1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 44 #include <linux/fscrypt.h> 45 #include <linux/fsverity.h> 46 47 #include <linux/compiler.h> 48 49 /* 50 * The fourth extended filesystem constants/structures 51 */ 52 53 /* 54 * with AGGRESSIVE_CHECK allocator runs consistency checks over 55 * structures. these checks slow things down a lot 56 */ 57 #define AGGRESSIVE_CHECK__ 58 59 /* 60 * with DOUBLE_CHECK defined mballoc creates persistent in-core 61 * bitmaps, maintains and uses them to check for double allocations 62 */ 63 #define DOUBLE_CHECK__ 64 65 /* 66 * Define EXT4FS_DEBUG to produce debug messages 67 */ 68 #undef EXT4FS_DEBUG 69 70 /* 71 * Debug code 72 */ 73 #ifdef EXT4FS_DEBUG 74 #define ext4_debug(f, a...) \ 75 do { \ 76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 77 __FILE__, __LINE__, __func__); \ 78 printk(KERN_DEBUG f, ## a); \ 79 } while (0) 80 #else 81 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 82 #endif 83 84 /* 85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 86 */ 87 #define EXT_DEBUG__ 88 89 /* 90 * Dynamic printk for controlled extents debugging. 91 */ 92 #ifdef CONFIG_EXT4_DEBUG 93 #define ext_debug(ino, fmt, ...) \ 94 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 95 current->comm, task_pid_nr(current), \ 96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 97 __func__, ##__VA_ARGS__) 98 #else 99 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 100 #endif 101 102 /* data type for block offset of block group */ 103 typedef int ext4_grpblk_t; 104 105 /* data type for filesystem-wide blocks number */ 106 typedef unsigned long long ext4_fsblk_t; 107 108 /* data type for file logical block number */ 109 typedef __u32 ext4_lblk_t; 110 111 /* data type for block group number */ 112 typedef unsigned int ext4_group_t; 113 114 enum SHIFT_DIRECTION { 115 SHIFT_LEFT = 0, 116 SHIFT_RIGHT, 117 }; 118 119 /* 120 * Flags used in mballoc's allocation_context flags field. 121 * 122 * Also used to show what's going on for debugging purposes when the 123 * flag field is exported via the traceport interface 124 */ 125 126 /* prefer goal again. length */ 127 #define EXT4_MB_HINT_MERGE 0x0001 128 /* blocks already reserved */ 129 #define EXT4_MB_HINT_RESERVED 0x0002 130 /* metadata is being allocated */ 131 #define EXT4_MB_HINT_METADATA 0x0004 132 /* first blocks in the file */ 133 #define EXT4_MB_HINT_FIRST 0x0008 134 /* search for the best chunk */ 135 #define EXT4_MB_HINT_BEST 0x0010 136 /* data is being allocated */ 137 #define EXT4_MB_HINT_DATA 0x0020 138 /* don't preallocate (for tails) */ 139 #define EXT4_MB_HINT_NOPREALLOC 0x0040 140 /* allocate for locality group */ 141 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 142 /* allocate goal blocks or none */ 143 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 144 /* goal is meaningful */ 145 #define EXT4_MB_HINT_TRY_GOAL 0x0200 146 /* blocks already pre-reserved by delayed allocation */ 147 #define EXT4_MB_DELALLOC_RESERVED 0x0400 148 /* We are doing stream allocation */ 149 #define EXT4_MB_STREAM_ALLOC 0x0800 150 /* Use reserved root blocks if needed */ 151 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 152 /* Use blocks from reserved pool */ 153 #define EXT4_MB_USE_RESERVED 0x2000 154 /* Do strict check for free blocks while retrying block allocation */ 155 #define EXT4_MB_STRICT_CHECK 0x4000 156 157 struct ext4_allocation_request { 158 /* target inode for block we're allocating */ 159 struct inode *inode; 160 /* how many blocks we want to allocate */ 161 unsigned int len; 162 /* logical block in target inode */ 163 ext4_lblk_t logical; 164 /* the closest logical allocated block to the left */ 165 ext4_lblk_t lleft; 166 /* the closest logical allocated block to the right */ 167 ext4_lblk_t lright; 168 /* phys. target (a hint) */ 169 ext4_fsblk_t goal; 170 /* phys. block for the closest logical allocated block to the left */ 171 ext4_fsblk_t pleft; 172 /* phys. block for the closest logical allocated block to the right */ 173 ext4_fsblk_t pright; 174 /* flags. see above EXT4_MB_HINT_* */ 175 unsigned int flags; 176 }; 177 178 /* 179 * Logical to physical block mapping, used by ext4_map_blocks() 180 * 181 * This structure is used to pass requests into ext4_map_blocks() as 182 * well as to store the information returned by ext4_map_blocks(). It 183 * takes less room on the stack than a struct buffer_head. 184 */ 185 #define EXT4_MAP_NEW BIT(BH_New) 186 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 187 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 188 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 189 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 190 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 191 192 struct ext4_map_blocks { 193 ext4_fsblk_t m_pblk; 194 ext4_lblk_t m_lblk; 195 unsigned int m_len; 196 unsigned int m_flags; 197 }; 198 199 /* 200 * Block validity checking, system zone rbtree. 201 */ 202 struct ext4_system_blocks { 203 struct rb_root root; 204 struct rcu_head rcu; 205 }; 206 207 /* 208 * Flags for ext4_io_end->flags 209 */ 210 #define EXT4_IO_END_UNWRITTEN 0x0001 211 212 struct ext4_io_end_vec { 213 struct list_head list; /* list of io_end_vec */ 214 loff_t offset; /* offset in the file */ 215 ssize_t size; /* size of the extent */ 216 }; 217 218 /* 219 * For converting unwritten extents on a work queue. 'handle' is used for 220 * buffered writeback. 221 */ 222 typedef struct ext4_io_end { 223 struct list_head list; /* per-file finished IO list */ 224 handle_t *handle; /* handle reserved for extent 225 * conversion */ 226 struct inode *inode; /* file being written to */ 227 struct bio *bio; /* Linked list of completed 228 * bios covering the extent */ 229 unsigned int flag; /* unwritten or not */ 230 atomic_t count; /* reference counter */ 231 struct list_head list_vec; /* list of ext4_io_end_vec */ 232 } ext4_io_end_t; 233 234 struct ext4_io_submit { 235 struct writeback_control *io_wbc; 236 struct bio *io_bio; 237 ext4_io_end_t *io_end; 238 sector_t io_next_block; 239 }; 240 241 /* 242 * Special inodes numbers 243 */ 244 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 245 #define EXT4_ROOT_INO 2 /* Root inode */ 246 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 247 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 248 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 249 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 250 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 251 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 252 253 /* First non-reserved inode for old ext4 filesystems */ 254 #define EXT4_GOOD_OLD_FIRST_INO 11 255 256 /* 257 * Maximal count of links to a file 258 */ 259 #define EXT4_LINK_MAX 65000 260 261 /* 262 * Macro-instructions used to manage several block sizes 263 */ 264 #define EXT4_MIN_BLOCK_SIZE 1024 265 #define EXT4_MAX_BLOCK_SIZE 65536 266 #define EXT4_MIN_BLOCK_LOG_SIZE 10 267 #define EXT4_MAX_BLOCK_LOG_SIZE 16 268 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 269 #ifdef __KERNEL__ 270 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 271 #else 272 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 273 #endif 274 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 275 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 276 EXT4_SB(s)->s_cluster_bits) 277 #ifdef __KERNEL__ 278 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 279 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 280 #else 281 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 282 #endif 283 #ifdef __KERNEL__ 284 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 285 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 286 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 287 #else 288 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 289 EXT4_GOOD_OLD_INODE_SIZE : \ 290 (s)->s_inode_size) 291 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 292 EXT4_GOOD_OLD_FIRST_INO : \ 293 (s)->s_first_ino) 294 #endif 295 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 296 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 297 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 298 blkbits)) 299 300 /* Translate a block number to a cluster number */ 301 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 302 /* Translate a cluster number to a block number */ 303 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 304 /* Translate # of blks to # of clusters */ 305 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 306 (sbi)->s_cluster_bits) 307 /* Mask out the low bits to get the starting block of the cluster */ 308 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 309 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 310 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 311 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 312 /* Fill in the low bits to get the last block of the cluster */ 313 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 314 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 315 /* Get the cluster offset */ 316 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 317 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 318 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 319 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 320 321 /* 322 * Structure of a blocks group descriptor 323 */ 324 struct ext4_group_desc 325 { 326 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 327 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 328 __le32 bg_inode_table_lo; /* Inodes table block */ 329 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 330 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 331 __le16 bg_used_dirs_count_lo; /* Directories count */ 332 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 333 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 334 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 335 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 336 __le16 bg_itable_unused_lo; /* Unused inodes count */ 337 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 338 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 339 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 340 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 341 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 342 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 343 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 344 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 345 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 346 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 347 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 348 __u32 bg_reserved; 349 }; 350 351 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 352 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 353 sizeof(__le16)) 354 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 355 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 356 sizeof(__le16)) 357 358 /* 359 * Structure of a flex block group info 360 */ 361 362 struct flex_groups { 363 atomic64_t free_clusters; 364 atomic_t free_inodes; 365 atomic_t used_dirs; 366 }; 367 368 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 369 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 370 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 371 372 /* 373 * Macro-instructions used to manage group descriptors 374 */ 375 #define EXT4_MIN_DESC_SIZE 32 376 #define EXT4_MIN_DESC_SIZE_64BIT 64 377 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 378 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 379 #ifdef __KERNEL__ 380 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 381 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 382 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 383 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 384 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 385 #else 386 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 387 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 388 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 389 #endif 390 391 /* 392 * Constants relative to the data blocks 393 */ 394 #define EXT4_NDIR_BLOCKS 12 395 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 396 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 397 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 398 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 399 400 /* 401 * Inode flags 402 */ 403 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 404 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 405 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 406 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 407 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 408 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 409 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 410 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 411 /* Reserved for compression usage... */ 412 #define EXT4_DIRTY_FL 0x00000100 413 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 414 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 415 /* nb: was previously EXT2_ECOMPR_FL */ 416 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 417 /* End compression flags --- maybe not all used */ 418 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 419 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 420 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 421 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 422 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 423 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 424 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 425 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 426 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 427 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 428 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 429 430 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 431 432 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 433 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 434 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 435 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 436 437 /* User modifiable flags */ 438 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 439 EXT4_UNRM_FL | \ 440 EXT4_COMPR_FL | \ 441 EXT4_SYNC_FL | \ 442 EXT4_IMMUTABLE_FL | \ 443 EXT4_APPEND_FL | \ 444 EXT4_NODUMP_FL | \ 445 EXT4_NOATIME_FL | \ 446 EXT4_JOURNAL_DATA_FL | \ 447 EXT4_NOTAIL_FL | \ 448 EXT4_DIRSYNC_FL | \ 449 EXT4_TOPDIR_FL | \ 450 EXT4_EXTENTS_FL | \ 451 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 452 EXT4_DAX_FL | \ 453 EXT4_PROJINHERIT_FL | \ 454 EXT4_CASEFOLD_FL) 455 456 /* User visible flags */ 457 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 458 EXT4_DIRTY_FL | \ 459 EXT4_COMPRBLK_FL | \ 460 EXT4_NOCOMPR_FL | \ 461 EXT4_ENCRYPT_FL | \ 462 EXT4_INDEX_FL | \ 463 EXT4_VERITY_FL | \ 464 EXT4_INLINE_DATA_FL) 465 466 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */ 467 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 468 EXT4_IMMUTABLE_FL | \ 469 EXT4_APPEND_FL | \ 470 EXT4_NODUMP_FL | \ 471 EXT4_NOATIME_FL | \ 472 EXT4_PROJINHERIT_FL | \ 473 EXT4_DAX_FL) 474 475 /* Flags that should be inherited by new inodes from their parent. */ 476 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 477 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 478 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 479 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 480 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 481 EXT4_DAX_FL) 482 483 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 484 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 485 EXT4_PROJINHERIT_FL)) 486 487 /* Flags that are appropriate for non-directories/regular files. */ 488 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 489 490 /* The only flags that should be swapped */ 491 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 492 493 /* Flags which are mutually exclusive to DAX */ 494 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 495 EXT4_JOURNAL_DATA_FL) 496 497 /* Mask out flags that are inappropriate for the given type of inode. */ 498 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 499 { 500 if (S_ISDIR(mode)) 501 return flags; 502 else if (S_ISREG(mode)) 503 return flags & EXT4_REG_FLMASK; 504 else 505 return flags & EXT4_OTHER_FLMASK; 506 } 507 508 /* 509 * Inode flags used for atomic set/get 510 */ 511 enum { 512 EXT4_INODE_SECRM = 0, /* Secure deletion */ 513 EXT4_INODE_UNRM = 1, /* Undelete */ 514 EXT4_INODE_COMPR = 2, /* Compress file */ 515 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 516 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 517 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 518 EXT4_INODE_NODUMP = 6, /* do not dump file */ 519 EXT4_INODE_NOATIME = 7, /* do not update atime */ 520 /* Reserved for compression usage... */ 521 EXT4_INODE_DIRTY = 8, 522 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 523 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 524 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 525 /* End compression flags --- maybe not all used */ 526 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 527 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 528 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 529 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 530 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 531 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 532 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 533 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 534 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 535 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 536 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 537 EXT4_INODE_DAX = 25, /* Inode is DAX */ 538 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 539 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 540 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 541 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 542 }; 543 544 /* 545 * Since it's pretty easy to mix up bit numbers and hex values, we use a 546 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 547 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 548 * any extra space in the compiled kernel image, otherwise, the build will fail. 549 * It's important that these values are the same, since we are using 550 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 551 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 552 * values found in ext2, ext3 and ext4 filesystems, and of course the values 553 * defined in e2fsprogs. 554 * 555 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 556 */ 557 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 558 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 559 560 static inline void ext4_check_flag_values(void) 561 { 562 CHECK_FLAG_VALUE(SECRM); 563 CHECK_FLAG_VALUE(UNRM); 564 CHECK_FLAG_VALUE(COMPR); 565 CHECK_FLAG_VALUE(SYNC); 566 CHECK_FLAG_VALUE(IMMUTABLE); 567 CHECK_FLAG_VALUE(APPEND); 568 CHECK_FLAG_VALUE(NODUMP); 569 CHECK_FLAG_VALUE(NOATIME); 570 CHECK_FLAG_VALUE(DIRTY); 571 CHECK_FLAG_VALUE(COMPRBLK); 572 CHECK_FLAG_VALUE(NOCOMPR); 573 CHECK_FLAG_VALUE(ENCRYPT); 574 CHECK_FLAG_VALUE(INDEX); 575 CHECK_FLAG_VALUE(IMAGIC); 576 CHECK_FLAG_VALUE(JOURNAL_DATA); 577 CHECK_FLAG_VALUE(NOTAIL); 578 CHECK_FLAG_VALUE(DIRSYNC); 579 CHECK_FLAG_VALUE(TOPDIR); 580 CHECK_FLAG_VALUE(HUGE_FILE); 581 CHECK_FLAG_VALUE(EXTENTS); 582 CHECK_FLAG_VALUE(VERITY); 583 CHECK_FLAG_VALUE(EA_INODE); 584 CHECK_FLAG_VALUE(INLINE_DATA); 585 CHECK_FLAG_VALUE(PROJINHERIT); 586 CHECK_FLAG_VALUE(CASEFOLD); 587 CHECK_FLAG_VALUE(RESERVED); 588 } 589 590 /* Used to pass group descriptor data when online resize is done */ 591 struct ext4_new_group_input { 592 __u32 group; /* Group number for this data */ 593 __u64 block_bitmap; /* Absolute block number of block bitmap */ 594 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 595 __u64 inode_table; /* Absolute block number of inode table start */ 596 __u32 blocks_count; /* Total number of blocks in this group */ 597 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 598 __u16 unused; 599 }; 600 601 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 602 struct compat_ext4_new_group_input { 603 u32 group; 604 compat_u64 block_bitmap; 605 compat_u64 inode_bitmap; 606 compat_u64 inode_table; 607 u32 blocks_count; 608 u16 reserved_blocks; 609 u16 unused; 610 }; 611 #endif 612 613 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 614 struct ext4_new_group_data { 615 __u32 group; 616 __u64 block_bitmap; 617 __u64 inode_bitmap; 618 __u64 inode_table; 619 __u32 blocks_count; 620 __u16 reserved_blocks; 621 __u16 mdata_blocks; 622 __u32 free_clusters_count; 623 }; 624 625 /* Indexes used to index group tables in ext4_new_group_data */ 626 enum { 627 BLOCK_BITMAP = 0, /* block bitmap */ 628 INODE_BITMAP, /* inode bitmap */ 629 INODE_TABLE, /* inode tables */ 630 GROUP_TABLE_COUNT, 631 }; 632 633 /* 634 * Flags used by ext4_map_blocks() 635 */ 636 /* Allocate any needed blocks and/or convert an unwritten 637 extent to be an initialized ext4 */ 638 #define EXT4_GET_BLOCKS_CREATE 0x0001 639 /* Request the creation of an unwritten extent */ 640 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 641 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 642 EXT4_GET_BLOCKS_CREATE) 643 /* Caller is from the delayed allocation writeout path 644 * finally doing the actual allocation of delayed blocks */ 645 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 646 /* caller is from the direct IO path, request to creation of an 647 unwritten extents if not allocated, split the unwritten 648 extent if blocks has been preallocated already*/ 649 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 650 #define EXT4_GET_BLOCKS_CONVERT 0x0010 651 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 652 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 653 /* Convert extent to initialized after IO complete */ 654 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 655 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 656 /* Eventual metadata allocation (due to growing extent tree) 657 * should not fail, so try to use reserved blocks for that.*/ 658 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 659 /* Don't normalize allocation size (used for fallocate) */ 660 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 661 /* Convert written extents to unwritten */ 662 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 663 /* Write zeros to newly created written extents */ 664 #define EXT4_GET_BLOCKS_ZERO 0x0200 665 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 666 EXT4_GET_BLOCKS_ZERO) 667 /* Caller will submit data before dropping transaction handle. This 668 * allows jbd2 to avoid submitting data before commit. */ 669 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 670 671 /* 672 * The bit position of these flags must not overlap with any of the 673 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 674 * read_extent_tree_block(), ext4_split_extent_at(), 675 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 676 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 677 * caching the extents when reading from the extent tree while a 678 * truncate or punch hole operation is in progress. 679 */ 680 #define EXT4_EX_NOCACHE 0x40000000 681 #define EXT4_EX_FORCE_CACHE 0x20000000 682 #define EXT4_EX_NOFAIL 0x10000000 683 684 /* 685 * Flags used by ext4_free_blocks 686 */ 687 #define EXT4_FREE_BLOCKS_METADATA 0x0001 688 #define EXT4_FREE_BLOCKS_FORGET 0x0002 689 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 690 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 691 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 692 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 693 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 694 695 /* 696 * ioctl commands 697 */ 698 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 699 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 700 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 701 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 702 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 703 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 704 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 705 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 706 #define EXT4_IOC_MIGRATE _IO('f', 9) 707 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 708 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 709 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 710 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 711 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 712 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 713 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 714 /* ioctl codes 19--39 are reserved for fscrypt */ 715 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 716 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 717 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 718 719 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 720 721 /* 722 * Flags for going down operation 723 */ 724 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 725 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 726 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 727 728 /* 729 * Flags returned by EXT4_IOC_GETSTATE 730 * 731 * We only expose to userspace a subset of the state flags in 732 * i_state_flags 733 */ 734 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 735 #define EXT4_STATE_FLAG_NEW 0x00000002 736 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 737 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 738 739 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 740 /* 741 * ioctl commands in 32 bit emulation 742 */ 743 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 744 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 745 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 746 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 747 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 748 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 749 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 750 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 751 #endif 752 753 /* 754 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 755 * It indicates that the entry in extent status cache is for a hole. 756 */ 757 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 758 759 /* Max physical block we can address w/o extents */ 760 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 761 762 /* Max logical block we can support */ 763 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 764 765 /* 766 * Structure of an inode on the disk 767 */ 768 struct ext4_inode { 769 __le16 i_mode; /* File mode */ 770 __le16 i_uid; /* Low 16 bits of Owner Uid */ 771 __le32 i_size_lo; /* Size in bytes */ 772 __le32 i_atime; /* Access time */ 773 __le32 i_ctime; /* Inode Change time */ 774 __le32 i_mtime; /* Modification time */ 775 __le32 i_dtime; /* Deletion Time */ 776 __le16 i_gid; /* Low 16 bits of Group Id */ 777 __le16 i_links_count; /* Links count */ 778 __le32 i_blocks_lo; /* Blocks count */ 779 __le32 i_flags; /* File flags */ 780 union { 781 struct { 782 __le32 l_i_version; 783 } linux1; 784 struct { 785 __u32 h_i_translator; 786 } hurd1; 787 struct { 788 __u32 m_i_reserved1; 789 } masix1; 790 } osd1; /* OS dependent 1 */ 791 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 792 __le32 i_generation; /* File version (for NFS) */ 793 __le32 i_file_acl_lo; /* File ACL */ 794 __le32 i_size_high; 795 __le32 i_obso_faddr; /* Obsoleted fragment address */ 796 union { 797 struct { 798 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 799 __le16 l_i_file_acl_high; 800 __le16 l_i_uid_high; /* these 2 fields */ 801 __le16 l_i_gid_high; /* were reserved2[0] */ 802 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 803 __le16 l_i_reserved; 804 } linux2; 805 struct { 806 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 807 __u16 h_i_mode_high; 808 __u16 h_i_uid_high; 809 __u16 h_i_gid_high; 810 __u32 h_i_author; 811 } hurd2; 812 struct { 813 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 814 __le16 m_i_file_acl_high; 815 __u32 m_i_reserved2[2]; 816 } masix2; 817 } osd2; /* OS dependent 2 */ 818 __le16 i_extra_isize; 819 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 820 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 821 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 822 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 823 __le32 i_crtime; /* File Creation time */ 824 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 825 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 826 __le32 i_projid; /* Project ID */ 827 }; 828 829 struct move_extent { 830 __u32 reserved; /* should be zero */ 831 __u32 donor_fd; /* donor file descriptor */ 832 __u64 orig_start; /* logical start offset in block for orig */ 833 __u64 donor_start; /* logical start offset in block for donor */ 834 __u64 len; /* block length to be moved */ 835 __u64 moved_len; /* moved block length */ 836 }; 837 838 #define EXT4_EPOCH_BITS 2 839 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 840 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 841 842 /* 843 * Extended fields will fit into an inode if the filesystem was formatted 844 * with large inodes (-I 256 or larger) and there are not currently any EAs 845 * consuming all of the available space. For new inodes we always reserve 846 * enough space for the kernel's known extended fields, but for inodes 847 * created with an old kernel this might not have been the case. None of 848 * the extended inode fields is critical for correct filesystem operation. 849 * This macro checks if a certain field fits in the inode. Note that 850 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 851 */ 852 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 853 ((offsetof(typeof(*ext4_inode), field) + \ 854 sizeof((ext4_inode)->field)) \ 855 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 856 (einode)->i_extra_isize)) \ 857 858 /* 859 * We use an encoding that preserves the times for extra epoch "00": 860 * 861 * extra msb of adjust for signed 862 * epoch 32-bit 32-bit tv_sec to 863 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 864 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 865 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 866 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 867 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 868 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 869 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 870 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 871 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 872 * 873 * Note that previous versions of the kernel on 64-bit systems would 874 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 875 * 1970. e2fsck will correct this, assuming that it is run on the 876 * affected filesystem before 2242. 877 */ 878 879 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 880 { 881 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 882 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 883 } 884 885 static inline void ext4_decode_extra_time(struct timespec64 *time, 886 __le32 extra) 887 { 888 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 889 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 890 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 891 } 892 893 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 894 do { \ 895 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 896 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 897 (raw_inode)->xtime ## _extra = \ 898 ext4_encode_extra_time(&(inode)->xtime); \ 899 } \ 900 else \ 901 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 902 } while (0) 903 904 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 905 do { \ 906 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 907 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 908 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 909 (raw_inode)->xtime ## _extra = \ 910 ext4_encode_extra_time(&(einode)->xtime); \ 911 } while (0) 912 913 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 914 do { \ 915 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 916 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 917 ext4_decode_extra_time(&(inode)->xtime, \ 918 raw_inode->xtime ## _extra); \ 919 } \ 920 else \ 921 (inode)->xtime.tv_nsec = 0; \ 922 } while (0) 923 924 925 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 926 do { \ 927 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 928 (einode)->xtime.tv_sec = \ 929 (signed)le32_to_cpu((raw_inode)->xtime); \ 930 else \ 931 (einode)->xtime.tv_sec = 0; \ 932 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 933 ext4_decode_extra_time(&(einode)->xtime, \ 934 raw_inode->xtime ## _extra); \ 935 else \ 936 (einode)->xtime.tv_nsec = 0; \ 937 } while (0) 938 939 #define i_disk_version osd1.linux1.l_i_version 940 941 #if defined(__KERNEL__) || defined(__linux__) 942 #define i_reserved1 osd1.linux1.l_i_reserved1 943 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 944 #define i_blocks_high osd2.linux2.l_i_blocks_high 945 #define i_uid_low i_uid 946 #define i_gid_low i_gid 947 #define i_uid_high osd2.linux2.l_i_uid_high 948 #define i_gid_high osd2.linux2.l_i_gid_high 949 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 950 951 #elif defined(__GNU__) 952 953 #define i_translator osd1.hurd1.h_i_translator 954 #define i_uid_high osd2.hurd2.h_i_uid_high 955 #define i_gid_high osd2.hurd2.h_i_gid_high 956 #define i_author osd2.hurd2.h_i_author 957 958 #elif defined(__masix__) 959 960 #define i_reserved1 osd1.masix1.m_i_reserved1 961 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 962 #define i_reserved2 osd2.masix2.m_i_reserved2 963 964 #endif /* defined(__KERNEL__) || defined(__linux__) */ 965 966 #include "extents_status.h" 967 968 /* 969 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 970 * 971 * These are needed to avoid lockdep false positives when we need to 972 * allocate blocks to the quota inode during ext4_map_blocks(), while 973 * holding i_data_sem for a normal (non-quota) inode. Since we don't 974 * do quota tracking for the quota inode, this avoids deadlock (as 975 * well as infinite recursion, since it isn't turtles all the way 976 * down...) 977 * 978 * I_DATA_SEM_NORMAL - Used for most inodes 979 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 980 * where the second inode has larger inode number 981 * than the first 982 * I_DATA_SEM_QUOTA - Used for quota inodes only 983 */ 984 enum { 985 I_DATA_SEM_NORMAL = 0, 986 I_DATA_SEM_OTHER, 987 I_DATA_SEM_QUOTA, 988 }; 989 990 991 /* 992 * fourth extended file system inode data in memory 993 */ 994 struct ext4_inode_info { 995 __le32 i_data[15]; /* unconverted */ 996 __u32 i_dtime; 997 ext4_fsblk_t i_file_acl; 998 999 /* 1000 * i_block_group is the number of the block group which contains 1001 * this file's inode. Constant across the lifetime of the inode, 1002 * it is used for making block allocation decisions - we try to 1003 * place a file's data blocks near its inode block, and new inodes 1004 * near to their parent directory's inode. 1005 */ 1006 ext4_group_t i_block_group; 1007 ext4_lblk_t i_dir_start_lookup; 1008 #if (BITS_PER_LONG < 64) 1009 unsigned long i_state_flags; /* Dynamic state flags */ 1010 #endif 1011 unsigned long i_flags; 1012 1013 /* 1014 * Extended attributes can be read independently of the main file 1015 * data. Taking i_mutex even when reading would cause contention 1016 * between readers of EAs and writers of regular file data, so 1017 * instead we synchronize on xattr_sem when reading or changing 1018 * EAs. 1019 */ 1020 struct rw_semaphore xattr_sem; 1021 1022 struct list_head i_orphan; /* unlinked but open inodes */ 1023 1024 /* 1025 * i_disksize keeps track of what the inode size is ON DISK, not 1026 * in memory. During truncate, i_size is set to the new size by 1027 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1028 * set i_disksize to 0 until the truncate is actually under way. 1029 * 1030 * The intent is that i_disksize always represents the blocks which 1031 * are used by this file. This allows recovery to restart truncate 1032 * on orphans if we crash during truncate. We actually write i_disksize 1033 * into the on-disk inode when writing inodes out, instead of i_size. 1034 * 1035 * The only time when i_disksize and i_size may be different is when 1036 * a truncate is in progress. The only things which change i_disksize 1037 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1038 */ 1039 loff_t i_disksize; 1040 1041 /* 1042 * i_data_sem is for serialising ext4_truncate() against 1043 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1044 * data tree are chopped off during truncate. We can't do that in 1045 * ext4 because whenever we perform intermediate commits during 1046 * truncate, the inode and all the metadata blocks *must* be in a 1047 * consistent state which allows truncation of the orphans to restart 1048 * during recovery. Hence we must fix the get_block-vs-truncate race 1049 * by other means, so we have i_data_sem. 1050 */ 1051 struct rw_semaphore i_data_sem; 1052 /* 1053 * i_mmap_sem is for serializing page faults with truncate / punch hole 1054 * operations. We have to make sure that new page cannot be faulted in 1055 * a section of the inode that is being punched. We cannot easily use 1056 * i_data_sem for this since we need protection for the whole punch 1057 * operation and i_data_sem ranks below transaction start so we have 1058 * to occasionally drop it. 1059 */ 1060 struct rw_semaphore i_mmap_sem; 1061 struct inode vfs_inode; 1062 struct jbd2_inode *jinode; 1063 1064 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1065 1066 /* 1067 * File creation time. Its function is same as that of 1068 * struct timespec64 i_{a,c,m}time in the generic inode. 1069 */ 1070 struct timespec64 i_crtime; 1071 1072 /* mballoc */ 1073 struct list_head i_prealloc_list; 1074 spinlock_t i_prealloc_lock; 1075 1076 /* extents status tree */ 1077 struct ext4_es_tree i_es_tree; 1078 rwlock_t i_es_lock; 1079 struct list_head i_es_list; 1080 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1081 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1082 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1083 extents to shrink. Protected by 1084 i_es_lock */ 1085 1086 /* ialloc */ 1087 ext4_group_t i_last_alloc_group; 1088 1089 /* allocation reservation info for delalloc */ 1090 /* In case of bigalloc, this refer to clusters rather than blocks */ 1091 unsigned int i_reserved_data_blocks; 1092 1093 /* pending cluster reservations for bigalloc file systems */ 1094 struct ext4_pending_tree i_pending_tree; 1095 1096 /* on-disk additional length */ 1097 __u16 i_extra_isize; 1098 1099 /* Indicate the inline data space. */ 1100 u16 i_inline_off; 1101 u16 i_inline_size; 1102 1103 #ifdef CONFIG_QUOTA 1104 /* quota space reservation, managed internally by quota code */ 1105 qsize_t i_reserved_quota; 1106 #endif 1107 1108 /* Lock protecting lists below */ 1109 spinlock_t i_completed_io_lock; 1110 /* 1111 * Completed IOs that need unwritten extents handling and have 1112 * transaction reserved 1113 */ 1114 struct list_head i_rsv_conversion_list; 1115 struct work_struct i_rsv_conversion_work; 1116 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1117 1118 spinlock_t i_block_reservation_lock; 1119 1120 /* 1121 * Transactions that contain inode's metadata needed to complete 1122 * fsync and fdatasync, respectively. 1123 */ 1124 tid_t i_sync_tid; 1125 tid_t i_datasync_tid; 1126 1127 #ifdef CONFIG_QUOTA 1128 struct dquot *i_dquot[MAXQUOTAS]; 1129 #endif 1130 1131 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1132 __u32 i_csum_seed; 1133 1134 kprojid_t i_projid; 1135 }; 1136 1137 /* 1138 * File system states 1139 */ 1140 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1141 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1142 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1143 1144 /* 1145 * Misc. filesystem flags 1146 */ 1147 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1148 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1149 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1150 1151 /* 1152 * Mount flags set via mount options or defaults 1153 */ 1154 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1155 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1156 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1157 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1158 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1159 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1160 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1161 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1162 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1163 #ifdef CONFIG_FS_DAX 1164 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1165 #else 1166 #define EXT4_MOUNT_DAX_ALWAYS 0 1167 #endif 1168 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1169 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1170 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1171 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1172 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1173 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1174 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1175 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1176 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1177 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1178 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1179 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1180 * enable enforcement for hidden 1181 * quota files */ 1182 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1183 * enforcement for hidden quota 1184 * files */ 1185 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1186 * enforcement */ 1187 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1188 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1189 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1190 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1191 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1192 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1193 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1194 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1195 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1196 1197 /* 1198 * Mount flags set either automatically (could not be set by mount option) 1199 * based on per file system feature or property or in special cases such as 1200 * distinguishing between explicit mount option definition and default. 1201 */ 1202 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1203 specified delalloc */ 1204 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1205 size of blocksize * 8 1206 blocks */ 1207 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1208 file systems */ 1209 #define EXT4_MOUNT2_DAX_NEVER 0x00000008 /* Do not allow Direct Access */ 1210 #define EXT4_MOUNT2_DAX_INODE 0x00000010 /* For printing options only */ 1211 1212 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1213 specified journal checksum */ 1214 1215 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1216 ~EXT4_MOUNT_##opt 1217 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1218 EXT4_MOUNT_##opt 1219 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1220 EXT4_MOUNT_##opt) 1221 1222 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1223 ~EXT4_MOUNT2_##opt 1224 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1225 EXT4_MOUNT2_##opt 1226 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1227 EXT4_MOUNT2_##opt) 1228 1229 #define ext4_test_and_set_bit __test_and_set_bit_le 1230 #define ext4_set_bit __set_bit_le 1231 #define ext4_set_bit_atomic ext2_set_bit_atomic 1232 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1233 #define ext4_clear_bit __clear_bit_le 1234 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1235 #define ext4_test_bit test_bit_le 1236 #define ext4_find_next_zero_bit find_next_zero_bit_le 1237 #define ext4_find_next_bit find_next_bit_le 1238 1239 extern void ext4_set_bits(void *bm, int cur, int len); 1240 1241 /* 1242 * Maximal mount counts between two filesystem checks 1243 */ 1244 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1245 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1246 1247 /* 1248 * Behaviour when detecting errors 1249 */ 1250 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1251 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1252 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1253 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1254 1255 /* Metadata checksum algorithm codes */ 1256 #define EXT4_CRC32C_CHKSUM 1 1257 1258 /* 1259 * Structure of the super block 1260 */ 1261 struct ext4_super_block { 1262 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1263 __le32 s_blocks_count_lo; /* Blocks count */ 1264 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1265 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1266 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1267 __le32 s_first_data_block; /* First Data Block */ 1268 __le32 s_log_block_size; /* Block size */ 1269 __le32 s_log_cluster_size; /* Allocation cluster size */ 1270 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1271 __le32 s_clusters_per_group; /* # Clusters per group */ 1272 __le32 s_inodes_per_group; /* # Inodes per group */ 1273 __le32 s_mtime; /* Mount time */ 1274 /*30*/ __le32 s_wtime; /* Write time */ 1275 __le16 s_mnt_count; /* Mount count */ 1276 __le16 s_max_mnt_count; /* Maximal mount count */ 1277 __le16 s_magic; /* Magic signature */ 1278 __le16 s_state; /* File system state */ 1279 __le16 s_errors; /* Behaviour when detecting errors */ 1280 __le16 s_minor_rev_level; /* minor revision level */ 1281 /*40*/ __le32 s_lastcheck; /* time of last check */ 1282 __le32 s_checkinterval; /* max. time between checks */ 1283 __le32 s_creator_os; /* OS */ 1284 __le32 s_rev_level; /* Revision level */ 1285 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1286 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1287 /* 1288 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1289 * 1290 * Note: the difference between the compatible feature set and 1291 * the incompatible feature set is that if there is a bit set 1292 * in the incompatible feature set that the kernel doesn't 1293 * know about, it should refuse to mount the filesystem. 1294 * 1295 * e2fsck's requirements are more strict; if it doesn't know 1296 * about a feature in either the compatible or incompatible 1297 * feature set, it must abort and not try to meddle with 1298 * things it doesn't understand... 1299 */ 1300 __le32 s_first_ino; /* First non-reserved inode */ 1301 __le16 s_inode_size; /* size of inode structure */ 1302 __le16 s_block_group_nr; /* block group # of this superblock */ 1303 __le32 s_feature_compat; /* compatible feature set */ 1304 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1305 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1306 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1307 /*78*/ char s_volume_name[16]; /* volume name */ 1308 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1309 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1310 /* 1311 * Performance hints. Directory preallocation should only 1312 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1313 */ 1314 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1315 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1316 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1317 /* 1318 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1319 */ 1320 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1321 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1322 __le32 s_journal_dev; /* device number of journal file */ 1323 __le32 s_last_orphan; /* start of list of inodes to delete */ 1324 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1325 __u8 s_def_hash_version; /* Default hash version to use */ 1326 __u8 s_jnl_backup_type; 1327 __le16 s_desc_size; /* size of group descriptor */ 1328 /*100*/ __le32 s_default_mount_opts; 1329 __le32 s_first_meta_bg; /* First metablock block group */ 1330 __le32 s_mkfs_time; /* When the filesystem was created */ 1331 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1332 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1333 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1334 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1335 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1336 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1337 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1338 __le32 s_flags; /* Miscellaneous flags */ 1339 __le16 s_raid_stride; /* RAID stride */ 1340 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1341 __le64 s_mmp_block; /* Block for multi-mount protection */ 1342 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1343 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1344 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1345 __u8 s_encryption_level; /* versioning level for encryption */ 1346 __u8 s_reserved_pad; /* Padding to next 32bits */ 1347 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1348 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1349 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1350 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1351 snapshot's future use */ 1352 __le32 s_snapshot_list; /* inode number of the head of the 1353 on-disk snapshot list */ 1354 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1355 __le32 s_error_count; /* number of fs errors */ 1356 __le32 s_first_error_time; /* first time an error happened */ 1357 __le32 s_first_error_ino; /* inode involved in first error */ 1358 __le64 s_first_error_block; /* block involved of first error */ 1359 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1360 __le32 s_first_error_line; /* line number where error happened */ 1361 __le32 s_last_error_time; /* most recent time of an error */ 1362 __le32 s_last_error_ino; /* inode involved in last error */ 1363 __le32 s_last_error_line; /* line number where error happened */ 1364 __le64 s_last_error_block; /* block involved of last error */ 1365 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1366 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1367 __u8 s_mount_opts[64]; 1368 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1369 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1370 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1371 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1372 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1373 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1374 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1375 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1376 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1377 __u8 s_wtime_hi; 1378 __u8 s_mtime_hi; 1379 __u8 s_mkfs_time_hi; 1380 __u8 s_lastcheck_hi; 1381 __u8 s_first_error_time_hi; 1382 __u8 s_last_error_time_hi; 1383 __u8 s_first_error_errcode; 1384 __u8 s_last_error_errcode; 1385 __le16 s_encoding; /* Filename charset encoding */ 1386 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1387 __le32 s_reserved[95]; /* Padding to the end of the block */ 1388 __le32 s_checksum; /* crc32c(superblock) */ 1389 }; 1390 1391 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1392 1393 #ifdef __KERNEL__ 1394 1395 /* 1396 * run-time mount flags 1397 */ 1398 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1399 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1400 1401 #ifdef CONFIG_FS_ENCRYPTION 1402 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL) 1403 #else 1404 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1405 #endif 1406 1407 /* Number of quota types we support */ 1408 #define EXT4_MAXQUOTAS 3 1409 1410 #define EXT4_ENC_UTF8_12_1 1 1411 1412 /* 1413 * Flags for ext4_sb_info.s_encoding_flags. 1414 */ 1415 #define EXT4_ENC_STRICT_MODE_FL (1 << 0) 1416 1417 #define ext4_has_strict_mode(sbi) \ 1418 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL) 1419 1420 /* 1421 * fourth extended-fs super-block data in memory 1422 */ 1423 struct ext4_sb_info { 1424 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1425 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1426 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1427 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1428 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1429 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1430 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1431 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1432 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1433 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1434 unsigned long s_overhead; /* # of fs overhead clusters */ 1435 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1436 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1437 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1438 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1439 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1440 struct buffer_head * __rcu *s_group_desc; 1441 unsigned int s_mount_opt; 1442 unsigned int s_mount_opt2; 1443 unsigned int s_mount_flags; 1444 unsigned int s_def_mount_opt; 1445 ext4_fsblk_t s_sb_block; 1446 atomic64_t s_resv_clusters; 1447 kuid_t s_resuid; 1448 kgid_t s_resgid; 1449 unsigned short s_mount_state; 1450 unsigned short s_pad; 1451 int s_addr_per_block_bits; 1452 int s_desc_per_block_bits; 1453 int s_inode_size; 1454 int s_first_ino; 1455 unsigned int s_inode_readahead_blks; 1456 unsigned int s_inode_goal; 1457 u32 s_hash_seed[4]; 1458 int s_def_hash_version; 1459 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1460 struct percpu_counter s_freeclusters_counter; 1461 struct percpu_counter s_freeinodes_counter; 1462 struct percpu_counter s_dirs_counter; 1463 struct percpu_counter s_dirtyclusters_counter; 1464 struct blockgroup_lock *s_blockgroup_lock; 1465 struct proc_dir_entry *s_proc; 1466 struct kobject s_kobj; 1467 struct completion s_kobj_unregister; 1468 struct super_block *s_sb; 1469 #ifdef CONFIG_UNICODE 1470 struct unicode_map *s_encoding; 1471 __u16 s_encoding_flags; 1472 #endif 1473 1474 /* Journaling */ 1475 struct journal_s *s_journal; 1476 struct list_head s_orphan; 1477 struct mutex s_orphan_lock; 1478 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1479 unsigned long s_commit_interval; 1480 u32 s_max_batch_time; 1481 u32 s_min_batch_time; 1482 struct block_device *journal_bdev; 1483 #ifdef CONFIG_QUOTA 1484 /* Names of quota files with journalled quota */ 1485 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1486 int s_jquota_fmt; /* Format of quota to use */ 1487 #endif 1488 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1489 struct ext4_system_blocks __rcu *system_blks; 1490 1491 #ifdef EXTENTS_STATS 1492 /* ext4 extents stats */ 1493 unsigned long s_ext_min; 1494 unsigned long s_ext_max; 1495 unsigned long s_depth_max; 1496 spinlock_t s_ext_stats_lock; 1497 unsigned long s_ext_blocks; 1498 unsigned long s_ext_extents; 1499 #endif 1500 1501 /* for buddy allocator */ 1502 struct ext4_group_info ** __rcu *s_group_info; 1503 struct inode *s_buddy_cache; 1504 spinlock_t s_md_lock; 1505 unsigned short *s_mb_offsets; 1506 unsigned int *s_mb_maxs; 1507 unsigned int s_group_info_size; 1508 unsigned int s_mb_free_pending; 1509 struct list_head s_freed_data_list; /* List of blocks to be freed 1510 after commit completed */ 1511 1512 /* tunables */ 1513 unsigned long s_stripe; 1514 unsigned int s_mb_stream_request; 1515 unsigned int s_mb_max_to_scan; 1516 unsigned int s_mb_min_to_scan; 1517 unsigned int s_mb_stats; 1518 unsigned int s_mb_order2_reqs; 1519 unsigned int s_mb_group_prealloc; 1520 unsigned int s_max_dir_size_kb; 1521 /* where last allocation was done - for stream allocation */ 1522 unsigned long s_mb_last_group; 1523 unsigned long s_mb_last_start; 1524 unsigned int s_mb_prefetch; 1525 unsigned int s_mb_prefetch_limit; 1526 1527 /* stats for buddy allocator */ 1528 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1529 atomic_t s_bal_success; /* we found long enough chunks */ 1530 atomic_t s_bal_allocated; /* in blocks */ 1531 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1532 atomic_t s_bal_goals; /* goal hits */ 1533 atomic_t s_bal_breaks; /* too long searches */ 1534 atomic_t s_bal_2orders; /* 2^order hits */ 1535 spinlock_t s_bal_lock; 1536 unsigned long s_mb_buddies_generated; 1537 unsigned long long s_mb_generation_time; 1538 atomic_t s_mb_lost_chunks; 1539 atomic_t s_mb_preallocated; 1540 atomic_t s_mb_discarded; 1541 atomic_t s_lock_busy; 1542 1543 /* locality groups */ 1544 struct ext4_locality_group __percpu *s_locality_groups; 1545 1546 /* for write statistics */ 1547 unsigned long s_sectors_written_start; 1548 u64 s_kbytes_written; 1549 1550 /* the size of zero-out chunk */ 1551 unsigned int s_extent_max_zeroout_kb; 1552 1553 unsigned int s_log_groups_per_flex; 1554 struct flex_groups * __rcu *s_flex_groups; 1555 ext4_group_t s_flex_groups_allocated; 1556 1557 /* workqueue for reserved extent conversions (buffered io) */ 1558 struct workqueue_struct *rsv_conversion_wq; 1559 1560 /* timer for periodic error stats printing */ 1561 struct timer_list s_err_report; 1562 1563 /* Lazy inode table initialization info */ 1564 struct ext4_li_request *s_li_request; 1565 /* Wait multiplier for lazy initialization thread */ 1566 unsigned int s_li_wait_mult; 1567 1568 /* Kernel thread for multiple mount protection */ 1569 struct task_struct *s_mmp_tsk; 1570 1571 /* record the last minlen when FITRIM is called. */ 1572 atomic_t s_last_trim_minblks; 1573 1574 /* Reference to checksum algorithm driver via cryptoapi */ 1575 struct crypto_shash *s_chksum_driver; 1576 1577 /* Precomputed FS UUID checksum for seeding other checksums */ 1578 __u32 s_csum_seed; 1579 1580 /* Reclaim extents from extent status tree */ 1581 struct shrinker s_es_shrinker; 1582 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1583 long s_es_nr_inode; 1584 struct ext4_es_stats s_es_stats; 1585 struct mb_cache *s_ea_block_cache; 1586 struct mb_cache *s_ea_inode_cache; 1587 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1588 1589 /* Ratelimit ext4 messages. */ 1590 struct ratelimit_state s_err_ratelimit_state; 1591 struct ratelimit_state s_warning_ratelimit_state; 1592 struct ratelimit_state s_msg_ratelimit_state; 1593 1594 /* Encryption context for '-o test_dummy_encryption' */ 1595 struct fscrypt_dummy_context s_dummy_enc_ctx; 1596 1597 /* 1598 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1599 * or EXTENTS flag. 1600 */ 1601 struct percpu_rw_semaphore s_writepages_rwsem; 1602 struct dax_device *s_daxdev; 1603 #ifdef CONFIG_EXT4_DEBUG 1604 unsigned long s_simulate_fail; 1605 #endif 1606 /* Record the errseq of the backing block device */ 1607 errseq_t s_bdev_wb_err; 1608 spinlock_t s_bdev_wb_lock; 1609 }; 1610 1611 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1612 { 1613 return sb->s_fs_info; 1614 } 1615 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1616 { 1617 return container_of(inode, struct ext4_inode_info, vfs_inode); 1618 } 1619 1620 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1621 { 1622 return ino == EXT4_ROOT_INO || 1623 (ino >= EXT4_FIRST_INO(sb) && 1624 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1625 } 1626 1627 /* 1628 * Returns: sbi->field[index] 1629 * Used to access an array element from the following sbi fields which require 1630 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1631 * - s_group_desc 1632 * - s_group_info 1633 * - s_flex_group 1634 */ 1635 #define sbi_array_rcu_deref(sbi, field, index) \ 1636 ({ \ 1637 typeof(*((sbi)->field)) _v; \ 1638 rcu_read_lock(); \ 1639 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1640 rcu_read_unlock(); \ 1641 _v; \ 1642 }) 1643 1644 /* 1645 * Simulate_fail codes 1646 */ 1647 #define EXT4_SIM_BBITMAP_EIO 1 1648 #define EXT4_SIM_BBITMAP_CRC 2 1649 #define EXT4_SIM_IBITMAP_EIO 3 1650 #define EXT4_SIM_IBITMAP_CRC 4 1651 #define EXT4_SIM_INODE_EIO 5 1652 #define EXT4_SIM_INODE_CRC 6 1653 #define EXT4_SIM_DIRBLOCK_EIO 7 1654 #define EXT4_SIM_DIRBLOCK_CRC 8 1655 1656 static inline bool ext4_simulate_fail(struct super_block *sb, 1657 unsigned long code) 1658 { 1659 #ifdef CONFIG_EXT4_DEBUG 1660 struct ext4_sb_info *sbi = EXT4_SB(sb); 1661 1662 if (unlikely(sbi->s_simulate_fail == code)) { 1663 sbi->s_simulate_fail = 0; 1664 return true; 1665 } 1666 #endif 1667 return false; 1668 } 1669 1670 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1671 struct buffer_head *bh, 1672 unsigned long code) 1673 { 1674 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1675 clear_buffer_uptodate(bh); 1676 } 1677 1678 /* 1679 * Error number codes for s_{first,last}_error_errno 1680 * 1681 * Linux errno numbers are architecture specific, so we need to translate 1682 * them into something which is architecture independent. We don't define 1683 * codes for all errno's; just the ones which are most likely to be the cause 1684 * of an ext4_error() call. 1685 */ 1686 #define EXT4_ERR_UNKNOWN 1 1687 #define EXT4_ERR_EIO 2 1688 #define EXT4_ERR_ENOMEM 3 1689 #define EXT4_ERR_EFSBADCRC 4 1690 #define EXT4_ERR_EFSCORRUPTED 5 1691 #define EXT4_ERR_ENOSPC 6 1692 #define EXT4_ERR_ENOKEY 7 1693 #define EXT4_ERR_EROFS 8 1694 #define EXT4_ERR_EFBIG 9 1695 #define EXT4_ERR_EEXIST 10 1696 #define EXT4_ERR_ERANGE 11 1697 #define EXT4_ERR_EOVERFLOW 12 1698 #define EXT4_ERR_EBUSY 13 1699 #define EXT4_ERR_ENOTDIR 14 1700 #define EXT4_ERR_ENOTEMPTY 15 1701 #define EXT4_ERR_ESHUTDOWN 16 1702 #define EXT4_ERR_EFAULT 17 1703 1704 /* 1705 * Inode dynamic state flags 1706 */ 1707 enum { 1708 EXT4_STATE_JDATA, /* journaled data exists */ 1709 EXT4_STATE_NEW, /* inode is newly created */ 1710 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1711 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1712 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1713 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1714 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1715 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1716 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1717 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1718 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1719 }; 1720 1721 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1722 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1723 { \ 1724 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1725 } \ 1726 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1727 { \ 1728 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1729 } \ 1730 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1731 { \ 1732 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1733 } 1734 1735 /* Add these declarations here only so that these functions can be 1736 * found by name. Otherwise, they are very hard to locate. */ 1737 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1738 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1739 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1740 EXT4_INODE_BIT_FNS(flag, flags, 0) 1741 1742 /* Add these declarations here only so that these functions can be 1743 * found by name. Otherwise, they are very hard to locate. */ 1744 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1745 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1746 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1747 #if (BITS_PER_LONG < 64) 1748 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1749 1750 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1751 { 1752 (ei)->i_state_flags = 0; 1753 } 1754 #else 1755 EXT4_INODE_BIT_FNS(state, flags, 32) 1756 1757 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1758 { 1759 /* We depend on the fact that callers will set i_flags */ 1760 } 1761 #endif 1762 #else 1763 /* Assume that user mode programs are passing in an ext4fs superblock, not 1764 * a kernel struct super_block. This will allow us to call the feature-test 1765 * macros from user land. */ 1766 #define EXT4_SB(sb) (sb) 1767 #endif 1768 1769 static inline bool ext4_verity_in_progress(struct inode *inode) 1770 { 1771 return IS_ENABLED(CONFIG_FS_VERITY) && 1772 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1773 } 1774 1775 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1776 1777 /* 1778 * Codes for operating systems 1779 */ 1780 #define EXT4_OS_LINUX 0 1781 #define EXT4_OS_HURD 1 1782 #define EXT4_OS_MASIX 2 1783 #define EXT4_OS_FREEBSD 3 1784 #define EXT4_OS_LITES 4 1785 1786 /* 1787 * Revision levels 1788 */ 1789 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1790 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1791 1792 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1793 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1794 1795 #define EXT4_GOOD_OLD_INODE_SIZE 128 1796 1797 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1798 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1799 #define EXT4_TIMESTAMP_MIN S32_MIN 1800 1801 /* 1802 * Feature set definitions 1803 */ 1804 1805 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1806 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1807 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1808 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1809 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1810 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1811 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1812 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1813 1814 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1815 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1816 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1817 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1818 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1819 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1820 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1821 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1822 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1823 /* 1824 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1825 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1826 * all other data structures' checksums. However, the METADATA_CSUM and 1827 * GDT_CSUM bits are mutually exclusive. 1828 */ 1829 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1830 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1831 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1832 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1833 1834 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1835 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1836 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1837 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1838 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1839 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1840 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1841 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1842 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1843 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1844 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1845 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1846 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1847 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1848 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1849 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1850 1851 extern void ext4_update_dynamic_rev(struct super_block *sb); 1852 1853 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1854 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1855 { \ 1856 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1857 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1858 } \ 1859 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1860 { \ 1861 ext4_update_dynamic_rev(sb); \ 1862 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1863 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1864 } \ 1865 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1866 { \ 1867 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1868 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1869 } 1870 1871 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1872 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1873 { \ 1874 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1875 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1876 } \ 1877 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1878 { \ 1879 ext4_update_dynamic_rev(sb); \ 1880 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1881 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1882 } \ 1883 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1884 { \ 1885 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1886 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1887 } 1888 1889 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1890 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1891 { \ 1892 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1893 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1894 } \ 1895 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1896 { \ 1897 ext4_update_dynamic_rev(sb); \ 1898 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1899 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1900 } \ 1901 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1902 { \ 1903 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1904 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1905 } 1906 1907 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1908 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1909 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1910 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1911 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1912 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1913 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1914 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 1915 1916 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1917 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1918 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1919 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1920 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1921 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1922 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1923 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1924 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1925 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1926 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1927 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1928 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 1929 1930 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1931 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1932 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1933 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1934 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1935 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1936 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1937 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1938 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1939 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1940 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1941 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1942 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1943 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1944 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1945 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 1946 1947 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1948 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1949 EXT4_FEATURE_INCOMPAT_META_BG) 1950 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1951 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1952 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1953 1954 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1955 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1956 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1957 EXT4_FEATURE_INCOMPAT_META_BG) 1958 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1959 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1960 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1961 1962 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1963 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1964 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1965 EXT4_FEATURE_INCOMPAT_META_BG| \ 1966 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1967 EXT4_FEATURE_INCOMPAT_64BIT| \ 1968 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1969 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1970 EXT4_FEATURE_INCOMPAT_MMP | \ 1971 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1972 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1973 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 1974 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1975 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1976 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1977 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1978 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1979 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1980 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1981 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1982 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1983 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1984 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1985 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1986 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 1987 EXT4_FEATURE_RO_COMPAT_VERITY) 1988 1989 #define EXTN_FEATURE_FUNCS(ver) \ 1990 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1991 { \ 1992 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1993 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1994 } \ 1995 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1996 { \ 1997 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1998 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1999 } \ 2000 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2001 { \ 2002 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2003 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2004 } 2005 2006 EXTN_FEATURE_FUNCS(2) 2007 EXTN_FEATURE_FUNCS(3) 2008 EXTN_FEATURE_FUNCS(4) 2009 2010 static inline bool ext4_has_compat_features(struct super_block *sb) 2011 { 2012 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2013 } 2014 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2015 { 2016 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2017 } 2018 static inline bool ext4_has_incompat_features(struct super_block *sb) 2019 { 2020 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2021 } 2022 2023 /* 2024 * Superblock flags 2025 */ 2026 #define EXT4_FLAGS_RESIZING 0 2027 #define EXT4_FLAGS_SHUTDOWN 1 2028 #define EXT4_FLAGS_BDEV_IS_DAX 2 2029 2030 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2031 { 2032 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2033 } 2034 2035 2036 /* 2037 * Default values for user and/or group using reserved blocks 2038 */ 2039 #define EXT4_DEF_RESUID 0 2040 #define EXT4_DEF_RESGID 0 2041 2042 /* 2043 * Default project ID 2044 */ 2045 #define EXT4_DEF_PROJID 0 2046 2047 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2048 2049 /* 2050 * Default mount options 2051 */ 2052 #define EXT4_DEFM_DEBUG 0x0001 2053 #define EXT4_DEFM_BSDGROUPS 0x0002 2054 #define EXT4_DEFM_XATTR_USER 0x0004 2055 #define EXT4_DEFM_ACL 0x0008 2056 #define EXT4_DEFM_UID16 0x0010 2057 #define EXT4_DEFM_JMODE 0x0060 2058 #define EXT4_DEFM_JMODE_DATA 0x0020 2059 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2060 #define EXT4_DEFM_JMODE_WBACK 0x0060 2061 #define EXT4_DEFM_NOBARRIER 0x0100 2062 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2063 #define EXT4_DEFM_DISCARD 0x0400 2064 #define EXT4_DEFM_NODELALLOC 0x0800 2065 2066 /* 2067 * Default journal batch times 2068 */ 2069 #define EXT4_DEF_MIN_BATCH_TIME 0 2070 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2071 2072 /* 2073 * Minimum number of groups in a flexgroup before we separate out 2074 * directories into the first block group of a flexgroup 2075 */ 2076 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2077 2078 /* 2079 * Structure of a directory entry 2080 */ 2081 #define EXT4_NAME_LEN 255 2082 2083 struct ext4_dir_entry { 2084 __le32 inode; /* Inode number */ 2085 __le16 rec_len; /* Directory entry length */ 2086 __le16 name_len; /* Name length */ 2087 char name[EXT4_NAME_LEN]; /* File name */ 2088 }; 2089 2090 /* 2091 * The new version of the directory entry. Since EXT4 structures are 2092 * stored in intel byte order, and the name_len field could never be 2093 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2094 * file_type field. 2095 */ 2096 struct ext4_dir_entry_2 { 2097 __le32 inode; /* Inode number */ 2098 __le16 rec_len; /* Directory entry length */ 2099 __u8 name_len; /* Name length */ 2100 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2101 char name[EXT4_NAME_LEN]; /* File name */ 2102 }; 2103 2104 /* 2105 * This is a bogus directory entry at the end of each leaf block that 2106 * records checksums. 2107 */ 2108 struct ext4_dir_entry_tail { 2109 __le32 det_reserved_zero1; /* Pretend to be unused */ 2110 __le16 det_rec_len; /* 12 */ 2111 __u8 det_reserved_zero2; /* Zero name length */ 2112 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2113 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2114 }; 2115 2116 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2117 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2118 ((blocksize) - \ 2119 sizeof(struct ext4_dir_entry_tail)))) 2120 2121 /* 2122 * Ext4 directory file types. Only the low 3 bits are used. The 2123 * other bits are reserved for now. 2124 */ 2125 #define EXT4_FT_UNKNOWN 0 2126 #define EXT4_FT_REG_FILE 1 2127 #define EXT4_FT_DIR 2 2128 #define EXT4_FT_CHRDEV 3 2129 #define EXT4_FT_BLKDEV 4 2130 #define EXT4_FT_FIFO 5 2131 #define EXT4_FT_SOCK 6 2132 #define EXT4_FT_SYMLINK 7 2133 2134 #define EXT4_FT_MAX 8 2135 2136 #define EXT4_FT_DIR_CSUM 0xDE 2137 2138 /* 2139 * EXT4_DIR_PAD defines the directory entries boundaries 2140 * 2141 * NOTE: It must be a multiple of 4 2142 */ 2143 #define EXT4_DIR_PAD 4 2144 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2145 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 2146 ~EXT4_DIR_ROUND) 2147 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2148 2149 /* 2150 * If we ever get support for fs block sizes > page_size, we'll need 2151 * to remove the #if statements in the next two functions... 2152 */ 2153 static inline unsigned int 2154 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2155 { 2156 unsigned len = le16_to_cpu(dlen); 2157 2158 #if (PAGE_SIZE >= 65536) 2159 if (len == EXT4_MAX_REC_LEN || len == 0) 2160 return blocksize; 2161 return (len & 65532) | ((len & 3) << 16); 2162 #else 2163 return len; 2164 #endif 2165 } 2166 2167 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2168 { 2169 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2170 BUG(); 2171 #if (PAGE_SIZE >= 65536) 2172 if (len < 65536) 2173 return cpu_to_le16(len); 2174 if (len == blocksize) { 2175 if (blocksize == 65536) 2176 return cpu_to_le16(EXT4_MAX_REC_LEN); 2177 else 2178 return cpu_to_le16(0); 2179 } 2180 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2181 #else 2182 return cpu_to_le16(len); 2183 #endif 2184 } 2185 2186 /* 2187 * Hash Tree Directory indexing 2188 * (c) Daniel Phillips, 2001 2189 */ 2190 2191 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2192 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2193 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2194 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2195 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2196 2197 /* Legal values for the dx_root hash_version field: */ 2198 2199 #define DX_HASH_LEGACY 0 2200 #define DX_HASH_HALF_MD4 1 2201 #define DX_HASH_TEA 2 2202 #define DX_HASH_LEGACY_UNSIGNED 3 2203 #define DX_HASH_HALF_MD4_UNSIGNED 4 2204 #define DX_HASH_TEA_UNSIGNED 5 2205 2206 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2207 const void *address, unsigned int length) 2208 { 2209 struct { 2210 struct shash_desc shash; 2211 char ctx[4]; 2212 } desc; 2213 2214 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2215 2216 desc.shash.tfm = sbi->s_chksum_driver; 2217 *(u32 *)desc.ctx = crc; 2218 2219 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2220 2221 return *(u32 *)desc.ctx; 2222 } 2223 2224 #ifdef __KERNEL__ 2225 2226 /* hash info structure used by the directory hash */ 2227 struct dx_hash_info 2228 { 2229 u32 hash; 2230 u32 minor_hash; 2231 int hash_version; 2232 u32 *seed; 2233 }; 2234 2235 2236 /* 32 and 64 bit signed EOF for dx directories */ 2237 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2238 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2239 2240 2241 /* 2242 * Control parameters used by ext4_htree_next_block 2243 */ 2244 #define HASH_NB_ALWAYS 1 2245 2246 struct ext4_filename { 2247 const struct qstr *usr_fname; 2248 struct fscrypt_str disk_name; 2249 struct dx_hash_info hinfo; 2250 #ifdef CONFIG_FS_ENCRYPTION 2251 struct fscrypt_str crypto_buf; 2252 #endif 2253 #ifdef CONFIG_UNICODE 2254 struct fscrypt_str cf_name; 2255 #endif 2256 }; 2257 2258 #define fname_name(p) ((p)->disk_name.name) 2259 #define fname_len(p) ((p)->disk_name.len) 2260 2261 /* 2262 * Describe an inode's exact location on disk and in memory 2263 */ 2264 struct ext4_iloc 2265 { 2266 struct buffer_head *bh; 2267 unsigned long offset; 2268 ext4_group_t block_group; 2269 }; 2270 2271 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2272 { 2273 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2274 } 2275 2276 static inline bool ext4_is_quota_file(struct inode *inode) 2277 { 2278 return IS_NOQUOTA(inode) && 2279 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2280 } 2281 2282 /* 2283 * This structure is stuffed into the struct file's private_data field 2284 * for directories. It is where we put information so that we can do 2285 * readdir operations in hash tree order. 2286 */ 2287 struct dir_private_info { 2288 struct rb_root root; 2289 struct rb_node *curr_node; 2290 struct fname *extra_fname; 2291 loff_t last_pos; 2292 __u32 curr_hash; 2293 __u32 curr_minor_hash; 2294 __u32 next_hash; 2295 }; 2296 2297 /* calculate the first block number of the group */ 2298 static inline ext4_fsblk_t 2299 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2300 { 2301 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2302 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2303 } 2304 2305 /* 2306 * Special error return code only used by dx_probe() and its callers. 2307 */ 2308 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2309 2310 /* htree levels for ext4 */ 2311 #define EXT4_HTREE_LEVEL_COMPAT 2 2312 #define EXT4_HTREE_LEVEL 3 2313 2314 static inline int ext4_dir_htree_level(struct super_block *sb) 2315 { 2316 return ext4_has_feature_largedir(sb) ? 2317 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2318 } 2319 2320 /* 2321 * Timeout and state flag for lazy initialization inode thread. 2322 */ 2323 #define EXT4_DEF_LI_WAIT_MULT 10 2324 #define EXT4_DEF_LI_MAX_START_DELAY 5 2325 #define EXT4_LAZYINIT_QUIT 0x0001 2326 #define EXT4_LAZYINIT_RUNNING 0x0002 2327 2328 /* 2329 * Lazy inode table initialization info 2330 */ 2331 struct ext4_lazy_init { 2332 unsigned long li_state; 2333 struct list_head li_request_list; 2334 struct mutex li_list_mtx; 2335 }; 2336 2337 struct ext4_li_request { 2338 struct super_block *lr_super; 2339 struct ext4_sb_info *lr_sbi; 2340 ext4_group_t lr_next_group; 2341 struct list_head lr_request; 2342 unsigned long lr_next_sched; 2343 unsigned long lr_timeout; 2344 }; 2345 2346 struct ext4_features { 2347 struct kobject f_kobj; 2348 struct completion f_kobj_unregister; 2349 }; 2350 2351 /* 2352 * This structure will be used for multiple mount protection. It will be 2353 * written into the block number saved in the s_mmp_block field in the 2354 * superblock. Programs that check MMP should assume that if 2355 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2356 * to use the filesystem, regardless of how old the timestamp is. 2357 */ 2358 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2359 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2360 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2361 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2362 2363 struct mmp_struct { 2364 __le32 mmp_magic; /* Magic number for MMP */ 2365 __le32 mmp_seq; /* Sequence no. updated periodically */ 2366 2367 /* 2368 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2369 * purposes and do not affect the correctness of the algorithm 2370 */ 2371 __le64 mmp_time; /* Time last updated */ 2372 char mmp_nodename[64]; /* Node which last updated MMP block */ 2373 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2374 2375 /* 2376 * mmp_check_interval is used to verify if the MMP block has been 2377 * updated on the block device. The value is updated based on the 2378 * maximum time to write the MMP block during an update cycle. 2379 */ 2380 __le16 mmp_check_interval; 2381 2382 __le16 mmp_pad1; 2383 __le32 mmp_pad2[226]; 2384 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2385 }; 2386 2387 /* arguments passed to the mmp thread */ 2388 struct mmpd_data { 2389 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2390 struct super_block *sb; /* super block of the fs */ 2391 }; 2392 2393 /* 2394 * Check interval multiplier 2395 * The MMP block is written every update interval and initially checked every 2396 * update interval x the multiplier (the value is then adapted based on the 2397 * write latency). The reason is that writes can be delayed under load and we 2398 * don't want readers to incorrectly assume that the filesystem is no longer 2399 * in use. 2400 */ 2401 #define EXT4_MMP_CHECK_MULT 2UL 2402 2403 /* 2404 * Minimum interval for MMP checking in seconds. 2405 */ 2406 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2407 2408 /* 2409 * Maximum interval for MMP checking in seconds. 2410 */ 2411 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2412 2413 /* 2414 * Function prototypes 2415 */ 2416 2417 /* 2418 * Ok, these declarations are also in <linux/kernel.h> but none of the 2419 * ext4 source programs needs to include it so they are duplicated here. 2420 */ 2421 # define NORET_TYPE /**/ 2422 # define ATTRIB_NORET __attribute__((noreturn)) 2423 # define NORET_AND noreturn, 2424 2425 /* bitmap.c */ 2426 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2427 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2428 struct ext4_group_desc *gdp, 2429 struct buffer_head *bh, int sz); 2430 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2431 struct ext4_group_desc *gdp, 2432 struct buffer_head *bh, int sz); 2433 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2434 struct ext4_group_desc *gdp, 2435 struct buffer_head *bh); 2436 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2437 struct ext4_group_desc *gdp, 2438 struct buffer_head *bh); 2439 2440 /* balloc.c */ 2441 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2442 ext4_fsblk_t blocknr, 2443 ext4_group_t *blockgrpp, 2444 ext4_grpblk_t *offsetp); 2445 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2446 ext4_fsblk_t block); 2447 2448 extern unsigned int ext4_block_group(struct super_block *sb, 2449 ext4_fsblk_t blocknr); 2450 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2451 ext4_fsblk_t blocknr); 2452 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2453 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2454 ext4_group_t group); 2455 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2456 ext4_fsblk_t goal, 2457 unsigned int flags, 2458 unsigned long *count, 2459 int *errp); 2460 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2461 s64 nclusters, unsigned int flags); 2462 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2463 extern void ext4_check_blocks_bitmap(struct super_block *); 2464 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2465 ext4_group_t block_group, 2466 struct buffer_head ** bh); 2467 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2468 2469 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2470 ext4_group_t block_group, 2471 bool ignore_locked); 2472 extern int ext4_wait_block_bitmap(struct super_block *sb, 2473 ext4_group_t block_group, 2474 struct buffer_head *bh); 2475 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2476 ext4_group_t block_group); 2477 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2478 ext4_group_t block_group, 2479 struct ext4_group_desc *gdp); 2480 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2481 2482 #ifdef CONFIG_UNICODE 2483 extern void ext4_fname_setup_ci_filename(struct inode *dir, 2484 const struct qstr *iname, 2485 struct fscrypt_str *fname); 2486 #endif 2487 2488 #ifdef CONFIG_FS_ENCRYPTION 2489 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2490 const struct fscrypt_name *src) 2491 { 2492 memset(dst, 0, sizeof(*dst)); 2493 2494 dst->usr_fname = src->usr_fname; 2495 dst->disk_name = src->disk_name; 2496 dst->hinfo.hash = src->hash; 2497 dst->hinfo.minor_hash = src->minor_hash; 2498 dst->crypto_buf = src->crypto_buf; 2499 } 2500 2501 static inline int ext4_fname_setup_filename(struct inode *dir, 2502 const struct qstr *iname, 2503 int lookup, 2504 struct ext4_filename *fname) 2505 { 2506 struct fscrypt_name name; 2507 int err; 2508 2509 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2510 if (err) 2511 return err; 2512 2513 ext4_fname_from_fscrypt_name(fname, &name); 2514 2515 #ifdef CONFIG_UNICODE 2516 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2517 #endif 2518 return 0; 2519 } 2520 2521 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2522 struct dentry *dentry, 2523 struct ext4_filename *fname) 2524 { 2525 struct fscrypt_name name; 2526 int err; 2527 2528 err = fscrypt_prepare_lookup(dir, dentry, &name); 2529 if (err) 2530 return err; 2531 2532 ext4_fname_from_fscrypt_name(fname, &name); 2533 2534 #ifdef CONFIG_UNICODE 2535 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name); 2536 #endif 2537 return 0; 2538 } 2539 2540 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2541 { 2542 struct fscrypt_name name; 2543 2544 name.crypto_buf = fname->crypto_buf; 2545 fscrypt_free_filename(&name); 2546 2547 fname->crypto_buf.name = NULL; 2548 fname->usr_fname = NULL; 2549 fname->disk_name.name = NULL; 2550 2551 #ifdef CONFIG_UNICODE 2552 kfree(fname->cf_name.name); 2553 fname->cf_name.name = NULL; 2554 #endif 2555 } 2556 #else /* !CONFIG_FS_ENCRYPTION */ 2557 static inline int ext4_fname_setup_filename(struct inode *dir, 2558 const struct qstr *iname, 2559 int lookup, 2560 struct ext4_filename *fname) 2561 { 2562 fname->usr_fname = iname; 2563 fname->disk_name.name = (unsigned char *) iname->name; 2564 fname->disk_name.len = iname->len; 2565 2566 #ifdef CONFIG_UNICODE 2567 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2568 #endif 2569 2570 return 0; 2571 } 2572 2573 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2574 struct dentry *dentry, 2575 struct ext4_filename *fname) 2576 { 2577 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2578 } 2579 2580 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2581 { 2582 #ifdef CONFIG_UNICODE 2583 kfree(fname->cf_name.name); 2584 fname->cf_name.name = NULL; 2585 #endif 2586 } 2587 #endif /* !CONFIG_FS_ENCRYPTION */ 2588 2589 /* dir.c */ 2590 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2591 struct file *, 2592 struct ext4_dir_entry_2 *, 2593 struct buffer_head *, char *, int, 2594 unsigned int); 2595 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2596 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2597 (de), (bh), (buf), (size), (offset))) 2598 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2599 __u32 minor_hash, 2600 struct ext4_dir_entry_2 *dirent, 2601 struct fscrypt_str *ent_name); 2602 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2603 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2604 struct buffer_head *bh, 2605 void *buf, int buf_size, 2606 struct ext4_filename *fname, 2607 struct ext4_dir_entry_2 **dest_de); 2608 void ext4_insert_dentry(struct inode *inode, 2609 struct ext4_dir_entry_2 *de, 2610 int buf_size, 2611 struct ext4_filename *fname); 2612 static inline void ext4_update_dx_flag(struct inode *inode) 2613 { 2614 if (!ext4_has_feature_dir_index(inode->i_sb)) { 2615 /* ext4_iget() should have caught this... */ 2616 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2617 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2618 } 2619 } 2620 static const unsigned char ext4_filetype_table[] = { 2621 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2622 }; 2623 2624 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2625 { 2626 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2627 return DT_UNKNOWN; 2628 2629 return ext4_filetype_table[filetype]; 2630 } 2631 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2632 void *buf, int buf_size); 2633 2634 /* fsync.c */ 2635 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2636 2637 /* hash.c */ 2638 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2639 struct dx_hash_info *hinfo); 2640 2641 /* ialloc.c */ 2642 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2643 const struct qstr *qstr, __u32 goal, 2644 uid_t *owner, __u32 i_flags, 2645 int handle_type, unsigned int line_no, 2646 int nblocks); 2647 2648 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2649 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2650 i_flags, 0, 0, 0) 2651 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2652 type, nblocks) \ 2653 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2654 0, (type), __LINE__, (nblocks)) 2655 2656 2657 extern void ext4_free_inode(handle_t *, struct inode *); 2658 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2659 extern unsigned long ext4_count_free_inodes(struct super_block *); 2660 extern unsigned long ext4_count_dirs(struct super_block *); 2661 extern void ext4_check_inodes_bitmap(struct super_block *); 2662 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2663 extern int ext4_init_inode_table(struct super_block *sb, 2664 ext4_group_t group, int barrier); 2665 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2666 2667 /* mballoc.c */ 2668 extern const struct seq_operations ext4_mb_seq_groups_ops; 2669 extern long ext4_mb_stats; 2670 extern long ext4_mb_max_to_scan; 2671 extern int ext4_mb_init(struct super_block *); 2672 extern int ext4_mb_release(struct super_block *); 2673 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2674 struct ext4_allocation_request *, int *); 2675 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2676 extern void ext4_discard_preallocations(struct inode *); 2677 extern int __init ext4_init_mballoc(void); 2678 extern void ext4_exit_mballoc(void); 2679 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2680 struct buffer_head *bh, ext4_fsblk_t block, 2681 unsigned long count, int flags); 2682 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2683 ext4_group_t ngroups); 2684 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2685 ext4_group_t i, struct ext4_group_desc *desc); 2686 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2687 ext4_fsblk_t block, unsigned long count); 2688 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2689 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2690 2691 /* inode.c */ 2692 int ext4_inode_is_fast_symlink(struct inode *inode); 2693 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2694 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2695 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2696 bool wait, struct buffer_head **bhs); 2697 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2698 struct buffer_head *bh_result, int create); 2699 int ext4_get_block(struct inode *inode, sector_t iblock, 2700 struct buffer_head *bh_result, int create); 2701 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2702 struct buffer_head *bh, int create); 2703 int ext4_walk_page_buffers(handle_t *handle, 2704 struct buffer_head *head, 2705 unsigned from, 2706 unsigned to, 2707 int *partial, 2708 int (*fn)(handle_t *handle, 2709 struct buffer_head *bh)); 2710 int do_journal_get_write_access(handle_t *handle, 2711 struct buffer_head *bh); 2712 #define FALL_BACK_TO_NONDELALLOC 1 2713 #define CONVERT_INLINE_DATA 2 2714 2715 typedef enum { 2716 EXT4_IGET_NORMAL = 0, 2717 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2718 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2719 } ext4_iget_flags; 2720 2721 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2722 ext4_iget_flags flags, const char *function, 2723 unsigned int line); 2724 2725 #define ext4_iget(sb, ino, flags) \ 2726 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2727 2728 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2729 extern int ext4_setattr(struct dentry *, struct iattr *); 2730 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2731 extern void ext4_evict_inode(struct inode *); 2732 extern void ext4_clear_inode(struct inode *); 2733 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2734 extern int ext4_sync_inode(handle_t *, struct inode *); 2735 extern void ext4_dirty_inode(struct inode *, int); 2736 extern int ext4_change_inode_journal_flag(struct inode *, int); 2737 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2738 extern int ext4_inode_attach_jinode(struct inode *inode); 2739 extern int ext4_can_truncate(struct inode *inode); 2740 extern int ext4_truncate(struct inode *); 2741 extern int ext4_break_layouts(struct inode *); 2742 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2743 extern void ext4_set_inode_flags(struct inode *, bool init); 2744 extern int ext4_alloc_da_blocks(struct inode *inode); 2745 extern void ext4_set_aops(struct inode *inode); 2746 extern int ext4_writepage_trans_blocks(struct inode *); 2747 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2748 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2749 loff_t lstart, loff_t lend); 2750 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2751 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2752 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2753 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2754 extern void ext4_da_release_space(struct inode *inode, int to_free); 2755 extern void ext4_da_update_reserve_space(struct inode *inode, 2756 int used, int quota_claim); 2757 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2758 ext4_fsblk_t pblk, ext4_lblk_t len); 2759 2760 /* indirect.c */ 2761 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2762 struct ext4_map_blocks *map, int flags); 2763 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2764 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2765 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2766 ext4_lblk_t start, ext4_lblk_t end); 2767 2768 /* ioctl.c */ 2769 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2770 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2771 2772 /* migrate.c */ 2773 extern int ext4_ext_migrate(struct inode *); 2774 extern int ext4_ind_migrate(struct inode *inode); 2775 2776 /* namei.c */ 2777 extern int ext4_dirblock_csum_verify(struct inode *inode, 2778 struct buffer_head *bh); 2779 extern int ext4_orphan_add(handle_t *, struct inode *); 2780 extern int ext4_orphan_del(handle_t *, struct inode *); 2781 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2782 __u32 start_minor_hash, __u32 *next_hash); 2783 extern int ext4_search_dir(struct buffer_head *bh, 2784 char *search_buf, 2785 int buf_size, 2786 struct inode *dir, 2787 struct ext4_filename *fname, 2788 unsigned int offset, 2789 struct ext4_dir_entry_2 **res_dir); 2790 extern int ext4_generic_delete_entry(handle_t *handle, 2791 struct inode *dir, 2792 struct ext4_dir_entry_2 *de_del, 2793 struct buffer_head *bh, 2794 void *entry_buf, 2795 int buf_size, 2796 int csum_size); 2797 extern bool ext4_empty_dir(struct inode *inode); 2798 2799 /* resize.c */ 2800 extern void ext4_kvfree_array_rcu(void *to_free); 2801 extern int ext4_group_add(struct super_block *sb, 2802 struct ext4_new_group_data *input); 2803 extern int ext4_group_extend(struct super_block *sb, 2804 struct ext4_super_block *es, 2805 ext4_fsblk_t n_blocks_count); 2806 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2807 2808 /* super.c */ 2809 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2810 sector_t block, int op_flags); 2811 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2812 extern int ext4_calculate_overhead(struct super_block *sb); 2813 extern void ext4_superblock_csum_set(struct super_block *sb); 2814 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2815 ext4_group_t ngroup); 2816 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2817 char nbuf[16]); 2818 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2819 ext4_group_t block_group, 2820 unsigned int flags); 2821 2822 extern __printf(6, 7) 2823 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64, 2824 const char *, ...); 2825 extern __printf(6, 7) 2826 void __ext4_error_inode(struct inode *, const char *, unsigned int, 2827 ext4_fsblk_t, int, const char *, ...); 2828 extern __printf(5, 6) 2829 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2830 const char *, ...); 2831 extern void __ext4_std_error(struct super_block *, const char *, 2832 unsigned int, int); 2833 extern __printf(5, 6) 2834 void __ext4_abort(struct super_block *, const char *, unsigned int, int, 2835 const char *, ...); 2836 extern __printf(4, 5) 2837 void __ext4_warning(struct super_block *, const char *, unsigned int, 2838 const char *, ...); 2839 extern __printf(4, 5) 2840 void __ext4_warning_inode(const struct inode *inode, const char *function, 2841 unsigned int line, const char *fmt, ...); 2842 extern __printf(3, 4) 2843 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2844 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2845 const char *, unsigned int, const char *); 2846 extern __printf(7, 8) 2847 void __ext4_grp_locked_error(const char *, unsigned int, 2848 struct super_block *, ext4_group_t, 2849 unsigned long, ext4_fsblk_t, 2850 const char *, ...); 2851 2852 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2853 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2854 2855 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 2856 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 2857 2858 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 2859 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 2860 (fmt), ## a) 2861 2862 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2863 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2864 2865 #ifdef CONFIG_PRINTK 2866 2867 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2868 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 2869 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 2870 __ext4_error_inode((inode), (func), (line), (block), \ 2871 (err), (fmt), ##__VA_ARGS__) 2872 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2873 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2874 #define ext4_error(sb, fmt, ...) \ 2875 __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__) 2876 #define ext4_error_err(sb, err, fmt, ...) \ 2877 __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__) 2878 #define ext4_abort(sb, err, fmt, ...) \ 2879 __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__) 2880 #define ext4_warning(sb, fmt, ...) \ 2881 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2882 #define ext4_warning_inode(inode, fmt, ...) \ 2883 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2884 #define ext4_msg(sb, level, fmt, ...) \ 2885 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2886 #define dump_mmp_msg(sb, mmp, msg) \ 2887 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2888 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2889 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2890 fmt, ##__VA_ARGS__) 2891 2892 #else 2893 2894 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2895 do { \ 2896 no_printk(fmt, ##__VA_ARGS__); \ 2897 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 2898 } while (0) 2899 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 2900 do { \ 2901 no_printk(fmt, ##__VA_ARGS__); \ 2902 __ext4_error_inode(inode, "", 0, block, err, " "); \ 2903 } while (0) 2904 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2905 do { \ 2906 no_printk(fmt, ##__VA_ARGS__); \ 2907 __ext4_error_file(file, "", 0, block, " "); \ 2908 } while (0) 2909 #define ext4_error(sb, fmt, ...) \ 2910 do { \ 2911 no_printk(fmt, ##__VA_ARGS__); \ 2912 __ext4_error(sb, "", 0, 0, 0, " "); \ 2913 } while (0) 2914 #define ext4_error_err(sb, err, fmt, ...) \ 2915 do { \ 2916 no_printk(fmt, ##__VA_ARGS__); \ 2917 __ext4_error(sb, "", 0, err, 0, " "); \ 2918 } while (0) 2919 #define ext4_abort(sb, err, fmt, ...) \ 2920 do { \ 2921 no_printk(fmt, ##__VA_ARGS__); \ 2922 __ext4_abort(sb, "", 0, err, " "); \ 2923 } while (0) 2924 #define ext4_warning(sb, fmt, ...) \ 2925 do { \ 2926 no_printk(fmt, ##__VA_ARGS__); \ 2927 __ext4_warning(sb, "", 0, " "); \ 2928 } while (0) 2929 #define ext4_warning_inode(inode, fmt, ...) \ 2930 do { \ 2931 no_printk(fmt, ##__VA_ARGS__); \ 2932 __ext4_warning_inode(inode, "", 0, " "); \ 2933 } while (0) 2934 #define ext4_msg(sb, level, fmt, ...) \ 2935 do { \ 2936 no_printk(fmt, ##__VA_ARGS__); \ 2937 __ext4_msg(sb, "", " "); \ 2938 } while (0) 2939 #define dump_mmp_msg(sb, mmp, msg) \ 2940 __dump_mmp_msg(sb, mmp, "", 0, "") 2941 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2942 do { \ 2943 no_printk(fmt, ##__VA_ARGS__); \ 2944 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2945 } while (0) 2946 2947 #endif 2948 2949 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2950 __u32 compat); 2951 extern int ext4_update_rocompat_feature(handle_t *handle, 2952 struct super_block *sb, __u32 rocompat); 2953 extern int ext4_update_incompat_feature(handle_t *handle, 2954 struct super_block *sb, __u32 incompat); 2955 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2956 struct ext4_group_desc *bg); 2957 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2958 struct ext4_group_desc *bg); 2959 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2960 struct ext4_group_desc *bg); 2961 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2962 struct ext4_group_desc *bg); 2963 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2964 struct ext4_group_desc *bg); 2965 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2966 struct ext4_group_desc *bg); 2967 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2968 struct ext4_group_desc *bg); 2969 extern void ext4_block_bitmap_set(struct super_block *sb, 2970 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2971 extern void ext4_inode_bitmap_set(struct super_block *sb, 2972 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2973 extern void ext4_inode_table_set(struct super_block *sb, 2974 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2975 extern void ext4_free_group_clusters_set(struct super_block *sb, 2976 struct ext4_group_desc *bg, 2977 __u32 count); 2978 extern void ext4_free_inodes_set(struct super_block *sb, 2979 struct ext4_group_desc *bg, __u32 count); 2980 extern void ext4_used_dirs_set(struct super_block *sb, 2981 struct ext4_group_desc *bg, __u32 count); 2982 extern void ext4_itable_unused_set(struct super_block *sb, 2983 struct ext4_group_desc *bg, __u32 count); 2984 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2985 struct ext4_group_desc *gdp); 2986 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2987 struct ext4_group_desc *gdp); 2988 extern int ext4_register_li_request(struct super_block *sb, 2989 ext4_group_t first_not_zeroed); 2990 2991 static inline int ext4_has_metadata_csum(struct super_block *sb) 2992 { 2993 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2994 !EXT4_SB(sb)->s_chksum_driver); 2995 2996 return ext4_has_feature_metadata_csum(sb) && 2997 (EXT4_SB(sb)->s_chksum_driver != NULL); 2998 } 2999 3000 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3001 { 3002 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3003 } 3004 3005 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3006 { 3007 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 3008 le32_to_cpu(es->s_blocks_count_lo); 3009 } 3010 3011 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3012 { 3013 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 3014 le32_to_cpu(es->s_r_blocks_count_lo); 3015 } 3016 3017 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3018 { 3019 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 3020 le32_to_cpu(es->s_free_blocks_count_lo); 3021 } 3022 3023 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3024 ext4_fsblk_t blk) 3025 { 3026 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3027 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3028 } 3029 3030 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3031 ext4_fsblk_t blk) 3032 { 3033 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3034 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3035 } 3036 3037 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3038 ext4_fsblk_t blk) 3039 { 3040 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3041 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3042 } 3043 3044 static inline loff_t ext4_isize(struct super_block *sb, 3045 struct ext4_inode *raw_inode) 3046 { 3047 if (ext4_has_feature_largedir(sb) || 3048 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3049 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3050 le32_to_cpu(raw_inode->i_size_lo); 3051 3052 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3053 } 3054 3055 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3056 { 3057 raw_inode->i_size_lo = cpu_to_le32(i_size); 3058 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3059 } 3060 3061 static inline 3062 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3063 ext4_group_t group) 3064 { 3065 struct ext4_group_info **grp_info; 3066 long indexv, indexh; 3067 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3068 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3069 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3070 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3071 return grp_info[indexh]; 3072 } 3073 3074 /* 3075 * Reading s_groups_count requires using smp_rmb() afterwards. See 3076 * the locking protocol documented in the comments of ext4_group_add() 3077 * in resize.c 3078 */ 3079 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3080 { 3081 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3082 3083 smp_rmb(); 3084 return ngroups; 3085 } 3086 3087 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3088 ext4_group_t block_group) 3089 { 3090 return block_group >> sbi->s_log_groups_per_flex; 3091 } 3092 3093 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3094 { 3095 return 1 << sbi->s_log_groups_per_flex; 3096 } 3097 3098 #define ext4_std_error(sb, errno) \ 3099 do { \ 3100 if ((errno)) \ 3101 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3102 } while (0) 3103 3104 #ifdef CONFIG_SMP 3105 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3106 * counters. So we need to make sure we have free clusters more 3107 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3108 */ 3109 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3110 #else 3111 #define EXT4_FREECLUSTERS_WATERMARK 0 3112 #endif 3113 3114 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 3115 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3116 { 3117 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3118 !inode_is_locked(inode)); 3119 down_write(&EXT4_I(inode)->i_data_sem); 3120 if (newsize > EXT4_I(inode)->i_disksize) 3121 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3122 up_write(&EXT4_I(inode)->i_data_sem); 3123 } 3124 3125 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 3126 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3127 { 3128 int changed = 0; 3129 3130 if (newsize > inode->i_size) { 3131 i_size_write(inode, newsize); 3132 changed = 1; 3133 } 3134 if (newsize > EXT4_I(inode)->i_disksize) { 3135 ext4_update_i_disksize(inode, newsize); 3136 changed |= 2; 3137 } 3138 return changed; 3139 } 3140 3141 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3142 loff_t len); 3143 3144 struct ext4_group_info { 3145 unsigned long bb_state; 3146 struct rb_root bb_free_root; 3147 ext4_grpblk_t bb_first_free; /* first free block */ 3148 ext4_grpblk_t bb_free; /* total free blocks */ 3149 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3150 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3151 struct list_head bb_prealloc_list; 3152 #ifdef DOUBLE_CHECK 3153 void *bb_bitmap; 3154 #endif 3155 struct rw_semaphore alloc_sem; 3156 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3157 * regions, index is order. 3158 * bb_counters[3] = 5 means 3159 * 5 free 8-block regions. */ 3160 }; 3161 3162 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3163 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3164 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3165 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3166 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3167 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3168 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3169 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3170 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3171 3172 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3173 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3174 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3175 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3176 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3177 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3178 3179 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3180 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3181 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3182 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3183 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3184 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3185 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3186 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3187 3188 #define EXT4_MAX_CONTENTION 8 3189 #define EXT4_CONTENTION_THRESHOLD 2 3190 3191 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3192 ext4_group_t group) 3193 { 3194 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3195 } 3196 3197 /* 3198 * Returns true if the filesystem is busy enough that attempts to 3199 * access the block group locks has run into contention. 3200 */ 3201 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3202 { 3203 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3204 } 3205 3206 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3207 { 3208 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3209 if (spin_trylock(lock)) 3210 /* 3211 * We're able to grab the lock right away, so drop the 3212 * lock contention counter. 3213 */ 3214 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3215 else { 3216 /* 3217 * The lock is busy, so bump the contention counter, 3218 * and then wait on the spin lock. 3219 */ 3220 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3221 EXT4_MAX_CONTENTION); 3222 spin_lock(lock); 3223 } 3224 } 3225 3226 static inline void ext4_unlock_group(struct super_block *sb, 3227 ext4_group_t group) 3228 { 3229 spin_unlock(ext4_group_lock_ptr(sb, group)); 3230 } 3231 3232 /* 3233 * Block validity checking 3234 */ 3235 #define ext4_check_indirect_blockref(inode, bh) \ 3236 ext4_check_blockref(__func__, __LINE__, inode, \ 3237 (__le32 *)(bh)->b_data, \ 3238 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3239 3240 #define ext4_ind_check_inode(inode) \ 3241 ext4_check_blockref(__func__, __LINE__, inode, \ 3242 EXT4_I(inode)->i_data, \ 3243 EXT4_NDIR_BLOCKS) 3244 3245 /* 3246 * Inodes and files operations 3247 */ 3248 3249 /* dir.c */ 3250 extern const struct file_operations ext4_dir_operations; 3251 3252 #ifdef CONFIG_UNICODE 3253 extern const struct dentry_operations ext4_dentry_ops; 3254 #endif 3255 3256 /* file.c */ 3257 extern const struct inode_operations ext4_file_inode_operations; 3258 extern const struct file_operations ext4_file_operations; 3259 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3260 3261 /* inline.c */ 3262 extern int ext4_get_max_inline_size(struct inode *inode); 3263 extern int ext4_find_inline_data_nolock(struct inode *inode); 3264 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3265 unsigned int len); 3266 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3267 3268 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3269 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3270 struct inode *inode, 3271 loff_t pos, unsigned len, 3272 unsigned flags, 3273 struct page **pagep); 3274 extern int ext4_write_inline_data_end(struct inode *inode, 3275 loff_t pos, unsigned len, 3276 unsigned copied, 3277 struct page *page); 3278 extern struct buffer_head * 3279 ext4_journalled_write_inline_data(struct inode *inode, 3280 unsigned len, 3281 struct page *page); 3282 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3283 struct inode *inode, 3284 loff_t pos, unsigned len, 3285 unsigned flags, 3286 struct page **pagep, 3287 void **fsdata); 3288 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3289 unsigned len, unsigned copied, 3290 struct page *page); 3291 extern int ext4_try_add_inline_entry(handle_t *handle, 3292 struct ext4_filename *fname, 3293 struct inode *dir, struct inode *inode); 3294 extern int ext4_try_create_inline_dir(handle_t *handle, 3295 struct inode *parent, 3296 struct inode *inode); 3297 extern int ext4_read_inline_dir(struct file *filp, 3298 struct dir_context *ctx, 3299 int *has_inline_data); 3300 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3301 struct inode *dir, ext4_lblk_t block, 3302 struct dx_hash_info *hinfo, 3303 __u32 start_hash, __u32 start_minor_hash, 3304 int *has_inline_data); 3305 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3306 struct ext4_filename *fname, 3307 struct ext4_dir_entry_2 **res_dir, 3308 int *has_inline_data); 3309 extern int ext4_delete_inline_entry(handle_t *handle, 3310 struct inode *dir, 3311 struct ext4_dir_entry_2 *de_del, 3312 struct buffer_head *bh, 3313 int *has_inline_data); 3314 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3315 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3316 struct ext4_dir_entry_2 **parent_de, 3317 int *retval); 3318 extern int ext4_inline_data_fiemap(struct inode *inode, 3319 struct fiemap_extent_info *fieinfo, 3320 int *has_inline, __u64 start, __u64 len); 3321 3322 struct iomap; 3323 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3324 3325 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3326 3327 extern int ext4_convert_inline_data(struct inode *inode); 3328 3329 static inline int ext4_has_inline_data(struct inode *inode) 3330 { 3331 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3332 EXT4_I(inode)->i_inline_off; 3333 } 3334 3335 /* namei.c */ 3336 extern const struct inode_operations ext4_dir_inode_operations; 3337 extern const struct inode_operations ext4_special_inode_operations; 3338 extern struct dentry *ext4_get_parent(struct dentry *child); 3339 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3340 struct ext4_dir_entry_2 *de, 3341 int blocksize, int csum_size, 3342 unsigned int parent_ino, int dotdot_real_len); 3343 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3344 unsigned int blocksize); 3345 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3346 struct buffer_head *bh); 3347 extern int ext4_ci_compare(const struct inode *parent, 3348 const struct qstr *fname, 3349 const struct qstr *entry, bool quick); 3350 3351 #define S_SHIFT 12 3352 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3353 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3354 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3355 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3356 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3357 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3358 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3359 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3360 }; 3361 3362 static inline void ext4_set_de_type(struct super_block *sb, 3363 struct ext4_dir_entry_2 *de, 3364 umode_t mode) { 3365 if (ext4_has_feature_filetype(sb)) 3366 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3367 } 3368 3369 /* readpages.c */ 3370 extern int ext4_mpage_readpages(struct inode *inode, 3371 struct readahead_control *rac, struct page *page); 3372 extern int __init ext4_init_post_read_processing(void); 3373 extern void ext4_exit_post_read_processing(void); 3374 3375 /* symlink.c */ 3376 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3377 extern const struct inode_operations ext4_symlink_inode_operations; 3378 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3379 3380 /* sysfs.c */ 3381 extern int ext4_register_sysfs(struct super_block *sb); 3382 extern void ext4_unregister_sysfs(struct super_block *sb); 3383 extern int __init ext4_init_sysfs(void); 3384 extern void ext4_exit_sysfs(void); 3385 3386 /* block_validity */ 3387 extern void ext4_release_system_zone(struct super_block *sb); 3388 extern int ext4_setup_system_zone(struct super_block *sb); 3389 extern int __init ext4_init_system_zone(void); 3390 extern void ext4_exit_system_zone(void); 3391 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3392 ext4_fsblk_t start_blk, 3393 unsigned int count); 3394 extern int ext4_check_blockref(const char *, unsigned int, 3395 struct inode *, __le32 *, unsigned int); 3396 3397 /* extents.c */ 3398 struct ext4_ext_path; 3399 struct ext4_extent; 3400 3401 /* 3402 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3403 * __le32. 3404 */ 3405 #define EXT_MAX_BLOCKS 0xffffffff 3406 3407 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3408 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3409 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3410 struct ext4_map_blocks *map, int flags); 3411 extern int ext4_ext_truncate(handle_t *, struct inode *); 3412 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3413 ext4_lblk_t end); 3414 extern void ext4_ext_init(struct super_block *); 3415 extern void ext4_ext_release(struct super_block *); 3416 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3417 loff_t len); 3418 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3419 loff_t offset, ssize_t len); 3420 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3421 ext4_io_end_t *io_end); 3422 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3423 struct ext4_map_blocks *map, int flags); 3424 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3425 int num, 3426 struct ext4_ext_path *path); 3427 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3428 struct ext4_ext_path **, 3429 struct ext4_extent *, int); 3430 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3431 struct ext4_ext_path **, 3432 int flags); 3433 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3434 extern int ext4_ext_check_inode(struct inode *inode); 3435 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3436 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3437 __u64 start, __u64 len); 3438 extern int ext4_get_es_cache(struct inode *inode, 3439 struct fiemap_extent_info *fieinfo, 3440 __u64 start, __u64 len); 3441 extern int ext4_ext_precache(struct inode *inode); 3442 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3443 struct inode *inode2, ext4_lblk_t lblk1, 3444 ext4_lblk_t lblk2, ext4_lblk_t count, 3445 int mark_unwritten,int *err); 3446 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3447 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3448 int check_cred, int restart_cred, 3449 int revoke_cred); 3450 3451 3452 /* move_extent.c */ 3453 extern void ext4_double_down_write_data_sem(struct inode *first, 3454 struct inode *second); 3455 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3456 struct inode *donor_inode); 3457 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3458 __u64 start_orig, __u64 start_donor, 3459 __u64 len, __u64 *moved_len); 3460 3461 /* page-io.c */ 3462 extern int __init ext4_init_pageio(void); 3463 extern void ext4_exit_pageio(void); 3464 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3465 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3466 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3467 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3468 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3469 struct writeback_control *wbc); 3470 extern void ext4_end_io_rsv_work(struct work_struct *work); 3471 extern void ext4_io_submit(struct ext4_io_submit *io); 3472 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3473 struct page *page, 3474 int len, 3475 struct writeback_control *wbc, 3476 bool keep_towrite); 3477 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3478 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3479 3480 /* mmp.c */ 3481 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3482 3483 /* verity.c */ 3484 extern const struct fsverity_operations ext4_verityops; 3485 3486 /* 3487 * Add new method to test whether block and inode bitmaps are properly 3488 * initialized. With uninit_bg reading the block from disk is not enough 3489 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3490 */ 3491 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3492 3493 static inline int bitmap_uptodate(struct buffer_head *bh) 3494 { 3495 return (buffer_uptodate(bh) && 3496 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3497 } 3498 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3499 { 3500 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3501 } 3502 3503 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3504 3505 /* For ioend & aio unwritten conversion wait queues */ 3506 #define EXT4_WQ_HASH_SZ 37 3507 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3508 EXT4_WQ_HASH_SZ]) 3509 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3510 3511 extern int ext4_resize_begin(struct super_block *sb); 3512 extern void ext4_resize_end(struct super_block *sb); 3513 3514 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3515 struct ext4_io_end *io_end) 3516 { 3517 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3518 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3519 atomic_inc(&EXT4_I(inode)->i_unwritten); 3520 } 3521 } 3522 3523 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3524 { 3525 struct inode *inode = io_end->inode; 3526 3527 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3528 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3529 /* Wake up anyone waiting on unwritten extent conversion */ 3530 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3531 wake_up_all(ext4_ioend_wq(inode)); 3532 } 3533 } 3534 3535 extern const struct iomap_ops ext4_iomap_ops; 3536 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3537 extern const struct iomap_ops ext4_iomap_report_ops; 3538 3539 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3540 { 3541 /* 3542 * If the buffer has the write error flag, we have failed 3543 * to write out data in the block. In this case, we don't 3544 * have to read the block because we may read the old data 3545 * successfully. 3546 */ 3547 if (!buffer_uptodate(bh) && buffer_write_io_error(bh)) 3548 set_buffer_uptodate(bh); 3549 return buffer_uptodate(bh); 3550 } 3551 3552 #endif /* __KERNEL__ */ 3553 3554 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3555 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3556 3557 #endif /* _EXT4_H */ 3558