern void ext4_check_blocks_bitmap(struct super_block *);
2273extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2274 ext4_group_t block_group,
2275 struct buffer_head ** bh);
2276extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2277
2278extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2279 ext4_group_t block_group);
2280extern int ext4_wait_block_bitmap(struct super_block *sb,
2281 ext4_group_t block_group,
2282 struct buffer_head *bh);
2283extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2284 ext4_group_t block_group);
2285extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2286 ext4_group_t block_group,
2287 struct ext4_group_desc *gdp);
2288ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2289
2290static inline int ext4_sb_has_crypto(struct super_block *sb)
2291{
2292 return ext4_has_feature_encrypt(sb);
2293}
2294
2295static inline bool ext4_encrypted_inode(struct inode *inode)
2296{
2297 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2298}
2299
2300#ifdef CONFIG_EXT4_FS_ENCRYPTION
2301static inline int ext4_fname_setup_filename(struct inode *dir,
2302 const struct qstr *iname,
2303 int lookup, struct ext4_filename *fname)
2304{
2305 struct fscrypt_name name;
2306 int err;
2307
2308 memset(fname, 0, sizeof(struct ext4_filename));
2309
2310 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2311
2312 fname->usr_fname = name.usr_fname;
2313 fname->disk_name = name.disk_name;
2314 fname->hinfo.hash = name.hash;
2315 fname->hinfo.minor_hash = name.minor_hash;
2316 fname->crypto_buf = name.crypto_buf;
2317 return err;
2318}
2319
2320static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2321{
2322 struct fscrypt_name name;
2323
2324 name.crypto_buf = fname->crypto_buf;
2325 fscrypt_free_filename(&name);
2326
2327 fname->crypto_buf.name = NULL;
2328 fname->usr_fname = NULL;
2329 fname->disk_name.name = NULL;
2330}
2331#else
2332static inline int ext4_fname_setup_filename(struct inode *dir,
2333 const struct qstr *iname,
2334 int lookup, struct ext4_filename *fname)
2335{
2336 fname->usr_fname = iname;
2337 fname->disk_name.name = (unsigned char *) iname->name;
2338 fname->disk_name.len = iname->len;
2339 return 0;
2340}
2341static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2342
2343#define fscrypt_set_d_op(i)
2344#define fscrypt_get_ctx fscrypt_notsupp_get_ctx
2345#define fscrypt_release_ctx fscrypt_notsupp_release_ctx
2346#define fscrypt_encrypt_page fscrypt_notsupp_encrypt_page
2347#define fscrypt_decrypt_page fscrypt_notsupp_decrypt_page
2348#define fscrypt_decrypt_bio_pages fscrypt_notsupp_decrypt_bio_pages
2349#define fscrypt_pullback_bio_page fscrypt_notsupp_pullback_bio_page
2350#define fscrypt_restore_control_page fscrypt_notsupp_restore_control_page
2351#define fscrypt_zeroout_range fscrypt_notsupp_zeroout_range
2352#define fscrypt_process_policy fscrypt_notsupp_process_policy
2353#define fscrypt_get_policy fscrypt_notsupp_get_policy
2354#define fscrypt_has_permitted_context fscrypt_notsupp_has_permitted_context
2355#define fscrypt_inherit_context fscrypt_notsupp_inherit_context
2356#define fscrypt_get_encryption_info fscrypt_notsupp_get_encryption_info
2357#define fscrypt_put_encryption_info fscrypt_notsupp_put_encryption_info
2358#define fscrypt_setup_filename fscrypt_notsupp_setup_filename
2359#define fscrypt_free_filename fscrypt_notsupp_free_filename
2360#define fscrypt_fname_encrypted_size fscrypt_notsupp_fname_encrypted_size
2361#define fscrypt_fname_alloc_buffer fscrypt_notsupp_fname_alloc_buffer
2362#define fscrypt_fname_free_buffer fscrypt_notsupp_fname_free_buffer
2363#define fscrypt_fname_disk_to_usr fscrypt_notsupp_fname_disk_to_usr
2364#define fscrypt_fname_usr_to_disk fscrypt_notsupp_fname_usr_to_disk
2365#endif
2366
2367/* dir.c */
2368extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2369 struct file *,
2370 struct ext4_dir_entry_2 *,
2371 struct buffer_head *, char *, int,
2372 unsigned int);
2373#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2374 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2375 (de), (bh), (buf), (size), (offset)))
2376extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2377 __u32 minor_hash,
2378 struct ext4_dir_entry_2 *dirent,
2379 struct fscrypt_str *ent_name);
2380extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2381extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2382 struct buffer_head *bh,
2383 void *buf, int buf_size,
2384 struct ext4_filename *fname,
2385 struct ext4_dir_entry_2 **dest_de);
2386int ext4_insert_dentry(struct inode *dir,
2387 struct inode *inode,
2388 struct ext4_dir_entry_2 *de,
2389 int buf_size,
2390 struct ext4_filename *fname);
2391static inline void ext4_update_dx_flag(struct inode *inode)
2392{
2393 if (!ext4_has_feature_dir_index(inode->i_sb))
2394 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2395}
2396static unsigned char ext4_filetype_table[] = {
2397 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2398};
2399
2400static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2401{
2402 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2403 return DT_UNKNOWN;
2404
2405 return ext4_filetype_table[filetype];
2406}
2407extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2408 void *buf, int buf_size);
2409
2410/* fsync.c */
2411extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2412
2413/* hash.c */
2414extern int ext4fs_dirhash(const char *name, int len, struct
2415 dx_hash_info *hinfo);
2416
2417/* ialloc.c */
2418extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2419 const struct qstr *qstr, __u32 goal,
2420 uid_t *owner, int handle_type,
2421 unsigned int line_no, int nblocks);
2422
2423#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2424 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2425 0, 0, 0)
2426#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2427 type, nblocks) \
2428 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2429 (type), __LINE__, (nblocks))
2430
2431
2432extern void ext4_free_inode(handle_t *, struct inode *);
2433extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2434extern unsigned long ext4_count_free_inodes(struct super_block *);
2435extern unsigned long ext4_count_dirs(struct super_block *);
2436extern void ext4_check_inodes_bitmap(struct super_block *);
2437extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2438extern int ext4_init_inode_table(struct super_block *sb,
2439 ext4_group_t group, int barrier);
2440extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2441
2442/* mballoc.c */
2443extern const struct file_operations ext4_seq_mb_groups_fops;
2444extern long ext4_mb_stats;
2445extern long ext4_mb_max_to_scan;
2446extern int ext4_mb_init(struct super_block *);
2447extern int ext4_mb_release(struct super_block *);
2448extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2449 struct ext4_allocation_request *, int *);
2450extern int ext4_mb_reserve_blocks(struct super_block *, int);
2451extern void ext4_discard_preallocations(struct inode *);
2452extern int __init ext4_init_mballoc(void);
2453extern void ext4_exit_mballoc(void);
2454extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2455 struct buffer_head *bh, ext4_fsblk_t block,
2456 unsigned long count, int flags);
2457extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2458 ext4_group_t ngroups);
2459extern int ext4_mb_add_groupinfo(struct super_block *sb,
2460 ext4_group_t i, struct ext4_group_desc *desc);
2461extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2462 ext4_fsblk_t block, unsigned long count);
2463extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2464
2465/* inode.c */
2466int ext4_inode_is_fast_symlink(struct inode *inode);
2467struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2468struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2469int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2470 struct buffer_head *bh_result, int create);
2471int ext4_dax_get_block(struct inode *inode, sector_t iblock,
2472 struct buffer_head *bh_result, int create);
2473int ext4_get_block(struct inode *inode, sector_t iblock,
2474 struct buffer_head *bh_result, int create);
2475int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2476 struct buffer_head *bh_result, int create);
2477int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2478 struct buffer_head *bh, int create);
2479int ext4_walk_page_buffers(handle_t *handle,
2480 struct buffer_head *head,
2481 unsigned from,
2482 unsigned to,
2483 int *partial,
2484 int (*fn)(handle_t *handle,
2485 struct buffer_head *bh));
2486int do_journal_get_write_access(handle_t *handle,
2487 struct buffer_head *bh);
2488#define FALL_BACK_TO_NONDELALLOC 1
2489#define CONVERT_INLINE_DATA 2
2490
2491extern struct inode *ext4_iget(struct super_block *, unsigned long);
2492extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2493extern int ext4_write_inode(struct inode *, struct writeback_control *);
2494extern int ext4_setattr(struct dentry *, struct iattr *);
2495extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2496 struct kstat *stat);
2497extern void ext4_evict_inode(struct inode *);
2498extern void ext4_clear_inode(struct inode *);
2499extern int ext4_sync_inode(handle_t *, struct inode *);
2500extern void ext4_dirty_inode(struct inode *, int);
2501extern int ext4_change_inode_journal_flag(struct inode *, int);
2502extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2503extern int ext4_inode_attach_jinode(struct inode *inode);
2504extern int ext4_can_truncate(struct inode *inode);
2505extern void ext4_truncate(struct inode *);
2506extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2507extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2508extern void ext4_set_inode_flags(struct inode *);
2509extern void ext4_get_inode_flags(struct ext4_inode_info *);
2510extern int ext4_alloc_da_blocks(struct inode *inode);
2511extern void ext4_set_aops(struct inode *inode);
2512extern int ext4_writepage_trans_blocks(struct inode *);
2513extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2514extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2515 loff_t lstart, loff_t lend);
2516extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2517extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2518extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2519extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2520extern void ext4_da_update_reserve_space(struct inode *inode,
2521 int used, int quota_claim);
2522extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2523 ext4_fsblk_t pblk, ext4_lblk_t len);
2524extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2525 unsigned int map_len,
2526 struct extent_status *result);
2527
2528/* indirect.c */
2529extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2530 struct ext4_map_blocks *map, int flags);
2531extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2532extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2533extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2534extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2535 ext4_lblk_t start, ext4_lblk_t end);
2536
2537/* ioctl.c */
2538extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2539extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2540
2541/* migrate.c */
2542extern int ext4_ext_migrate(struct inode *);
2543extern int ext4_ind_migrate(struct inode *inode);
2544
2545/* namei.c */
2546extern int ext4_dirent_csum_verify(struct inode *inode,
2547 struct ext4_dir_entry *dirent);
2548extern int ext4_orphan_add(handle_t *, struct inode *);
2549extern int ext4_orphan_del(handle_t *, struct inode *);
2550extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2551 __u32 start_minor_hash, __u32 *next_hash);
2552extern int ext4_search_dir(struct buffer_head *bh,
2553 char *search_buf,
2554 int buf_size,
2555 struct inode *dir,
2556 struct ext4_filename *fname,
2557 const struct qstr *d_name,
2558 unsigned int offset,
2559 struct ext4_dir_entry_2 **res_dir);
2560extern int ext4_generic_delete_entry(handle_t *handle,
2561 struct inode *dir,
2562 struct ext4_dir_entry_2 *de_del,
2563 struct buffer_head *bh,
2564 void *entry_buf,
2565 int buf_size,
2566 int csum_size);
2567extern bool ext4_empty_dir(struct inode *inode);
2568
2569/* resize.c */
2570extern int ext4_group_add(struct super_block *sb,
2571 struct ext4_new_group_data *input);
2572extern int ext4_group_extend(struct super_block *sb,
2573 struct ext4_super_block *es,
2574 ext4_fsblk_t n_blocks_count);
2575extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2576
2577/* super.c */
2578extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2579extern int ext4_calculate_overhead(struct super_block *sb);
2580extern void ext4_superblock_csum_set(struct super_block *sb);
2581extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2582extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2583extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2584 ext4_group_t ngroup);
2585extern const char *ext4_decode_error(struct super_block *sb, int errno,
2586 char nbuf[16]);
2587
2588extern __printf(4, 5)
2589void __ext4_error(struct super_block *, const char *, unsigned int,
2590 const char *, ...);
2591extern __printf(5, 6)
2592void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2593 const char *, ...);
2594extern __printf(5, 6)
2595void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2596 const char *, ...);
2597extern void __ext4_std_error(struct super_block *, const char *,
2598 unsigned int, int);
2599extern __printf(4, 5)
2600void __ext4_abort(struct super_block *, const char *, unsigned int,
2601 const char *, ...);
2602extern __printf(4, 5)
2603void __ext4_warning(struct super_block *, const char *, unsigned int,
2604 const char *, ...);
2605extern __printf(4, 5)
2606void __ext4_warning_inode(const struct inode *inode, const char *function,
2607 unsigned int line, const char *fmt, ...);
2608extern __printf(3, 4)
2609void __ext4_msg(struct super_block *, const char *, const char *, ...);
2610extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2611 const char *, unsigned int, const char *);
2612extern __printf(7, 8)
2613void __ext4_grp_locked_error(const char *, unsigned int,
2614 struct super_block *, ext4_group_t,
2615 unsigned long, ext4_fsblk_t,
2616 const char *, ...);
2617
2618#define EXT4_ERROR_INODE(inode, fmt, a...) \
2619 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2620
2621#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2622 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2623
2624#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2625 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2626
2627#ifdef CONFIG_PRINTK
2628
2629#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2630 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2631#define ext4_error_file(file, func, line, block, fmt, ...) \
2632 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2633#define ext4_error(sb, fmt, ...) \
2634 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2635#define ext4_abort(sb, fmt, ...) \
2636 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2637#define ext4_warning(sb, fmt, ...) \
2638 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2639#define ext4_warning_inode(inode, fmt, ...) \
2640 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2641#define ext4_msg(sb, level, fmt, ...) \
2642 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2643#define dump_mmp_msg(sb, mmp, msg) \
2644 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2645#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2646 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2647 fmt, ##__VA_ARGS__)
2648
2649#else
2650
2651#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2652do { \
2653 no_printk(fmt, ##__VA_ARGS__); \
2654 __ext4_error_inode(inode, "", 0, block, " "); \
2655} while (0)
2656#define ext4_error_file(file, func, line, block, fmt, ...) \
2657do { \
2658 no_printk(fmt, ##__VA_ARGS__); \
2659 __ext4_error_file(file, "", 0, block, " "); \
2660} while (0)
2661#define ext4_error(sb, fmt, ...) \
2662do { \
2663 no_printk(fmt, ##__VA_ARGS__); \
2664 __ext4_error(sb, "", 0, " "); \
2665} while (0)
2666#define ext4_abort(sb, fmt, ...) \
2667do { \
2668 no_printk(fmt, ##__VA_ARGS__); \
2669 __ext4_abort(sb, "", 0, " "); \
2670} while (0)
2671#define ext4_warning(sb, fmt, ...) \
2672do { \
2673 no_printk(fmt, ##__VA_ARGS__); \
2674 __ext4_warning(sb, "", 0, " "); \
2675} while (0)
2676#define ext4_warning_inode(inode, fmt, ...) \
2677do { \
2678 no_printk(fmt, ##__VA_ARGS__); \
2679 __ext4_warning_inode(inode, "", 0, " "); \
2680} while (0)
2681#define ext4_msg(sb, level, fmt, ...) \
2682do { \
2683 no_printk(fmt, ##__VA_ARGS__); \
2684 __ext4_msg(sb, "", " "); \
2685} while (0)
2686#define dump_mmp_msg(sb, mmp, msg) \
2687 __dump_mmp_msg(sb, mmp, "", 0, "")
2688#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2689do { \
2690 no_printk(fmt, ##__VA_ARGS__); \
2691 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2692} while (0)
2693
2694#endif
2695
2696extern void ext4_update_dynamic_rev(struct super_block *sb);
2697extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2698 __u32 compat);
2699extern int ext4_update_rocompat_feature(handle_t *handle,
2700 struct super_block *sb, __u32 rocompat);
2701extern int ext4_update_incompat_feature(handle_t *handle,
2702 struct super_block *sb, __u32 incompat);
2703extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2704 struct ext4_group_desc *bg);
2705extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2706 struct ext4_group_desc *bg);
2707extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2708 struct ext4_group_desc *bg);
2709extern __u32 ext4_free_group_clusters(struct super_block *sb,
2710 struct ext4_group_desc *bg);
2711extern __u32 ext4_free_inodes_count(struct super_block *sb,
2712 struct ext4_group_desc *bg);
2713extern __u32 ext4_used_dirs_count(struct super_block *sb,
2714 struct ext4_group_desc *bg);
2715extern __u32 ext4_itable_unused_count(struct super_block *sb,
2716 struct ext4_group_desc *bg);
2717extern void ext4_block_bitmap_set(struct super_block *sb,
2718 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2719extern void ext4_inode_bitmap_set(struct super_block *sb,
2720 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2721extern void ext4_inode_table_set(struct super_block *sb,
2722 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2723extern void ext4_free_group_clusters_set(struct super_block *sb,
2724 struct ext4_group_desc *bg,
2725 __u32 count);
2726extern void ext4_free_inodes_set(struct super_block *sb,
2727 struct ext4_group_desc *bg, __u32 count);
2728extern void ext4_used_dirs_set(struct super_block *sb,
2729 struct ext4_group_desc *bg, __u32 count);
2730extern void ext4_itable_unused_set(struct super_block *sb,
2731 struct ext4_group_desc *bg, __u32 count);
2732extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2733 struct ext4_group_desc *gdp);
2734extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2735 struct ext4_group_desc *gdp);
2736extern int ext4_register_li_request(struct super_block *sb,
2737 ext4_group_t first_not_zeroed);
2738
2739static inline int ext4_has_group_desc_csum(struct super_block *sb)
2740{
2741 return ext4_has_feature_gdt_csum(sb) ||
2742 EXT4_SB(sb)->s_chksum_driver != NULL;
2743}
2744
2745static inline int ext4_has_metadata_csum(struct super_block *sb)
2746{
2747 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2748 !EXT4_SB(sb)->s_chksum_driver);
2749
2750 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2751}
2752static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2753{
2754 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2755 le32_to_cpu(es->s_blocks_count_lo);
2756}
2757
2758static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2759{
2760 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2761 le32_to_cpu(es->s_r_blocks_count_lo);
2762}
2763
2764static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2765{
2766 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2767 le32_to_cpu(es->s_free_blocks_count_lo);
2768}
2769
2770static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2771 ext4_fsblk_t blk)
2772{
2773 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2774 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2775}
2776
2777static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2778 ext4_fsblk_t blk)
2779{
2780 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2781 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2782}
2783
2784static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2785 ext4_fsblk_t blk)
2786{
2787 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2788 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2789}
2790
2791static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2792{
2793 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2794 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2795 le32_to_cpu(raw_inode->i_size_lo);
2796 else
2797 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2798}
2799
2800static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2801{
2802 raw_inode->i_size_lo = cpu_to_le32(i_size);
2803 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2804}
2805
2806static inline
2807struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2808 ext4_group_t group)
2809{
2810 struct ext4_group_info ***grp_info;
2811 long indexv, indexh;
2812 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2813 grp_info = EXT4_SB(sb)->s_group_info;
2814 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2815 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2816 return grp_info[indexv][indexh];
2817}
2818
2819/*
2820 * Reading s_groups_count requires using smp_rmb() afterwards. See
2821 * the locking protocol documented in the comments of ext4_group_add()
2822 * in resize.c
2823 */
2824static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2825{
2826 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2827
2828 smp_rmb();
2829 return ngroups;
2830}
2831
2832static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2833 ext4_group_t block_group)
2834{
2835 return block_group >> sbi->s_log_groups_per_flex;
2836}
2837
2838static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2839{
2840 return 1 << sbi->s_log_groups_per_flex;
2841}
2842
2843#define ext4_std_error(sb, errno) \
2844do { \
2845 if ((errno)) \
2846 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2847} while (0)
2848
2849#ifdef CONFIG_SMP
2850/* Each CPU can accumulate percpu_counter_batch clusters in their local
2851 * counters. So we need to make sure we have free clusters more
2852 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2853 */
2854#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2855#else
2856#define EXT4_FREECLUSTERS_WATERMARK 0
2857#endif
2858
2859/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2860static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2861{
2862 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2863 !inode_is_locked(inode));
2864 down_write(&EXT4_I(inode)->i_data_sem);
2865 if (newsize > EXT4_I(inode)->i_disksize)
2866 EXT4_I(inode)->i_disksize = newsize;
2867 up_write(&EXT4_I(inode)->i_data_sem);
2868}
2869
2870/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2871static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2872{
2873 int changed = 0;
2874
2875 if (newsize > inode->i_size) {
2876 i_size_write(inode, newsize);
2877 changed = 1;
2878 }
2879 if (newsize > EXT4_I(inode)->i_disksize) {
2880 ext4_update_i_disksize(inode, newsize);
2881 changed |= 2;
2882 }
2883 return changed;
2884}
2885
2886int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2887 loff_t len);
2888
2889struct ext4_group_info {
2890 unsigned long bb_state;
2891 struct rb_root bb_free_root;
2892 ext4_grpblk_t bb_first_free; /* first free block */
2893 ext4_grpblk_t bb_free; /* total free blocks */
2894 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2895 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2896 struct list_head bb_prealloc_list;
2897#ifdef DOUBLE_CHECK
2898 void *bb_bitmap;
2899#endif
2900 struct rw_semaphore alloc_sem;
2901 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2902 * regions, index is order.
2903 * bb_counters[3] = 5 means
2904 * 5 free 8-block regions. */
2905};
2906
2907#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2908#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2909#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2910#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2911
2912#define EXT4_MB_GRP_NEED_INIT(grp) \
2913 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2914#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2915 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2916#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2917 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2918
2919#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2920 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2921#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2922 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2923#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2924 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2925
2926#define EXT4_MAX_CONTENTION 8
2927#define EXT4_CONTENTION_THRESHOLD 2
2928
2929static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2930 ext4_group_t group)
2931{
2932 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2933}
2934
2935/*
2936 * Returns true if the filesystem is busy enough that attempts to
2937 * access the block group locks has run into contention.
2938 */
2939static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2940{
2941 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2942}
2943
2944static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2945{
2946 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2947 if (spin_trylock(lock))
2948 /*
2949 * We're able to grab the lock right away, so drop the
2950 * lock contention counter.
2951 */
2952 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2953 else {
2954 /*
2955 * The lock is busy, so bump the contention counter,
2956 * and then wait on the spin lock.
2957 */
2958 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2959 EXT4_MAX_CONTENTION);
2960 spin_lock(lock);
2961 }
2962}
2963
2964static inline void ext4_unlock_group(struct super_block *sb,
2965 ext4_group_t group)
2966{
2967 spin_unlock(ext4_group_lock_ptr(sb, group));
2968}
2969
2970/*
2971 * Block validity checking
2972 */
2973#define ext4_check_indirect_blockref(inode, bh) \
2974 ext4_check_blockref(__func__, __LINE__, inode, \
2975 (__le32 *)(bh)->b_data, \
2976 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2977
2978#define ext4_ind_check_inode(inode) \
2979 ext4_check_blockref(__func__, __LINE__, inode, \
2980 EXT4_I(inode)->i_data, \
2981 EXT4_NDIR_BLOCKS)
2982
2983/*
2984 * Inodes and files operations
2985 */
2986
2987/* dir.c */
2988extern const struct file_operations ext4_dir_operations;
2989
2990/* file.c */
2991extern const struct inode_operations ext4_file_inode_operations;
2992extern const struct file_operations ext4_file_operations;
2993extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2994
2995/* inline.c */
2996extern int ext4_get_max_inline_size(struct inode *inode);
2997extern int ext4_find_inline_data_nolock(struct inode *inode);
2998extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2999 unsigned int len);
3000extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3001
3002extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3003extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3004 struct inode *inode,
3005 loff_t pos, unsigned len,
3006 unsigned flags,
3007 struct page **pagep);
3008extern int ext4_write_inline_data_end(struct inode *inode,
3009 loff_t pos, unsigned len,
3010 unsigned copied,
3011 struct page *page);
3012extern struct buffer_head *
3013ext4_journalled_write_inline_data(struct inode *inode,
3014 unsigned len,
3015 struct page *page);
3016extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3017 struct inode *inode,
3018 loff_t pos, unsigned len,
3019 unsigned flags,
3020 struct page **pagep,
3021 void **fsdata);
3022extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3023 unsigned len, unsigned copied,
3024 struct page *page);
3025extern int ext4_try_add_inline_entry(handle_t *handle,
3026 struct ext4_filename *fname,
3027 struct inode *dir, struct inode *inode);
3028extern int ext4_try_create_inline_dir(handle_t *handle,
3029 struct inode *parent,
3030 struct inode *inode);
3031extern int ext4_read_inline_dir(struct file *filp,
3032 struct dir_context *ctx,
3033 int *has_inline_data);
3034extern int htree_inlinedir_to_tree(struct file *dir_file,
3035 struct inode *dir, ext4_lblk_t block,
3036 struct dx_hash_info *hinfo,
3037 __u32 start_hash, __u32 start_minor_hash,
3038 int *has_inline_data);
3039extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3040 struct ext4_filename *fname,
3041 const struct qstr *d_name,
3042 struct ext4_dir_entry_2 **res_dir,
3043 int *has_inline_data);
3044extern int ext4_delete_inline_entry(handle_t *handle,
3045 struct inode *dir,
3046 struct ext4_dir_entry_2 *de_del,
3047 struct buffer_head *bh,
3048 int *has_inline_data);
3049extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3050extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3051 struct ext4_dir_entry_2 **parent_de,
3052 int *retval);
3053extern int ext4_inline_data_fiemap(struct inode *inode,
3054 struct fiemap_extent_info *fieinfo,
3055 int *has_inline, __u64 start, __u64 len);
3056extern int ext4_try_to_evict_inline_data(handle_t *handle,
3057 struct inode *inode,
3058 int needed);
3059extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3060
3061extern int ext4_convert_inline_data(struct inode *inode);
3062
3063static inline int ext4_has_inline_data(struct inode *inode)
3064{
3065 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3066 EXT4_I(inode)->i_inline_off;
3067}
3068
3069/* namei.c */
3070extern const struct inode_operations ext4_dir_inode_operations;
3071extern const struct inode_operations ext4_special_inode_operations;
3072extern struct dentry *ext4_get_parent(struct dentry *child);
3073extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3074 struct ext4_dir_entry_2 *de,
3075 int blocksize, int csum_size,
3076 unsigned int parent_ino, int dotdot_real_len);
3077extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3078 unsigned int blocksize);
3079extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3080 struct inode *inode,
3081 struct buffer_head *bh);
3082#define S_SHIFT 12
3083static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3084 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3085 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3086 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3087 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3088 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3089 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3090 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3091};
3092
3093static inline void ext4_set_de_type(struct super_block *sb,
3094 struct ext4_dir_entry_2 *de,
3095 umode_t mode) {
3096 if (ext4_has_feature_filetype(sb))
3097 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3098}
3099
3100/* readpages.c */
3101extern int ext4_mpage_readpages(struct address_space *mapping,
3102 struct list_head *pages, struct page *page,
3103 unsigned nr_pages);
3104
3105/* symlink.c */
3106extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3107extern const struct inode_operations ext4_symlink_inode_operations;
3108extern const struct inode_operations ext4_fast_symlink_inode_operations;
3109
3110/* sysfs.c */
3111extern int ext4_register_sysfs(struct super_block *sb);
3112extern void ext4_unregister_sysfs(struct super_block *sb);
3113extern int __init ext4_init_sysfs(void);
3114extern void ext4_exit_sysfs(void);
3115
3116/* block_validity */
3117extern void ext4_release_system_zone(struct super_block *sb);
3118extern int ext4_setup_system_zone(struct super_block *sb);
3119extern int __init ext4_init_system_zone(void);
3120extern void ext4_exit_system_zone(void);
3121extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3122 ext4_fsblk_t start_blk,
3123 unsigned int count);
3124extern int ext4_check_blockref(const char *, unsigned int,
3125 struct inode *, __le32 *, unsigned int);
3126
3127/* extents.c */
3128struct ext4_ext_path;
3129struct ext4_extent;
3130
3131/*
3132 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3133 * __le32.
3134 */
3135#define EXT_MAX_BLOCKS 0xffffffff
3136
3137extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3138extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3139extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3140extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3141 struct ext4_map_blocks *map, int flags);
3142extern void ext4_ext_truncate(handle_t *, struct inode *);
3143extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3144 ext4_lblk_t end);
3145extern void ext4_ext_init(struct super_block *);
3146extern void ext4_ext_release(struct super_block *);
3147extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3148 loff_t len);
3149extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3150 loff_t offset, ssize_t len);
3151extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3152 struct ext4_map_blocks *map, int flags);
3153extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3154 ext4_lblk_t lblocks);
3155extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3156 int num,
3157 struct ext4_ext_path *path);
3158extern int ext4_can_extents_be_merged(struct inode *inode,
3159 struct ext4_extent *ex1,
3160 struct ext4_extent *ex2);
3161extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3162 struct ext4_ext_path **,
3163 struct ext4_extent *, int);
3164extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3165 struct ext4_ext_path **,
3166 int flags);
3167extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3168extern int ext4_ext_check_inode(struct inode *inode);
3169extern int ext4_find_delalloc_range(struct inode *inode,
3170 ext4_lblk_t lblk_start,
3171 ext4_lblk_t lblk_end);
3172extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3173extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3174extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3175 __u64 start, __u64 len);
3176extern int ext4_ext_precache(struct inode *inode);
3177extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3178extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3179extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3180 struct inode *inode2, ext4_lblk_t lblk1,
3181 ext4_lblk_t lblk2, ext4_lblk_t count,
3182 int mark_unwritten,int *err);
3183
3184/* move_extent.c */
3185extern void ext4_double_down_write_data_sem(struct inode *first,
3186 struct inode *second);
3187extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3188 struct inode *donor_inode);
3189extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3190 __u64 start_orig, __u64 start_donor,
3191 __u64 len, __u64 *moved_len);
3192
3193/* page-io.c */
3194extern int __init ext4_init_pageio(void);
3195extern void ext4_exit_pageio(void);
3196extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3197extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3198extern int ext4_put_io_end(ext4_io_end_t *io_end);
3199extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3200extern void ext4_io_submit_init(struct ext4_io_submit *io,
3201 struct writeback_control *wbc);
3202extern void ext4_end_io_rsv_work(struct work_struct *work);
3203extern void ext4_io_submit(struct ext4_io_submit *io);
3204extern int ext4_bio_write_page(struct ext4_io_submit *io,
3205 struct page *page,
3206 int len,
3207 struct writeback_control *wbc,
3208 bool keep_towrite);
3209
3210/* mmp.c */
3211extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3212
3213/*
3214 * Add new method to test whether block and inode bitmaps are properly
3215 * initialized. With uninit_bg reading the block from disk is not enough
3216 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3217 */
3218#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3219
3220static inline int bitmap_uptodate(struct buffer_head *bh)
3221{
3222 return (buffer_uptodate(bh) &&
3223 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3224}
3225static inline void set_bitmap_uptodate(struct buffer_head *bh)
3226{
3227 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3228}
3229
3230/*
3231 * Disable DIO read nolock optimization, so new dioreaders will be forced
3232 * to grab i_mutex
3233 */
3234static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3235{
3236 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3237 smp_mb();
3238}
3239static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3240{
3241 smp_mb();
3242 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3243}
3244
3245#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3246
3247/* For ioend & aio unwritten conversion wait queues */
3248#define EXT4_WQ_HASH_SZ 37
3249#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3250 EXT4_WQ_HASH_SZ])
3251extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3252
3253#define EXT4_RESIZING 0
3254extern int ext4_resize_begin(struct super_block *sb);
3255extern void ext4_resize_end(struct super_block *sb);
3256
3257static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3258 struct ext4_io_end *io_end)
3259{
3260 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3261 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3262 atomic_inc(&EXT4_I(inode)->i_unwritten);
3263 }
3264}
3265
3266static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3267{
3268 struct inode *inode = io_end->inode;
3269
3270 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3271 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3272 /* Wake up anyone waiting on unwritten extent conversion */
3273 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3274 wake_up_all(ext4_ioend_wq(inode));
3275 }
3276}
3277
3278static inline bool ext4_aligned_io(struct inode *inode, loff_t off, loff_t len)
3279{
3280 int blksize = 1 << inode->i_blkbits;
3281
3282 return IS_ALIGNED(off, blksize) && IS_ALIGNED(len, blksize);
3283}
3284
3285#endif /* __KERNEL__ */
3286
3287#define EFSBADCRC EBADMSG /* Bad CRC detected */
3288#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3289
3290#endif /* _EXT4_H */