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/wait.h> 31 #include <linux/sched/signal.h> 32 #include <linux/blockgroup_lock.h> 33 #include <linux/percpu_counter.h> 34 #include <linux/ratelimit.h> 35 #include <crypto/hash.h> 36 #include <linux/falloc.h> 37 #include <linux/percpu-rwsem.h> 38 #include <linux/fiemap.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #include <linux/fscrypt.h> 44 #include <linux/fsverity.h> 45 46 #include <linux/compiler.h> 47 48 /* 49 * The fourth extended filesystem constants/structures 50 */ 51 52 /* 53 * with AGGRESSIVE_CHECK allocator runs consistency checks over 54 * structures. these checks slow things down a lot 55 */ 56 #define AGGRESSIVE_CHECK__ 57 58 /* 59 * with DOUBLE_CHECK defined mballoc creates persistent in-core 60 * bitmaps, maintains and uses them to check for double allocations 61 */ 62 #define DOUBLE_CHECK__ 63 64 /* 65 * Define EXT4FS_DEBUG to produce debug messages 66 */ 67 #undef EXT4FS_DEBUG 68 69 /* 70 * Debug code 71 */ 72 #ifdef EXT4FS_DEBUG 73 #define ext4_debug(f, a...) \ 74 do { \ 75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 76 __FILE__, __LINE__, __func__); \ 77 printk(KERN_DEBUG f, ## a); \ 78 } while (0) 79 #else 80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 81 #endif 82 83 /* 84 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 85 */ 86 #define EXT_DEBUG__ 87 88 /* 89 * Dynamic printk for controlled extents debugging. 90 */ 91 #ifdef CONFIG_EXT4_DEBUG 92 #define ext_debug(ino, fmt, ...) \ 93 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 94 current->comm, task_pid_nr(current), \ 95 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 96 __func__, ##__VA_ARGS__) 97 #else 98 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 99 #endif 100 101 #define ASSERT(assert) \ 102 do { \ 103 if (unlikely(!(assert))) { \ 104 printk(KERN_EMERG \ 105 "Assertion failure in %s() at %s:%d: '%s'\n", \ 106 __func__, __FILE__, __LINE__, #assert); \ 107 BUG(); \ 108 } \ 109 } while (0) 110 111 /* data type for block offset of block group */ 112 typedef int ext4_grpblk_t; 113 114 /* data type for filesystem-wide blocks number */ 115 typedef unsigned long long ext4_fsblk_t; 116 117 /* data type for file logical block number */ 118 typedef __u32 ext4_lblk_t; 119 120 /* data type for block group number */ 121 typedef unsigned int ext4_group_t; 122 123 enum SHIFT_DIRECTION { 124 SHIFT_LEFT = 0, 125 SHIFT_RIGHT, 126 }; 127 128 /* 129 * Flags used in mballoc's allocation_context flags field. 130 * 131 * Also used to show what's going on for debugging purposes when the 132 * flag field is exported via the traceport interface 133 */ 134 135 /* prefer goal again. length */ 136 #define EXT4_MB_HINT_MERGE 0x0001 137 /* blocks already reserved */ 138 #define EXT4_MB_HINT_RESERVED 0x0002 139 /* metadata is being allocated */ 140 #define EXT4_MB_HINT_METADATA 0x0004 141 /* first blocks in the file */ 142 #define EXT4_MB_HINT_FIRST 0x0008 143 /* search for the best chunk */ 144 #define EXT4_MB_HINT_BEST 0x0010 145 /* data is being allocated */ 146 #define EXT4_MB_HINT_DATA 0x0020 147 /* don't preallocate (for tails) */ 148 #define EXT4_MB_HINT_NOPREALLOC 0x0040 149 /* allocate for locality group */ 150 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 151 /* allocate goal blocks or none */ 152 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 153 /* goal is meaningful */ 154 #define EXT4_MB_HINT_TRY_GOAL 0x0200 155 /* blocks already pre-reserved by delayed allocation */ 156 #define EXT4_MB_DELALLOC_RESERVED 0x0400 157 /* We are doing stream allocation */ 158 #define EXT4_MB_STREAM_ALLOC 0x0800 159 /* Use reserved root blocks if needed */ 160 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 161 /* Use blocks from reserved pool */ 162 #define EXT4_MB_USE_RESERVED 0x2000 163 /* Do strict check for free blocks while retrying block allocation */ 164 #define EXT4_MB_STRICT_CHECK 0x4000 165 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 166 #define EXT4_MB_CR0_OPTIMIZED 0x8000 167 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 168 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 169 /* Perform linear traversal for one group */ 170 #define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000 171 struct ext4_allocation_request { 172 /* target inode for block we're allocating */ 173 struct inode *inode; 174 /* how many blocks we want to allocate */ 175 unsigned int len; 176 /* logical block in target inode */ 177 ext4_lblk_t logical; 178 /* the closest logical allocated block to the left */ 179 ext4_lblk_t lleft; 180 /* the closest logical allocated block to the right */ 181 ext4_lblk_t lright; 182 /* phys. target (a hint) */ 183 ext4_fsblk_t goal; 184 /* phys. block for the closest logical allocated block to the left */ 185 ext4_fsblk_t pleft; 186 /* phys. block for the closest logical allocated block to the right */ 187 ext4_fsblk_t pright; 188 /* flags. see above EXT4_MB_HINT_* */ 189 unsigned int flags; 190 }; 191 192 /* 193 * Logical to physical block mapping, used by ext4_map_blocks() 194 * 195 * This structure is used to pass requests into ext4_map_blocks() as 196 * well as to store the information returned by ext4_map_blocks(). It 197 * takes less room on the stack than a struct buffer_head. 198 */ 199 #define EXT4_MAP_NEW BIT(BH_New) 200 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 201 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 202 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 203 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 204 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 205 206 struct ext4_map_blocks { 207 ext4_fsblk_t m_pblk; 208 ext4_lblk_t m_lblk; 209 unsigned int m_len; 210 unsigned int m_flags; 211 }; 212 213 /* 214 * Block validity checking, system zone rbtree. 215 */ 216 struct ext4_system_blocks { 217 struct rb_root root; 218 struct rcu_head rcu; 219 }; 220 221 /* 222 * Flags for ext4_io_end->flags 223 */ 224 #define EXT4_IO_END_UNWRITTEN 0x0001 225 226 struct ext4_io_end_vec { 227 struct list_head list; /* list of io_end_vec */ 228 loff_t offset; /* offset in the file */ 229 ssize_t size; /* size of the extent */ 230 }; 231 232 /* 233 * For converting unwritten extents on a work queue. 'handle' is used for 234 * buffered writeback. 235 */ 236 typedef struct ext4_io_end { 237 struct list_head list; /* per-file finished IO list */ 238 handle_t *handle; /* handle reserved for extent 239 * conversion */ 240 struct inode *inode; /* file being written to */ 241 struct bio *bio; /* Linked list of completed 242 * bios covering the extent */ 243 unsigned int flag; /* unwritten or not */ 244 atomic_t count; /* reference counter */ 245 struct list_head list_vec; /* list of ext4_io_end_vec */ 246 } ext4_io_end_t; 247 248 struct ext4_io_submit { 249 struct writeback_control *io_wbc; 250 struct bio *io_bio; 251 ext4_io_end_t *io_end; 252 sector_t io_next_block; 253 }; 254 255 /* 256 * Special inodes numbers 257 */ 258 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 259 #define EXT4_ROOT_INO 2 /* Root inode */ 260 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 261 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 262 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 263 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 264 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 265 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 266 267 /* First non-reserved inode for old ext4 filesystems */ 268 #define EXT4_GOOD_OLD_FIRST_INO 11 269 270 /* 271 * Maximal count of links to a file 272 */ 273 #define EXT4_LINK_MAX 65000 274 275 /* 276 * Macro-instructions used to manage several block sizes 277 */ 278 #define EXT4_MIN_BLOCK_SIZE 1024 279 #define EXT4_MAX_BLOCK_SIZE 65536 280 #define EXT4_MIN_BLOCK_LOG_SIZE 10 281 #define EXT4_MAX_BLOCK_LOG_SIZE 16 282 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 283 #ifdef __KERNEL__ 284 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 285 #else 286 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 287 #endif 288 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 289 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 290 EXT4_SB(s)->s_cluster_bits) 291 #ifdef __KERNEL__ 292 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 293 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 294 #else 295 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 296 #endif 297 #ifdef __KERNEL__ 298 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 299 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 300 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 301 #else 302 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 303 EXT4_GOOD_OLD_INODE_SIZE : \ 304 (s)->s_inode_size) 305 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 306 EXT4_GOOD_OLD_FIRST_INO : \ 307 (s)->s_first_ino) 308 #endif 309 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 310 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 311 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 312 blkbits)) 313 314 /* Translate a block number to a cluster number */ 315 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 316 /* Translate a cluster number to a block number */ 317 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 318 /* Translate # of blks to # of clusters */ 319 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 320 (sbi)->s_cluster_bits) 321 /* Mask out the low bits to get the starting block of the cluster */ 322 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 323 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 324 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 325 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 326 /* Fill in the low bits to get the last block of the cluster */ 327 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 328 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 329 /* Get the cluster offset */ 330 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 331 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 332 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 333 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 334 335 /* 336 * Structure of a blocks group descriptor 337 */ 338 struct ext4_group_desc 339 { 340 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 341 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 342 __le32 bg_inode_table_lo; /* Inodes table block */ 343 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 344 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 345 __le16 bg_used_dirs_count_lo; /* Directories count */ 346 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 347 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 348 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 349 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 350 __le16 bg_itable_unused_lo; /* Unused inodes count */ 351 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 352 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 353 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 354 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 355 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 356 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 357 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 358 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 359 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 360 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 361 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 362 __u32 bg_reserved; 363 }; 364 365 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 366 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 367 sizeof(__le16)) 368 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 369 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 370 sizeof(__le16)) 371 372 /* 373 * Structure of a flex block group info 374 */ 375 376 struct flex_groups { 377 atomic64_t free_clusters; 378 atomic_t free_inodes; 379 atomic_t used_dirs; 380 }; 381 382 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 383 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 384 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 385 386 /* 387 * Macro-instructions used to manage group descriptors 388 */ 389 #define EXT4_MIN_DESC_SIZE 32 390 #define EXT4_MIN_DESC_SIZE_64BIT 64 391 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 392 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 393 #ifdef __KERNEL__ 394 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 395 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 396 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 397 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 398 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 399 #else 400 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 401 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 402 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 403 #endif 404 405 /* 406 * Constants relative to the data blocks 407 */ 408 #define EXT4_NDIR_BLOCKS 12 409 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 410 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 411 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 412 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 413 414 /* 415 * Inode flags 416 */ 417 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 418 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 419 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 420 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 421 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 422 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 423 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 424 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 425 /* Reserved for compression usage... */ 426 #define EXT4_DIRTY_FL 0x00000100 427 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 428 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 429 /* nb: was previously EXT2_ECOMPR_FL */ 430 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 431 /* End compression flags --- maybe not all used */ 432 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 433 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 434 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 435 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 436 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 437 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 438 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 439 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 440 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 441 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 442 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 443 444 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 445 446 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 447 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 448 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 449 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 450 451 /* User modifiable flags */ 452 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 453 EXT4_UNRM_FL | \ 454 EXT4_COMPR_FL | \ 455 EXT4_SYNC_FL | \ 456 EXT4_IMMUTABLE_FL | \ 457 EXT4_APPEND_FL | \ 458 EXT4_NODUMP_FL | \ 459 EXT4_NOATIME_FL | \ 460 EXT4_JOURNAL_DATA_FL | \ 461 EXT4_NOTAIL_FL | \ 462 EXT4_DIRSYNC_FL | \ 463 EXT4_TOPDIR_FL | \ 464 EXT4_EXTENTS_FL | \ 465 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 466 EXT4_DAX_FL | \ 467 EXT4_PROJINHERIT_FL | \ 468 EXT4_CASEFOLD_FL) 469 470 /* User visible flags */ 471 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 472 EXT4_DIRTY_FL | \ 473 EXT4_COMPRBLK_FL | \ 474 EXT4_NOCOMPR_FL | \ 475 EXT4_ENCRYPT_FL | \ 476 EXT4_INDEX_FL | \ 477 EXT4_VERITY_FL | \ 478 EXT4_INLINE_DATA_FL) 479 480 /* Flags that should be inherited by new inodes from their parent. */ 481 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 482 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 483 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 484 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 485 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 486 EXT4_DAX_FL) 487 488 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 489 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 490 EXT4_PROJINHERIT_FL)) 491 492 /* Flags that are appropriate for non-directories/regular files. */ 493 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 494 495 /* The only flags that should be swapped */ 496 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 497 498 /* Flags which are mutually exclusive to DAX */ 499 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 500 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 501 502 /* Mask out flags that are inappropriate for the given type of inode. */ 503 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 504 { 505 if (S_ISDIR(mode)) 506 return flags; 507 else if (S_ISREG(mode)) 508 return flags & EXT4_REG_FLMASK; 509 else 510 return flags & EXT4_OTHER_FLMASK; 511 } 512 513 /* 514 * Inode flags used for atomic set/get 515 */ 516 enum { 517 EXT4_INODE_SECRM = 0, /* Secure deletion */ 518 EXT4_INODE_UNRM = 1, /* Undelete */ 519 EXT4_INODE_COMPR = 2, /* Compress file */ 520 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 521 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 522 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 523 EXT4_INODE_NODUMP = 6, /* do not dump file */ 524 EXT4_INODE_NOATIME = 7, /* do not update atime */ 525 /* Reserved for compression usage... */ 526 EXT4_INODE_DIRTY = 8, 527 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 528 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 529 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 530 /* End compression flags --- maybe not all used */ 531 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 532 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 533 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 534 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 535 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 536 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 537 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 538 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 539 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 540 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 541 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 542 EXT4_INODE_DAX = 25, /* Inode is DAX */ 543 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 544 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 545 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 546 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 547 }; 548 549 /* 550 * Since it's pretty easy to mix up bit numbers and hex values, we use a 551 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 552 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 553 * any extra space in the compiled kernel image, otherwise, the build will fail. 554 * It's important that these values are the same, since we are using 555 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 556 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 557 * values found in ext2, ext3 and ext4 filesystems, and of course the values 558 * defined in e2fsprogs. 559 * 560 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 561 */ 562 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 563 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 564 565 static inline void ext4_check_flag_values(void) 566 { 567 CHECK_FLAG_VALUE(SECRM); 568 CHECK_FLAG_VALUE(UNRM); 569 CHECK_FLAG_VALUE(COMPR); 570 CHECK_FLAG_VALUE(SYNC); 571 CHECK_FLAG_VALUE(IMMUTABLE); 572 CHECK_FLAG_VALUE(APPEND); 573 CHECK_FLAG_VALUE(NODUMP); 574 CHECK_FLAG_VALUE(NOATIME); 575 CHECK_FLAG_VALUE(DIRTY); 576 CHECK_FLAG_VALUE(COMPRBLK); 577 CHECK_FLAG_VALUE(NOCOMPR); 578 CHECK_FLAG_VALUE(ENCRYPT); 579 CHECK_FLAG_VALUE(INDEX); 580 CHECK_FLAG_VALUE(IMAGIC); 581 CHECK_FLAG_VALUE(JOURNAL_DATA); 582 CHECK_FLAG_VALUE(NOTAIL); 583 CHECK_FLAG_VALUE(DIRSYNC); 584 CHECK_FLAG_VALUE(TOPDIR); 585 CHECK_FLAG_VALUE(HUGE_FILE); 586 CHECK_FLAG_VALUE(EXTENTS); 587 CHECK_FLAG_VALUE(VERITY); 588 CHECK_FLAG_VALUE(EA_INODE); 589 CHECK_FLAG_VALUE(INLINE_DATA); 590 CHECK_FLAG_VALUE(PROJINHERIT); 591 CHECK_FLAG_VALUE(CASEFOLD); 592 CHECK_FLAG_VALUE(RESERVED); 593 } 594 595 /* Used to pass group descriptor data when online resize is done */ 596 struct ext4_new_group_input { 597 __u32 group; /* Group number for this data */ 598 __u64 block_bitmap; /* Absolute block number of block bitmap */ 599 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 600 __u64 inode_table; /* Absolute block number of inode table start */ 601 __u32 blocks_count; /* Total number of blocks in this group */ 602 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 603 __u16 unused; 604 }; 605 606 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 607 struct compat_ext4_new_group_input { 608 u32 group; 609 compat_u64 block_bitmap; 610 compat_u64 inode_bitmap; 611 compat_u64 inode_table; 612 u32 blocks_count; 613 u16 reserved_blocks; 614 u16 unused; 615 }; 616 #endif 617 618 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 619 struct ext4_new_group_data { 620 __u32 group; 621 __u64 block_bitmap; 622 __u64 inode_bitmap; 623 __u64 inode_table; 624 __u32 blocks_count; 625 __u16 reserved_blocks; 626 __u16 mdata_blocks; 627 __u32 free_clusters_count; 628 }; 629 630 /* Indexes used to index group tables in ext4_new_group_data */ 631 enum { 632 BLOCK_BITMAP = 0, /* block bitmap */ 633 INODE_BITMAP, /* inode bitmap */ 634 INODE_TABLE, /* inode tables */ 635 GROUP_TABLE_COUNT, 636 }; 637 638 /* 639 * Flags used by ext4_map_blocks() 640 */ 641 /* Allocate any needed blocks and/or convert an unwritten 642 extent to be an initialized ext4 */ 643 #define EXT4_GET_BLOCKS_CREATE 0x0001 644 /* Request the creation of an unwritten extent */ 645 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 646 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 647 EXT4_GET_BLOCKS_CREATE) 648 /* Caller is from the delayed allocation writeout path 649 * finally doing the actual allocation of delayed blocks */ 650 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 651 /* caller is from the direct IO path, request to creation of an 652 unwritten extents if not allocated, split the unwritten 653 extent if blocks has been preallocated already*/ 654 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 655 #define EXT4_GET_BLOCKS_CONVERT 0x0010 656 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 657 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 658 /* Convert extent to initialized after IO complete */ 659 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 660 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 661 /* Eventual metadata allocation (due to growing extent tree) 662 * should not fail, so try to use reserved blocks for that.*/ 663 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 664 /* Don't normalize allocation size (used for fallocate) */ 665 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 666 /* Convert written extents to unwritten */ 667 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 668 /* Write zeros to newly created written extents */ 669 #define EXT4_GET_BLOCKS_ZERO 0x0200 670 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 671 EXT4_GET_BLOCKS_ZERO) 672 /* Caller will submit data before dropping transaction handle. This 673 * allows jbd2 to avoid submitting data before commit. */ 674 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 675 676 /* 677 * The bit position of these flags must not overlap with any of the 678 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 679 * read_extent_tree_block(), ext4_split_extent_at(), 680 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 681 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 682 * caching the extents when reading from the extent tree while a 683 * truncate or punch hole operation is in progress. 684 */ 685 #define EXT4_EX_NOCACHE 0x40000000 686 #define EXT4_EX_FORCE_CACHE 0x20000000 687 #define EXT4_EX_NOFAIL 0x10000000 688 689 /* 690 * Flags used by ext4_free_blocks 691 */ 692 #define EXT4_FREE_BLOCKS_METADATA 0x0001 693 #define EXT4_FREE_BLOCKS_FORGET 0x0002 694 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 695 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 696 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 697 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 698 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 699 700 /* 701 * ioctl commands 702 */ 703 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 704 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 705 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 706 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 707 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 708 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 709 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 710 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 711 #define EXT4_IOC_MIGRATE _IO('f', 9) 712 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 713 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 714 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 715 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 716 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 717 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 718 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 719 /* ioctl codes 19--39 are reserved for fscrypt */ 720 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 721 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 722 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 723 #define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32) 724 725 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 726 727 /* 728 * Flags for going down operation 729 */ 730 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 731 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 732 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 733 734 /* 735 * Flags returned by EXT4_IOC_GETSTATE 736 * 737 * We only expose to userspace a subset of the state flags in 738 * i_state_flags 739 */ 740 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 741 #define EXT4_STATE_FLAG_NEW 0x00000002 742 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 743 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 744 745 /* flags for ioctl EXT4_IOC_CHECKPOINT */ 746 #define EXT4_IOC_CHECKPOINT_FLAG_DISCARD 0x1 747 #define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT 0x2 748 #define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN 0x4 749 #define EXT4_IOC_CHECKPOINT_FLAG_VALID (EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \ 750 EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \ 751 EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN) 752 753 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 754 /* 755 * ioctl commands in 32 bit emulation 756 */ 757 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 758 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 759 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 760 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 761 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 762 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 763 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 764 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 765 #endif 766 767 /* 768 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 769 * It indicates that the entry in extent status cache is for a hole. 770 */ 771 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 772 773 /* Max physical block we can address w/o extents */ 774 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 775 776 /* Max logical block we can support */ 777 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 778 779 /* 780 * Structure of an inode on the disk 781 */ 782 struct ext4_inode { 783 __le16 i_mode; /* File mode */ 784 __le16 i_uid; /* Low 16 bits of Owner Uid */ 785 __le32 i_size_lo; /* Size in bytes */ 786 __le32 i_atime; /* Access time */ 787 __le32 i_ctime; /* Inode Change time */ 788 __le32 i_mtime; /* Modification time */ 789 __le32 i_dtime; /* Deletion Time */ 790 __le16 i_gid; /* Low 16 bits of Group Id */ 791 __le16 i_links_count; /* Links count */ 792 __le32 i_blocks_lo; /* Blocks count */ 793 __le32 i_flags; /* File flags */ 794 union { 795 struct { 796 __le32 l_i_version; 797 } linux1; 798 struct { 799 __u32 h_i_translator; 800 } hurd1; 801 struct { 802 __u32 m_i_reserved1; 803 } masix1; 804 } osd1; /* OS dependent 1 */ 805 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 806 __le32 i_generation; /* File version (for NFS) */ 807 __le32 i_file_acl_lo; /* File ACL */ 808 __le32 i_size_high; 809 __le32 i_obso_faddr; /* Obsoleted fragment address */ 810 union { 811 struct { 812 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 813 __le16 l_i_file_acl_high; 814 __le16 l_i_uid_high; /* these 2 fields */ 815 __le16 l_i_gid_high; /* were reserved2[0] */ 816 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 817 __le16 l_i_reserved; 818 } linux2; 819 struct { 820 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 821 __u16 h_i_mode_high; 822 __u16 h_i_uid_high; 823 __u16 h_i_gid_high; 824 __u32 h_i_author; 825 } hurd2; 826 struct { 827 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 828 __le16 m_i_file_acl_high; 829 __u32 m_i_reserved2[2]; 830 } masix2; 831 } osd2; /* OS dependent 2 */ 832 __le16 i_extra_isize; 833 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 834 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 835 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 836 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 837 __le32 i_crtime; /* File Creation time */ 838 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 839 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 840 __le32 i_projid; /* Project ID */ 841 }; 842 843 struct move_extent { 844 __u32 reserved; /* should be zero */ 845 __u32 donor_fd; /* donor file descriptor */ 846 __u64 orig_start; /* logical start offset in block for orig */ 847 __u64 donor_start; /* logical start offset in block for donor */ 848 __u64 len; /* block length to be moved */ 849 __u64 moved_len; /* moved block length */ 850 }; 851 852 #define EXT4_EPOCH_BITS 2 853 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 854 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 855 856 /* 857 * Extended fields will fit into an inode if the filesystem was formatted 858 * with large inodes (-I 256 or larger) and there are not currently any EAs 859 * consuming all of the available space. For new inodes we always reserve 860 * enough space for the kernel's known extended fields, but for inodes 861 * created with an old kernel this might not have been the case. None of 862 * the extended inode fields is critical for correct filesystem operation. 863 * This macro checks if a certain field fits in the inode. Note that 864 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 865 */ 866 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 867 ((offsetof(typeof(*ext4_inode), field) + \ 868 sizeof((ext4_inode)->field)) \ 869 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 870 (einode)->i_extra_isize)) \ 871 872 /* 873 * We use an encoding that preserves the times for extra epoch "00": 874 * 875 * extra msb of adjust for signed 876 * epoch 32-bit 32-bit tv_sec to 877 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 878 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 879 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 880 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 881 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 882 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 883 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 884 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 885 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 886 * 887 * Note that previous versions of the kernel on 64-bit systems would 888 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 889 * 1970. e2fsck will correct this, assuming that it is run on the 890 * affected filesystem before 2242. 891 */ 892 893 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 894 { 895 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 896 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 897 } 898 899 static inline void ext4_decode_extra_time(struct timespec64 *time, 900 __le32 extra) 901 { 902 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 903 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 904 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 905 } 906 907 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 908 do { \ 909 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 910 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 911 (raw_inode)->xtime ## _extra = \ 912 ext4_encode_extra_time(&(inode)->xtime); \ 913 } \ 914 else \ 915 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 916 } while (0) 917 918 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 919 do { \ 920 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 921 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 922 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 923 (raw_inode)->xtime ## _extra = \ 924 ext4_encode_extra_time(&(einode)->xtime); \ 925 } while (0) 926 927 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 928 do { \ 929 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 930 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 931 ext4_decode_extra_time(&(inode)->xtime, \ 932 raw_inode->xtime ## _extra); \ 933 } \ 934 else \ 935 (inode)->xtime.tv_nsec = 0; \ 936 } while (0) 937 938 939 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 940 do { \ 941 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 942 (einode)->xtime.tv_sec = \ 943 (signed)le32_to_cpu((raw_inode)->xtime); \ 944 else \ 945 (einode)->xtime.tv_sec = 0; \ 946 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 947 ext4_decode_extra_time(&(einode)->xtime, \ 948 raw_inode->xtime ## _extra); \ 949 else \ 950 (einode)->xtime.tv_nsec = 0; \ 951 } while (0) 952 953 #define i_disk_version osd1.linux1.l_i_version 954 955 #if defined(__KERNEL__) || defined(__linux__) 956 #define i_reserved1 osd1.linux1.l_i_reserved1 957 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 958 #define i_blocks_high osd2.linux2.l_i_blocks_high 959 #define i_uid_low i_uid 960 #define i_gid_low i_gid 961 #define i_uid_high osd2.linux2.l_i_uid_high 962 #define i_gid_high osd2.linux2.l_i_gid_high 963 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 964 965 #elif defined(__GNU__) 966 967 #define i_translator osd1.hurd1.h_i_translator 968 #define i_uid_high osd2.hurd2.h_i_uid_high 969 #define i_gid_high osd2.hurd2.h_i_gid_high 970 #define i_author osd2.hurd2.h_i_author 971 972 #elif defined(__masix__) 973 974 #define i_reserved1 osd1.masix1.m_i_reserved1 975 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 976 #define i_reserved2 osd2.masix2.m_i_reserved2 977 978 #endif /* defined(__KERNEL__) || defined(__linux__) */ 979 980 #include "extents_status.h" 981 #include "fast_commit.h" 982 983 /* 984 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 985 * 986 * These are needed to avoid lockdep false positives when we need to 987 * allocate blocks to the quota inode during ext4_map_blocks(), while 988 * holding i_data_sem for a normal (non-quota) inode. Since we don't 989 * do quota tracking for the quota inode, this avoids deadlock (as 990 * well as infinite recursion, since it isn't turtles all the way 991 * down...) 992 * 993 * I_DATA_SEM_NORMAL - Used for most inodes 994 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 995 * where the second inode has larger inode number 996 * than the first 997 * I_DATA_SEM_QUOTA - Used for quota inodes only 998 */ 999 enum { 1000 I_DATA_SEM_NORMAL = 0, 1001 I_DATA_SEM_OTHER, 1002 I_DATA_SEM_QUOTA, 1003 }; 1004 1005 1006 /* 1007 * fourth extended file system inode data in memory 1008 */ 1009 struct ext4_inode_info { 1010 __le32 i_data[15]; /* unconverted */ 1011 __u32 i_dtime; 1012 ext4_fsblk_t i_file_acl; 1013 1014 /* 1015 * i_block_group is the number of the block group which contains 1016 * this file's inode. Constant across the lifetime of the inode, 1017 * it is used for making block allocation decisions - we try to 1018 * place a file's data blocks near its inode block, and new inodes 1019 * near to their parent directory's inode. 1020 */ 1021 ext4_group_t i_block_group; 1022 ext4_lblk_t i_dir_start_lookup; 1023 #if (BITS_PER_LONG < 64) 1024 unsigned long i_state_flags; /* Dynamic state flags */ 1025 #endif 1026 unsigned long i_flags; 1027 1028 /* 1029 * Extended attributes can be read independently of the main file 1030 * data. Taking i_mutex even when reading would cause contention 1031 * between readers of EAs and writers of regular file data, so 1032 * instead we synchronize on xattr_sem when reading or changing 1033 * EAs. 1034 */ 1035 struct rw_semaphore xattr_sem; 1036 1037 /* 1038 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1039 * i_orphan is used. 1040 */ 1041 union { 1042 struct list_head i_orphan; /* unlinked but open inodes */ 1043 unsigned int i_orphan_idx; /* Index in orphan file */ 1044 }; 1045 1046 /* Fast commit related info */ 1047 1048 struct list_head i_fc_list; /* 1049 * inodes that need fast commit 1050 * protected by sbi->s_fc_lock. 1051 */ 1052 1053 /* Start of lblk range that needs to be committed in this fast commit */ 1054 ext4_lblk_t i_fc_lblk_start; 1055 1056 /* End of lblk range that needs to be committed in this fast commit */ 1057 ext4_lblk_t i_fc_lblk_len; 1058 1059 /* Number of ongoing updates on this inode */ 1060 atomic_t i_fc_updates; 1061 1062 /* Fast commit wait queue for this inode */ 1063 wait_queue_head_t i_fc_wait; 1064 1065 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1066 struct mutex i_fc_lock; 1067 1068 /* 1069 * i_disksize keeps track of what the inode size is ON DISK, not 1070 * in memory. During truncate, i_size is set to the new size by 1071 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1072 * set i_disksize to 0 until the truncate is actually under way. 1073 * 1074 * The intent is that i_disksize always represents the blocks which 1075 * are used by this file. This allows recovery to restart truncate 1076 * on orphans if we crash during truncate. We actually write i_disksize 1077 * into the on-disk inode when writing inodes out, instead of i_size. 1078 * 1079 * The only time when i_disksize and i_size may be different is when 1080 * a truncate is in progress. The only things which change i_disksize 1081 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1082 */ 1083 loff_t i_disksize; 1084 1085 /* 1086 * i_data_sem is for serialising ext4_truncate() against 1087 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1088 * data tree are chopped off during truncate. We can't do that in 1089 * ext4 because whenever we perform intermediate commits during 1090 * truncate, the inode and all the metadata blocks *must* be in a 1091 * consistent state which allows truncation of the orphans to restart 1092 * during recovery. Hence we must fix the get_block-vs-truncate race 1093 * by other means, so we have i_data_sem. 1094 */ 1095 struct rw_semaphore i_data_sem; 1096 struct inode vfs_inode; 1097 struct jbd2_inode *jinode; 1098 1099 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1100 1101 /* 1102 * File creation time. Its function is same as that of 1103 * struct timespec64 i_{a,c,m}time in the generic inode. 1104 */ 1105 struct timespec64 i_crtime; 1106 1107 /* mballoc */ 1108 atomic_t i_prealloc_active; 1109 struct list_head i_prealloc_list; 1110 spinlock_t i_prealloc_lock; 1111 1112 /* extents status tree */ 1113 struct ext4_es_tree i_es_tree; 1114 rwlock_t i_es_lock; 1115 struct list_head i_es_list; 1116 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1117 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1118 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1119 extents to shrink. Protected by 1120 i_es_lock */ 1121 1122 /* ialloc */ 1123 ext4_group_t i_last_alloc_group; 1124 1125 /* allocation reservation info for delalloc */ 1126 /* In case of bigalloc, this refer to clusters rather than blocks */ 1127 unsigned int i_reserved_data_blocks; 1128 1129 /* pending cluster reservations for bigalloc file systems */ 1130 struct ext4_pending_tree i_pending_tree; 1131 1132 /* on-disk additional length */ 1133 __u16 i_extra_isize; 1134 1135 /* Indicate the inline data space. */ 1136 u16 i_inline_off; 1137 u16 i_inline_size; 1138 1139 #ifdef CONFIG_QUOTA 1140 /* quota space reservation, managed internally by quota code */ 1141 qsize_t i_reserved_quota; 1142 #endif 1143 1144 /* Lock protecting lists below */ 1145 spinlock_t i_completed_io_lock; 1146 /* 1147 * Completed IOs that need unwritten extents handling and have 1148 * transaction reserved 1149 */ 1150 struct list_head i_rsv_conversion_list; 1151 struct work_struct i_rsv_conversion_work; 1152 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1153 1154 spinlock_t i_block_reservation_lock; 1155 1156 /* 1157 * Transactions that contain inode's metadata needed to complete 1158 * fsync and fdatasync, respectively. 1159 */ 1160 tid_t i_sync_tid; 1161 tid_t i_datasync_tid; 1162 1163 #ifdef CONFIG_QUOTA 1164 struct dquot *i_dquot[MAXQUOTAS]; 1165 #endif 1166 1167 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1168 __u32 i_csum_seed; 1169 1170 kprojid_t i_projid; 1171 }; 1172 1173 /* 1174 * File system states 1175 */ 1176 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1177 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1178 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1179 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1180 1181 /* 1182 * Misc. filesystem flags 1183 */ 1184 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1185 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1186 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1187 1188 /* 1189 * Mount flags set via mount options or defaults 1190 */ 1191 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1192 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1193 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1194 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1195 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1196 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1197 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1198 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1199 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1200 #ifdef CONFIG_FS_DAX 1201 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1202 #else 1203 #define EXT4_MOUNT_DAX_ALWAYS 0 1204 #endif 1205 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1206 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1207 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1208 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1209 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1210 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1211 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1212 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1213 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1214 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1215 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1216 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1217 * enable enforcement for hidden 1218 * quota files */ 1219 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1220 * enforcement for hidden quota 1221 * files */ 1222 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1223 * enforcement */ 1224 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1225 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1226 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1227 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1228 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1229 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1230 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1231 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1232 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1233 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1234 1235 /* 1236 * Mount flags set either automatically (could not be set by mount option) 1237 * based on per file system feature or property or in special cases such as 1238 * distinguishing between explicit mount option definition and default. 1239 */ 1240 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1241 specified delalloc */ 1242 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1243 size of blocksize * 8 1244 blocks */ 1245 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1246 file systems */ 1247 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1248 specified journal checksum */ 1249 1250 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1251 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1252 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1253 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1254 * scanning in mballoc 1255 */ 1256 1257 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1258 ~EXT4_MOUNT_##opt 1259 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1260 EXT4_MOUNT_##opt 1261 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1262 EXT4_MOUNT_##opt) 1263 1264 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1265 ~EXT4_MOUNT2_##opt 1266 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1267 EXT4_MOUNT2_##opt 1268 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1269 EXT4_MOUNT2_##opt) 1270 1271 #define ext4_test_and_set_bit __test_and_set_bit_le 1272 #define ext4_set_bit __set_bit_le 1273 #define ext4_set_bit_atomic ext2_set_bit_atomic 1274 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1275 #define ext4_clear_bit __clear_bit_le 1276 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1277 #define ext4_test_bit test_bit_le 1278 #define ext4_find_next_zero_bit find_next_zero_bit_le 1279 #define ext4_find_next_bit find_next_bit_le 1280 1281 extern void ext4_set_bits(void *bm, int cur, int len); 1282 1283 /* 1284 * Maximal mount counts between two filesystem checks 1285 */ 1286 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1287 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1288 1289 /* 1290 * Behaviour when detecting errors 1291 */ 1292 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1293 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1294 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1295 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1296 1297 /* Metadata checksum algorithm codes */ 1298 #define EXT4_CRC32C_CHKSUM 1 1299 1300 /* 1301 * Structure of the super block 1302 */ 1303 struct ext4_super_block { 1304 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1305 __le32 s_blocks_count_lo; /* Blocks count */ 1306 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1307 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1308 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1309 __le32 s_first_data_block; /* First Data Block */ 1310 __le32 s_log_block_size; /* Block size */ 1311 __le32 s_log_cluster_size; /* Allocation cluster size */ 1312 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1313 __le32 s_clusters_per_group; /* # Clusters per group */ 1314 __le32 s_inodes_per_group; /* # Inodes per group */ 1315 __le32 s_mtime; /* Mount time */ 1316 /*30*/ __le32 s_wtime; /* Write time */ 1317 __le16 s_mnt_count; /* Mount count */ 1318 __le16 s_max_mnt_count; /* Maximal mount count */ 1319 __le16 s_magic; /* Magic signature */ 1320 __le16 s_state; /* File system state */ 1321 __le16 s_errors; /* Behaviour when detecting errors */ 1322 __le16 s_minor_rev_level; /* minor revision level */ 1323 /*40*/ __le32 s_lastcheck; /* time of last check */ 1324 __le32 s_checkinterval; /* max. time between checks */ 1325 __le32 s_creator_os; /* OS */ 1326 __le32 s_rev_level; /* Revision level */ 1327 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1328 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1329 /* 1330 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1331 * 1332 * Note: the difference between the compatible feature set and 1333 * the incompatible feature set is that if there is a bit set 1334 * in the incompatible feature set that the kernel doesn't 1335 * know about, it should refuse to mount the filesystem. 1336 * 1337 * e2fsck's requirements are more strict; if it doesn't know 1338 * about a feature in either the compatible or incompatible 1339 * feature set, it must abort and not try to meddle with 1340 * things it doesn't understand... 1341 */ 1342 __le32 s_first_ino; /* First non-reserved inode */ 1343 __le16 s_inode_size; /* size of inode structure */ 1344 __le16 s_block_group_nr; /* block group # of this superblock */ 1345 __le32 s_feature_compat; /* compatible feature set */ 1346 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1347 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1348 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1349 /*78*/ char s_volume_name[16]; /* volume name */ 1350 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1351 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1352 /* 1353 * Performance hints. Directory preallocation should only 1354 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1355 */ 1356 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1357 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1358 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1359 /* 1360 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1361 */ 1362 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1363 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1364 __le32 s_journal_dev; /* device number of journal file */ 1365 __le32 s_last_orphan; /* start of list of inodes to delete */ 1366 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1367 __u8 s_def_hash_version; /* Default hash version to use */ 1368 __u8 s_jnl_backup_type; 1369 __le16 s_desc_size; /* size of group descriptor */ 1370 /*100*/ __le32 s_default_mount_opts; 1371 __le32 s_first_meta_bg; /* First metablock block group */ 1372 __le32 s_mkfs_time; /* When the filesystem was created */ 1373 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1374 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1375 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1376 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1377 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1378 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1379 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1380 __le32 s_flags; /* Miscellaneous flags */ 1381 __le16 s_raid_stride; /* RAID stride */ 1382 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1383 __le64 s_mmp_block; /* Block for multi-mount protection */ 1384 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1385 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1386 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1387 __u8 s_encryption_level; /* versioning level for encryption */ 1388 __u8 s_reserved_pad; /* Padding to next 32bits */ 1389 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1390 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1391 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1392 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1393 snapshot's future use */ 1394 __le32 s_snapshot_list; /* inode number of the head of the 1395 on-disk snapshot list */ 1396 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1397 __le32 s_error_count; /* number of fs errors */ 1398 __le32 s_first_error_time; /* first time an error happened */ 1399 __le32 s_first_error_ino; /* inode involved in first error */ 1400 __le64 s_first_error_block; /* block involved of first error */ 1401 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1402 __le32 s_first_error_line; /* line number where error happened */ 1403 __le32 s_last_error_time; /* most recent time of an error */ 1404 __le32 s_last_error_ino; /* inode involved in last error */ 1405 __le32 s_last_error_line; /* line number where error happened */ 1406 __le64 s_last_error_block; /* block involved of last error */ 1407 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1408 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1409 __u8 s_mount_opts[64]; 1410 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1411 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1412 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1413 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1414 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1415 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1416 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1417 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1418 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1419 __u8 s_wtime_hi; 1420 __u8 s_mtime_hi; 1421 __u8 s_mkfs_time_hi; 1422 __u8 s_lastcheck_hi; 1423 __u8 s_first_error_time_hi; 1424 __u8 s_last_error_time_hi; 1425 __u8 s_first_error_errcode; 1426 __u8 s_last_error_errcode; 1427 __le16 s_encoding; /* Filename charset encoding */ 1428 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1429 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1430 __le32 s_reserved[94]; /* Padding to the end of the block */ 1431 __le32 s_checksum; /* crc32c(superblock) */ 1432 }; 1433 1434 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1435 1436 #ifdef __KERNEL__ 1437 1438 #ifdef CONFIG_FS_ENCRYPTION 1439 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL) 1440 #else 1441 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1442 #endif 1443 1444 /* Number of quota types we support */ 1445 #define EXT4_MAXQUOTAS 3 1446 1447 #define EXT4_ENC_UTF8_12_1 1 1448 1449 /* Types of ext4 journal triggers */ 1450 enum ext4_journal_trigger_type { 1451 EXT4_JTR_ORPHAN_FILE, 1452 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1453 }; 1454 1455 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1456 1457 struct ext4_journal_trigger { 1458 struct jbd2_buffer_trigger_type tr_triggers; 1459 struct super_block *sb; 1460 }; 1461 1462 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1463 struct jbd2_buffer_trigger_type *trigger) 1464 { 1465 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1466 } 1467 1468 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1469 1470 /* Structure at the tail of orphan block */ 1471 struct ext4_orphan_block_tail { 1472 __le32 ob_magic; 1473 __le32 ob_checksum; 1474 }; 1475 1476 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1477 { 1478 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1479 sizeof(u32); 1480 } 1481 1482 struct ext4_orphan_block { 1483 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1484 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1485 }; 1486 1487 /* 1488 * Info about orphan file. 1489 */ 1490 struct ext4_orphan_info { 1491 int of_blocks; /* Number of orphan blocks in a file */ 1492 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1493 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1494 * file blocks */ 1495 }; 1496 1497 /* 1498 * fourth extended-fs super-block data in memory 1499 */ 1500 struct ext4_sb_info { 1501 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1502 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1503 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1504 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1505 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1506 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1507 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1508 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1509 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1510 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1511 unsigned long s_overhead; /* # of fs overhead clusters */ 1512 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1513 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1514 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1515 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1516 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1517 struct buffer_head * __rcu *s_group_desc; 1518 unsigned int s_mount_opt; 1519 unsigned int s_mount_opt2; 1520 unsigned long s_mount_flags; 1521 unsigned int s_def_mount_opt; 1522 ext4_fsblk_t s_sb_block; 1523 atomic64_t s_resv_clusters; 1524 kuid_t s_resuid; 1525 kgid_t s_resgid; 1526 unsigned short s_mount_state; 1527 unsigned short s_pad; 1528 int s_addr_per_block_bits; 1529 int s_desc_per_block_bits; 1530 int s_inode_size; 1531 int s_first_ino; 1532 unsigned int s_inode_readahead_blks; 1533 unsigned int s_inode_goal; 1534 u32 s_hash_seed[4]; 1535 int s_def_hash_version; 1536 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1537 struct percpu_counter s_freeclusters_counter; 1538 struct percpu_counter s_freeinodes_counter; 1539 struct percpu_counter s_dirs_counter; 1540 struct percpu_counter s_dirtyclusters_counter; 1541 struct percpu_counter s_sra_exceeded_retry_limit; 1542 struct blockgroup_lock *s_blockgroup_lock; 1543 struct proc_dir_entry *s_proc; 1544 struct kobject s_kobj; 1545 struct completion s_kobj_unregister; 1546 struct super_block *s_sb; 1547 struct buffer_head *s_mmp_bh; 1548 1549 /* Journaling */ 1550 struct journal_s *s_journal; 1551 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1552 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1553 struct list_head s_orphan; /* List of orphaned inodes in on disk 1554 list */ 1555 struct ext4_orphan_info s_orphan_info; 1556 unsigned long s_commit_interval; 1557 u32 s_max_batch_time; 1558 u32 s_min_batch_time; 1559 struct block_device *s_journal_bdev; 1560 #ifdef CONFIG_QUOTA 1561 /* Names of quota files with journalled quota */ 1562 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1563 int s_jquota_fmt; /* Format of quota to use */ 1564 #endif 1565 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1566 struct ext4_system_blocks __rcu *s_system_blks; 1567 1568 #ifdef EXTENTS_STATS 1569 /* ext4 extents stats */ 1570 unsigned long s_ext_min; 1571 unsigned long s_ext_max; 1572 unsigned long s_depth_max; 1573 spinlock_t s_ext_stats_lock; 1574 unsigned long s_ext_blocks; 1575 unsigned long s_ext_extents; 1576 #endif 1577 1578 /* for buddy allocator */ 1579 struct ext4_group_info ** __rcu *s_group_info; 1580 struct inode *s_buddy_cache; 1581 spinlock_t s_md_lock; 1582 unsigned short *s_mb_offsets; 1583 unsigned int *s_mb_maxs; 1584 unsigned int s_group_info_size; 1585 unsigned int s_mb_free_pending; 1586 struct list_head s_freed_data_list; /* List of blocks to be freed 1587 after commit completed */ 1588 struct list_head s_discard_list; 1589 struct work_struct s_discard_work; 1590 atomic_t s_retry_alloc_pending; 1591 struct rb_root s_mb_avg_fragment_size_root; 1592 rwlock_t s_mb_rb_lock; 1593 struct list_head *s_mb_largest_free_orders; 1594 rwlock_t *s_mb_largest_free_orders_locks; 1595 1596 /* tunables */ 1597 unsigned long s_stripe; 1598 unsigned int s_mb_max_linear_groups; 1599 unsigned int s_mb_stream_request; 1600 unsigned int s_mb_max_to_scan; 1601 unsigned int s_mb_min_to_scan; 1602 unsigned int s_mb_stats; 1603 unsigned int s_mb_order2_reqs; 1604 unsigned int s_mb_group_prealloc; 1605 unsigned int s_mb_max_inode_prealloc; 1606 unsigned int s_max_dir_size_kb; 1607 /* where last allocation was done - for stream allocation */ 1608 unsigned long s_mb_last_group; 1609 unsigned long s_mb_last_start; 1610 unsigned int s_mb_prefetch; 1611 unsigned int s_mb_prefetch_limit; 1612 1613 /* stats for buddy allocator */ 1614 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1615 atomic_t s_bal_success; /* we found long enough chunks */ 1616 atomic_t s_bal_allocated; /* in blocks */ 1617 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1618 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1619 atomic_t s_bal_goals; /* goal hits */ 1620 atomic_t s_bal_breaks; /* too long searches */ 1621 atomic_t s_bal_2orders; /* 2^order hits */ 1622 atomic_t s_bal_cr0_bad_suggestions; 1623 atomic_t s_bal_cr1_bad_suggestions; 1624 atomic64_t s_bal_cX_groups_considered[4]; 1625 atomic64_t s_bal_cX_hits[4]; 1626 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1627 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1628 atomic64_t s_mb_generation_time; 1629 atomic_t s_mb_lost_chunks; 1630 atomic_t s_mb_preallocated; 1631 atomic_t s_mb_discarded; 1632 atomic_t s_lock_busy; 1633 1634 /* locality groups */ 1635 struct ext4_locality_group __percpu *s_locality_groups; 1636 1637 /* for write statistics */ 1638 unsigned long s_sectors_written_start; 1639 u64 s_kbytes_written; 1640 1641 /* the size of zero-out chunk */ 1642 unsigned int s_extent_max_zeroout_kb; 1643 1644 unsigned int s_log_groups_per_flex; 1645 struct flex_groups * __rcu *s_flex_groups; 1646 ext4_group_t s_flex_groups_allocated; 1647 1648 /* workqueue for reserved extent conversions (buffered io) */ 1649 struct workqueue_struct *rsv_conversion_wq; 1650 1651 /* timer for periodic error stats printing */ 1652 struct timer_list s_err_report; 1653 1654 /* Lazy inode table initialization info */ 1655 struct ext4_li_request *s_li_request; 1656 /* Wait multiplier for lazy initialization thread */ 1657 unsigned int s_li_wait_mult; 1658 1659 /* Kernel thread for multiple mount protection */ 1660 struct task_struct *s_mmp_tsk; 1661 1662 /* record the last minlen when FITRIM is called. */ 1663 atomic_t s_last_trim_minblks; 1664 1665 /* Reference to checksum algorithm driver via cryptoapi */ 1666 struct crypto_shash *s_chksum_driver; 1667 1668 /* Precomputed FS UUID checksum for seeding other checksums */ 1669 __u32 s_csum_seed; 1670 1671 /* Reclaim extents from extent status tree */ 1672 struct shrinker s_es_shrinker; 1673 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1674 long s_es_nr_inode; 1675 struct ext4_es_stats s_es_stats; 1676 struct mb_cache *s_ea_block_cache; 1677 struct mb_cache *s_ea_inode_cache; 1678 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1679 1680 /* Journal triggers for checksum computation */ 1681 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1682 1683 /* Ratelimit ext4 messages. */ 1684 struct ratelimit_state s_err_ratelimit_state; 1685 struct ratelimit_state s_warning_ratelimit_state; 1686 struct ratelimit_state s_msg_ratelimit_state; 1687 atomic_t s_warning_count; 1688 atomic_t s_msg_count; 1689 1690 /* Encryption policy for '-o test_dummy_encryption' */ 1691 struct fscrypt_dummy_policy s_dummy_enc_policy; 1692 1693 /* 1694 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1695 * or EXTENTS flag. 1696 */ 1697 struct percpu_rw_semaphore s_writepages_rwsem; 1698 struct dax_device *s_daxdev; 1699 #ifdef CONFIG_EXT4_DEBUG 1700 unsigned long s_simulate_fail; 1701 #endif 1702 /* Record the errseq of the backing block device */ 1703 errseq_t s_bdev_wb_err; 1704 spinlock_t s_bdev_wb_lock; 1705 1706 /* Information about errors that happened during this mount */ 1707 spinlock_t s_error_lock; 1708 int s_add_error_count; 1709 int s_first_error_code; 1710 __u32 s_first_error_line; 1711 __u32 s_first_error_ino; 1712 __u64 s_first_error_block; 1713 const char *s_first_error_func; 1714 time64_t s_first_error_time; 1715 int s_last_error_code; 1716 __u32 s_last_error_line; 1717 __u32 s_last_error_ino; 1718 __u64 s_last_error_block; 1719 const char *s_last_error_func; 1720 time64_t s_last_error_time; 1721 /* 1722 * If we are in a context where we cannot update error information in 1723 * the on-disk superblock, we queue this work to do it. 1724 */ 1725 struct work_struct s_error_work; 1726 1727 /* Ext4 fast commit sub transaction ID */ 1728 atomic_t s_fc_subtid; 1729 1730 /* 1731 * After commit starts, the main queue gets locked, and the further 1732 * updates get added in the staging queue. 1733 */ 1734 #define FC_Q_MAIN 0 1735 #define FC_Q_STAGING 1 1736 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1737 * that have data changes in them. 1738 */ 1739 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1740 unsigned int s_fc_bytes; 1741 /* 1742 * Main fast commit lock. This lock protects accesses to the 1743 * following fields: 1744 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1745 */ 1746 spinlock_t s_fc_lock; 1747 struct buffer_head *s_fc_bh; 1748 struct ext4_fc_stats s_fc_stats; 1749 tid_t s_fc_ineligible_tid; 1750 #ifdef CONFIG_EXT4_DEBUG 1751 int s_fc_debug_max_replay; 1752 #endif 1753 struct ext4_fc_replay_state s_fc_replay_state; 1754 }; 1755 1756 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1757 { 1758 return sb->s_fs_info; 1759 } 1760 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1761 { 1762 return container_of(inode, struct ext4_inode_info, vfs_inode); 1763 } 1764 1765 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1766 { 1767 return ino == EXT4_ROOT_INO || 1768 (ino >= EXT4_FIRST_INO(sb) && 1769 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1770 } 1771 1772 /* 1773 * Returns: sbi->field[index] 1774 * Used to access an array element from the following sbi fields which require 1775 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1776 * - s_group_desc 1777 * - s_group_info 1778 * - s_flex_group 1779 */ 1780 #define sbi_array_rcu_deref(sbi, field, index) \ 1781 ({ \ 1782 typeof(*((sbi)->field)) _v; \ 1783 rcu_read_lock(); \ 1784 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1785 rcu_read_unlock(); \ 1786 _v; \ 1787 }) 1788 1789 /* 1790 * run-time mount flags 1791 */ 1792 enum { 1793 EXT4_MF_MNTDIR_SAMPLED, 1794 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1795 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1796 }; 1797 1798 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1799 { 1800 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1801 } 1802 1803 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1804 { 1805 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1806 } 1807 1808 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1809 { 1810 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1811 } 1812 1813 1814 /* 1815 * Simulate_fail codes 1816 */ 1817 #define EXT4_SIM_BBITMAP_EIO 1 1818 #define EXT4_SIM_BBITMAP_CRC 2 1819 #define EXT4_SIM_IBITMAP_EIO 3 1820 #define EXT4_SIM_IBITMAP_CRC 4 1821 #define EXT4_SIM_INODE_EIO 5 1822 #define EXT4_SIM_INODE_CRC 6 1823 #define EXT4_SIM_DIRBLOCK_EIO 7 1824 #define EXT4_SIM_DIRBLOCK_CRC 8 1825 1826 static inline bool ext4_simulate_fail(struct super_block *sb, 1827 unsigned long code) 1828 { 1829 #ifdef CONFIG_EXT4_DEBUG 1830 struct ext4_sb_info *sbi = EXT4_SB(sb); 1831 1832 if (unlikely(sbi->s_simulate_fail == code)) { 1833 sbi->s_simulate_fail = 0; 1834 return true; 1835 } 1836 #endif 1837 return false; 1838 } 1839 1840 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1841 struct buffer_head *bh, 1842 unsigned long code) 1843 { 1844 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1845 clear_buffer_uptodate(bh); 1846 } 1847 1848 /* 1849 * Error number codes for s_{first,last}_error_errno 1850 * 1851 * Linux errno numbers are architecture specific, so we need to translate 1852 * them into something which is architecture independent. We don't define 1853 * codes for all errno's; just the ones which are most likely to be the cause 1854 * of an ext4_error() call. 1855 */ 1856 #define EXT4_ERR_UNKNOWN 1 1857 #define EXT4_ERR_EIO 2 1858 #define EXT4_ERR_ENOMEM 3 1859 #define EXT4_ERR_EFSBADCRC 4 1860 #define EXT4_ERR_EFSCORRUPTED 5 1861 #define EXT4_ERR_ENOSPC 6 1862 #define EXT4_ERR_ENOKEY 7 1863 #define EXT4_ERR_EROFS 8 1864 #define EXT4_ERR_EFBIG 9 1865 #define EXT4_ERR_EEXIST 10 1866 #define EXT4_ERR_ERANGE 11 1867 #define EXT4_ERR_EOVERFLOW 12 1868 #define EXT4_ERR_EBUSY 13 1869 #define EXT4_ERR_ENOTDIR 14 1870 #define EXT4_ERR_ENOTEMPTY 15 1871 #define EXT4_ERR_ESHUTDOWN 16 1872 #define EXT4_ERR_EFAULT 17 1873 1874 /* 1875 * Inode dynamic state flags 1876 */ 1877 enum { 1878 EXT4_STATE_JDATA, /* journaled data exists */ 1879 EXT4_STATE_NEW, /* inode is newly created */ 1880 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1881 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1882 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1883 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1884 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1885 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1886 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1887 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1888 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1889 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1890 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1891 }; 1892 1893 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1894 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1895 { \ 1896 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1897 } \ 1898 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1899 { \ 1900 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1901 } \ 1902 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1903 { \ 1904 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1905 } 1906 1907 /* Add these declarations here only so that these functions can be 1908 * found by name. Otherwise, they are very hard to locate. */ 1909 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1910 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1911 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1912 EXT4_INODE_BIT_FNS(flag, flags, 0) 1913 1914 /* Add these declarations here only so that these functions can be 1915 * found by name. Otherwise, they are very hard to locate. */ 1916 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1917 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1918 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1919 #if (BITS_PER_LONG < 64) 1920 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1921 1922 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1923 { 1924 (ei)->i_state_flags = 0; 1925 } 1926 #else 1927 EXT4_INODE_BIT_FNS(state, flags, 32) 1928 1929 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1930 { 1931 /* We depend on the fact that callers will set i_flags */ 1932 } 1933 #endif 1934 #else 1935 /* Assume that user mode programs are passing in an ext4fs superblock, not 1936 * a kernel struct super_block. This will allow us to call the feature-test 1937 * macros from user land. */ 1938 #define EXT4_SB(sb) (sb) 1939 #endif 1940 1941 static inline bool ext4_verity_in_progress(struct inode *inode) 1942 { 1943 return IS_ENABLED(CONFIG_FS_VERITY) && 1944 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1945 } 1946 1947 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1948 1949 /* 1950 * Codes for operating systems 1951 */ 1952 #define EXT4_OS_LINUX 0 1953 #define EXT4_OS_HURD 1 1954 #define EXT4_OS_MASIX 2 1955 #define EXT4_OS_FREEBSD 3 1956 #define EXT4_OS_LITES 4 1957 1958 /* 1959 * Revision levels 1960 */ 1961 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1962 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1963 1964 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1965 1966 #define EXT4_GOOD_OLD_INODE_SIZE 128 1967 1968 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1969 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1970 #define EXT4_TIMESTAMP_MIN S32_MIN 1971 1972 /* 1973 * Feature set definitions 1974 */ 1975 1976 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1977 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1978 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1979 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1980 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1981 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1982 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1983 /* 1984 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1985 * incompatible only if fast commit blocks are present in the FS. Since we 1986 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1987 * incompatible. We also have a JBD2 incompat feature, which gets set when 1988 * there are fast commit blocks present in the journal. 1989 */ 1990 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1991 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1992 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1993 1994 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1995 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1996 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1997 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1998 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1999 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2000 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2001 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2002 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2003 /* 2004 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2005 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2006 * all other data structures' checksums. However, the METADATA_CSUM and 2007 * GDT_CSUM bits are mutually exclusive. 2008 */ 2009 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2010 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2011 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2012 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2013 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2014 non-empty */ 2015 2016 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2017 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2018 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2019 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2020 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2021 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2022 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2023 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2024 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2025 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2026 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2027 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2028 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2029 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2030 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2031 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2032 2033 extern void ext4_update_dynamic_rev(struct super_block *sb); 2034 2035 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2036 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2037 { \ 2038 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2039 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2040 } \ 2041 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2042 { \ 2043 ext4_update_dynamic_rev(sb); \ 2044 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2045 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2046 } \ 2047 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2048 { \ 2049 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2050 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2051 } 2052 2053 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2054 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2055 { \ 2056 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2057 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2058 } \ 2059 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2060 { \ 2061 ext4_update_dynamic_rev(sb); \ 2062 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2063 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2064 } \ 2065 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2066 { \ 2067 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2068 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2069 } 2070 2071 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2072 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2073 { \ 2074 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2075 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2076 } \ 2077 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2078 { \ 2079 ext4_update_dynamic_rev(sb); \ 2080 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2081 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2082 } \ 2083 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2084 { \ 2085 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2086 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2087 } 2088 2089 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2090 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2091 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2092 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2093 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2094 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2095 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2096 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2097 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2098 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2099 2100 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2101 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2102 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2103 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2104 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2105 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2106 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2107 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2108 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2109 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2110 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2111 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2112 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2113 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2114 2115 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2116 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2117 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2118 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2119 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2120 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2121 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2122 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2123 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2124 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2125 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2126 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2127 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2128 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2129 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2130 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2131 2132 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2133 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2134 EXT4_FEATURE_INCOMPAT_META_BG) 2135 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2136 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2137 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2138 2139 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2140 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2141 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2142 EXT4_FEATURE_INCOMPAT_META_BG) 2143 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2144 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2145 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2146 2147 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2148 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2149 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2150 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2151 EXT4_FEATURE_INCOMPAT_META_BG| \ 2152 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2153 EXT4_FEATURE_INCOMPAT_64BIT| \ 2154 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2155 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2156 EXT4_FEATURE_INCOMPAT_MMP | \ 2157 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2158 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2159 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2160 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2161 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2162 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2163 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2164 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2165 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2166 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2167 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2168 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2169 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2170 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2171 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2172 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2173 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2174 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2175 2176 #define EXTN_FEATURE_FUNCS(ver) \ 2177 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2178 { \ 2179 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2180 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2181 } \ 2182 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2183 { \ 2184 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2185 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2186 } \ 2187 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2188 { \ 2189 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2190 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2191 } 2192 2193 EXTN_FEATURE_FUNCS(2) 2194 EXTN_FEATURE_FUNCS(3) 2195 EXTN_FEATURE_FUNCS(4) 2196 2197 static inline bool ext4_has_compat_features(struct super_block *sb) 2198 { 2199 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2200 } 2201 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2202 { 2203 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2204 } 2205 static inline bool ext4_has_incompat_features(struct super_block *sb) 2206 { 2207 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2208 } 2209 2210 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2211 2212 /* 2213 * Superblock flags 2214 */ 2215 #define EXT4_FLAGS_RESIZING 0 2216 #define EXT4_FLAGS_SHUTDOWN 1 2217 #define EXT4_FLAGS_BDEV_IS_DAX 2 2218 2219 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2220 { 2221 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2222 } 2223 2224 /* 2225 * Default values for user and/or group using reserved blocks 2226 */ 2227 #define EXT4_DEF_RESUID 0 2228 #define EXT4_DEF_RESGID 0 2229 2230 /* 2231 * Default project ID 2232 */ 2233 #define EXT4_DEF_PROJID 0 2234 2235 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2236 2237 /* 2238 * Default mount options 2239 */ 2240 #define EXT4_DEFM_DEBUG 0x0001 2241 #define EXT4_DEFM_BSDGROUPS 0x0002 2242 #define EXT4_DEFM_XATTR_USER 0x0004 2243 #define EXT4_DEFM_ACL 0x0008 2244 #define EXT4_DEFM_UID16 0x0010 2245 #define EXT4_DEFM_JMODE 0x0060 2246 #define EXT4_DEFM_JMODE_DATA 0x0020 2247 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2248 #define EXT4_DEFM_JMODE_WBACK 0x0060 2249 #define EXT4_DEFM_NOBARRIER 0x0100 2250 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2251 #define EXT4_DEFM_DISCARD 0x0400 2252 #define EXT4_DEFM_NODELALLOC 0x0800 2253 2254 /* 2255 * Default journal batch times 2256 */ 2257 #define EXT4_DEF_MIN_BATCH_TIME 0 2258 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2259 2260 /* 2261 * Minimum number of groups in a flexgroup before we separate out 2262 * directories into the first block group of a flexgroup 2263 */ 2264 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2265 2266 /* 2267 * Structure of a directory entry 2268 */ 2269 #define EXT4_NAME_LEN 255 2270 2271 struct ext4_dir_entry { 2272 __le32 inode; /* Inode number */ 2273 __le16 rec_len; /* Directory entry length */ 2274 __le16 name_len; /* Name length */ 2275 char name[EXT4_NAME_LEN]; /* File name */ 2276 }; 2277 2278 2279 /* 2280 * Encrypted Casefolded entries require saving the hash on disk. This structure 2281 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2282 * boundary. 2283 */ 2284 struct ext4_dir_entry_hash { 2285 __le32 hash; 2286 __le32 minor_hash; 2287 }; 2288 2289 /* 2290 * The new version of the directory entry. Since EXT4 structures are 2291 * stored in intel byte order, and the name_len field could never be 2292 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2293 * file_type field. 2294 */ 2295 struct ext4_dir_entry_2 { 2296 __le32 inode; /* Inode number */ 2297 __le16 rec_len; /* Directory entry length */ 2298 __u8 name_len; /* Name length */ 2299 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2300 char name[EXT4_NAME_LEN]; /* File name */ 2301 }; 2302 2303 /* 2304 * Access the hashes at the end of ext4_dir_entry_2 2305 */ 2306 #define EXT4_DIRENT_HASHES(entry) \ 2307 ((struct ext4_dir_entry_hash *) \ 2308 (((void *)(entry)) + \ 2309 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2310 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2311 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2312 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2313 2314 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2315 { 2316 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2317 } 2318 2319 /* 2320 * This is a bogus directory entry at the end of each leaf block that 2321 * records checksums. 2322 */ 2323 struct ext4_dir_entry_tail { 2324 __le32 det_reserved_zero1; /* Pretend to be unused */ 2325 __le16 det_rec_len; /* 12 */ 2326 __u8 det_reserved_zero2; /* Zero name length */ 2327 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2328 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2329 }; 2330 2331 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2332 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2333 ((blocksize) - \ 2334 sizeof(struct ext4_dir_entry_tail)))) 2335 2336 /* 2337 * Ext4 directory file types. Only the low 3 bits are used. The 2338 * other bits are reserved for now. 2339 */ 2340 #define EXT4_FT_UNKNOWN 0 2341 #define EXT4_FT_REG_FILE 1 2342 #define EXT4_FT_DIR 2 2343 #define EXT4_FT_CHRDEV 3 2344 #define EXT4_FT_BLKDEV 4 2345 #define EXT4_FT_FIFO 5 2346 #define EXT4_FT_SOCK 6 2347 #define EXT4_FT_SYMLINK 7 2348 2349 #define EXT4_FT_MAX 8 2350 2351 #define EXT4_FT_DIR_CSUM 0xDE 2352 2353 /* 2354 * EXT4_DIR_PAD defines the directory entries boundaries 2355 * 2356 * NOTE: It must be a multiple of 4 2357 */ 2358 #define EXT4_DIR_PAD 4 2359 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2360 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2361 2362 /* 2363 * The rec_len is dependent on the type of directory. Directories that are 2364 * casefolded and encrypted need to store the hash as well, so we add room for 2365 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2366 * pass NULL for dir, as those entries do not use the extra fields. 2367 */ 2368 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2369 const struct inode *dir) 2370 { 2371 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2372 2373 if (dir && ext4_hash_in_dirent(dir)) 2374 rec_len += sizeof(struct ext4_dir_entry_hash); 2375 return (rec_len & ~EXT4_DIR_ROUND); 2376 } 2377 2378 /* 2379 * If we ever get support for fs block sizes > page_size, we'll need 2380 * to remove the #if statements in the next two functions... 2381 */ 2382 static inline unsigned int 2383 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2384 { 2385 unsigned len = le16_to_cpu(dlen); 2386 2387 #if (PAGE_SIZE >= 65536) 2388 if (len == EXT4_MAX_REC_LEN || len == 0) 2389 return blocksize; 2390 return (len & 65532) | ((len & 3) << 16); 2391 #else 2392 return len; 2393 #endif 2394 } 2395 2396 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2397 { 2398 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2399 BUG(); 2400 #if (PAGE_SIZE >= 65536) 2401 if (len < 65536) 2402 return cpu_to_le16(len); 2403 if (len == blocksize) { 2404 if (blocksize == 65536) 2405 return cpu_to_le16(EXT4_MAX_REC_LEN); 2406 else 2407 return cpu_to_le16(0); 2408 } 2409 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2410 #else 2411 return cpu_to_le16(len); 2412 #endif 2413 } 2414 2415 /* 2416 * Hash Tree Directory indexing 2417 * (c) Daniel Phillips, 2001 2418 */ 2419 2420 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2421 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2422 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2423 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2424 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2425 2426 /* Legal values for the dx_root hash_version field: */ 2427 2428 #define DX_HASH_LEGACY 0 2429 #define DX_HASH_HALF_MD4 1 2430 #define DX_HASH_TEA 2 2431 #define DX_HASH_LEGACY_UNSIGNED 3 2432 #define DX_HASH_HALF_MD4_UNSIGNED 4 2433 #define DX_HASH_TEA_UNSIGNED 5 2434 #define DX_HASH_SIPHASH 6 2435 2436 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2437 const void *address, unsigned int length) 2438 { 2439 struct { 2440 struct shash_desc shash; 2441 char ctx[4]; 2442 } desc; 2443 2444 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2445 2446 desc.shash.tfm = sbi->s_chksum_driver; 2447 *(u32 *)desc.ctx = crc; 2448 2449 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2450 2451 return *(u32 *)desc.ctx; 2452 } 2453 2454 #ifdef __KERNEL__ 2455 2456 /* hash info structure used by the directory hash */ 2457 struct dx_hash_info 2458 { 2459 u32 hash; 2460 u32 minor_hash; 2461 int hash_version; 2462 u32 *seed; 2463 }; 2464 2465 2466 /* 32 and 64 bit signed EOF for dx directories */ 2467 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2468 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2469 2470 2471 /* 2472 * Control parameters used by ext4_htree_next_block 2473 */ 2474 #define HASH_NB_ALWAYS 1 2475 2476 struct ext4_filename { 2477 const struct qstr *usr_fname; 2478 struct fscrypt_str disk_name; 2479 struct dx_hash_info hinfo; 2480 #ifdef CONFIG_FS_ENCRYPTION 2481 struct fscrypt_str crypto_buf; 2482 #endif 2483 #ifdef CONFIG_UNICODE 2484 struct fscrypt_str cf_name; 2485 #endif 2486 }; 2487 2488 #define fname_name(p) ((p)->disk_name.name) 2489 #define fname_usr_name(p) ((p)->usr_fname->name) 2490 #define fname_len(p) ((p)->disk_name.len) 2491 2492 /* 2493 * Describe an inode's exact location on disk and in memory 2494 */ 2495 struct ext4_iloc 2496 { 2497 struct buffer_head *bh; 2498 unsigned long offset; 2499 ext4_group_t block_group; 2500 }; 2501 2502 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2503 { 2504 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2505 } 2506 2507 static inline bool ext4_is_quota_file(struct inode *inode) 2508 { 2509 return IS_NOQUOTA(inode) && 2510 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2511 } 2512 2513 /* 2514 * This structure is stuffed into the struct file's private_data field 2515 * for directories. It is where we put information so that we can do 2516 * readdir operations in hash tree order. 2517 */ 2518 struct dir_private_info { 2519 struct rb_root root; 2520 struct rb_node *curr_node; 2521 struct fname *extra_fname; 2522 loff_t last_pos; 2523 __u32 curr_hash; 2524 __u32 curr_minor_hash; 2525 __u32 next_hash; 2526 }; 2527 2528 /* calculate the first block number of the group */ 2529 static inline ext4_fsblk_t 2530 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2531 { 2532 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2533 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2534 } 2535 2536 /* 2537 * Special error return code only used by dx_probe() and its callers. 2538 */ 2539 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2540 2541 /* htree levels for ext4 */ 2542 #define EXT4_HTREE_LEVEL_COMPAT 2 2543 #define EXT4_HTREE_LEVEL 3 2544 2545 static inline int ext4_dir_htree_level(struct super_block *sb) 2546 { 2547 return ext4_has_feature_largedir(sb) ? 2548 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2549 } 2550 2551 /* 2552 * Timeout and state flag for lazy initialization inode thread. 2553 */ 2554 #define EXT4_DEF_LI_WAIT_MULT 10 2555 #define EXT4_DEF_LI_MAX_START_DELAY 5 2556 #define EXT4_LAZYINIT_QUIT 0x0001 2557 #define EXT4_LAZYINIT_RUNNING 0x0002 2558 2559 /* 2560 * Lazy inode table initialization info 2561 */ 2562 struct ext4_lazy_init { 2563 unsigned long li_state; 2564 struct list_head li_request_list; 2565 struct mutex li_list_mtx; 2566 }; 2567 2568 enum ext4_li_mode { 2569 EXT4_LI_MODE_PREFETCH_BBITMAP, 2570 EXT4_LI_MODE_ITABLE, 2571 }; 2572 2573 struct ext4_li_request { 2574 struct super_block *lr_super; 2575 enum ext4_li_mode lr_mode; 2576 ext4_group_t lr_first_not_zeroed; 2577 ext4_group_t lr_next_group; 2578 struct list_head lr_request; 2579 unsigned long lr_next_sched; 2580 unsigned long lr_timeout; 2581 }; 2582 2583 struct ext4_features { 2584 struct kobject f_kobj; 2585 struct completion f_kobj_unregister; 2586 }; 2587 2588 /* 2589 * This structure will be used for multiple mount protection. It will be 2590 * written into the block number saved in the s_mmp_block field in the 2591 * superblock. Programs that check MMP should assume that if 2592 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2593 * to use the filesystem, regardless of how old the timestamp is. 2594 */ 2595 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2596 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2597 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2598 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2599 2600 struct mmp_struct { 2601 __le32 mmp_magic; /* Magic number for MMP */ 2602 __le32 mmp_seq; /* Sequence no. updated periodically */ 2603 2604 /* 2605 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2606 * purposes and do not affect the correctness of the algorithm 2607 */ 2608 __le64 mmp_time; /* Time last updated */ 2609 char mmp_nodename[64]; /* Node which last updated MMP block */ 2610 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2611 2612 /* 2613 * mmp_check_interval is used to verify if the MMP block has been 2614 * updated on the block device. The value is updated based on the 2615 * maximum time to write the MMP block during an update cycle. 2616 */ 2617 __le16 mmp_check_interval; 2618 2619 __le16 mmp_pad1; 2620 __le32 mmp_pad2[226]; 2621 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2622 }; 2623 2624 /* arguments passed to the mmp thread */ 2625 struct mmpd_data { 2626 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2627 struct super_block *sb; /* super block of the fs */ 2628 }; 2629 2630 /* 2631 * Check interval multiplier 2632 * The MMP block is written every update interval and initially checked every 2633 * update interval x the multiplier (the value is then adapted based on the 2634 * write latency). The reason is that writes can be delayed under load and we 2635 * don't want readers to incorrectly assume that the filesystem is no longer 2636 * in use. 2637 */ 2638 #define EXT4_MMP_CHECK_MULT 2UL 2639 2640 /* 2641 * Minimum interval for MMP checking in seconds. 2642 */ 2643 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2644 2645 /* 2646 * Maximum interval for MMP checking in seconds. 2647 */ 2648 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2649 2650 /* 2651 * Function prototypes 2652 */ 2653 2654 /* 2655 * Ok, these declarations are also in <linux/kernel.h> but none of the 2656 * ext4 source programs needs to include it so they are duplicated here. 2657 */ 2658 # define NORET_TYPE /**/ 2659 # define ATTRIB_NORET __attribute__((noreturn)) 2660 # define NORET_AND noreturn, 2661 2662 /* bitmap.c */ 2663 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2664 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2665 struct ext4_group_desc *gdp, 2666 struct buffer_head *bh, int sz); 2667 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2668 struct ext4_group_desc *gdp, 2669 struct buffer_head *bh, int sz); 2670 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2671 struct ext4_group_desc *gdp, 2672 struct buffer_head *bh); 2673 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2674 struct ext4_group_desc *gdp, 2675 struct buffer_head *bh); 2676 2677 /* balloc.c */ 2678 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2679 ext4_fsblk_t blocknr, 2680 ext4_group_t *blockgrpp, 2681 ext4_grpblk_t *offsetp); 2682 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2683 ext4_fsblk_t block); 2684 2685 extern unsigned int ext4_block_group(struct super_block *sb, 2686 ext4_fsblk_t blocknr); 2687 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2688 ext4_fsblk_t blocknr); 2689 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2690 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2691 ext4_group_t group); 2692 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2693 ext4_fsblk_t goal, 2694 unsigned int flags, 2695 unsigned long *count, 2696 int *errp); 2697 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2698 s64 nclusters, unsigned int flags); 2699 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2700 extern void ext4_check_blocks_bitmap(struct super_block *); 2701 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2702 ext4_group_t block_group, 2703 struct buffer_head ** bh); 2704 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2705 2706 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2707 ext4_group_t block_group, 2708 bool ignore_locked); 2709 extern int ext4_wait_block_bitmap(struct super_block *sb, 2710 ext4_group_t block_group, 2711 struct buffer_head *bh); 2712 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2713 ext4_group_t block_group); 2714 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2715 ext4_group_t block_group, 2716 struct ext4_group_desc *gdp); 2717 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2718 2719 #ifdef CONFIG_UNICODE 2720 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2721 const struct qstr *iname, 2722 struct ext4_filename *fname); 2723 #endif 2724 2725 #ifdef CONFIG_FS_ENCRYPTION 2726 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2727 const struct fscrypt_name *src) 2728 { 2729 memset(dst, 0, sizeof(*dst)); 2730 2731 dst->usr_fname = src->usr_fname; 2732 dst->disk_name = src->disk_name; 2733 dst->hinfo.hash = src->hash; 2734 dst->hinfo.minor_hash = src->minor_hash; 2735 dst->crypto_buf = src->crypto_buf; 2736 } 2737 2738 static inline int ext4_fname_setup_filename(struct inode *dir, 2739 const struct qstr *iname, 2740 int lookup, 2741 struct ext4_filename *fname) 2742 { 2743 struct fscrypt_name name; 2744 int err; 2745 2746 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2747 if (err) 2748 return err; 2749 2750 ext4_fname_from_fscrypt_name(fname, &name); 2751 2752 #ifdef CONFIG_UNICODE 2753 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2754 #endif 2755 return err; 2756 } 2757 2758 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2759 struct dentry *dentry, 2760 struct ext4_filename *fname) 2761 { 2762 struct fscrypt_name name; 2763 int err; 2764 2765 err = fscrypt_prepare_lookup(dir, dentry, &name); 2766 if (err) 2767 return err; 2768 2769 ext4_fname_from_fscrypt_name(fname, &name); 2770 2771 #ifdef CONFIG_UNICODE 2772 err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname); 2773 #endif 2774 return err; 2775 } 2776 2777 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2778 { 2779 struct fscrypt_name name; 2780 2781 name.crypto_buf = fname->crypto_buf; 2782 fscrypt_free_filename(&name); 2783 2784 fname->crypto_buf.name = NULL; 2785 fname->usr_fname = NULL; 2786 fname->disk_name.name = NULL; 2787 2788 #ifdef CONFIG_UNICODE 2789 kfree(fname->cf_name.name); 2790 fname->cf_name.name = NULL; 2791 #endif 2792 } 2793 #else /* !CONFIG_FS_ENCRYPTION */ 2794 static inline int ext4_fname_setup_filename(struct inode *dir, 2795 const struct qstr *iname, 2796 int lookup, 2797 struct ext4_filename *fname) 2798 { 2799 int err = 0; 2800 fname->usr_fname = iname; 2801 fname->disk_name.name = (unsigned char *) iname->name; 2802 fname->disk_name.len = iname->len; 2803 2804 #ifdef CONFIG_UNICODE 2805 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2806 #endif 2807 2808 return err; 2809 } 2810 2811 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2812 struct dentry *dentry, 2813 struct ext4_filename *fname) 2814 { 2815 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2816 } 2817 2818 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2819 { 2820 #ifdef CONFIG_UNICODE 2821 kfree(fname->cf_name.name); 2822 fname->cf_name.name = NULL; 2823 #endif 2824 } 2825 #endif /* !CONFIG_FS_ENCRYPTION */ 2826 2827 /* dir.c */ 2828 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2829 struct file *, 2830 struct ext4_dir_entry_2 *, 2831 struct buffer_head *, char *, int, 2832 unsigned int); 2833 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2834 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2835 (de), (bh), (buf), (size), (offset))) 2836 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2837 __u32 minor_hash, 2838 struct ext4_dir_entry_2 *dirent, 2839 struct fscrypt_str *ent_name); 2840 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2841 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2842 struct buffer_head *bh, 2843 void *buf, int buf_size, 2844 struct ext4_filename *fname, 2845 struct ext4_dir_entry_2 **dest_de); 2846 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2847 struct ext4_dir_entry_2 *de, 2848 int buf_size, 2849 struct ext4_filename *fname); 2850 static inline void ext4_update_dx_flag(struct inode *inode) 2851 { 2852 if (!ext4_has_feature_dir_index(inode->i_sb) && 2853 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2854 /* ext4_iget() should have caught this... */ 2855 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2856 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2857 } 2858 } 2859 static const unsigned char ext4_filetype_table[] = { 2860 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2861 }; 2862 2863 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2864 { 2865 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2866 return DT_UNKNOWN; 2867 2868 return ext4_filetype_table[filetype]; 2869 } 2870 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2871 void *buf, int buf_size); 2872 2873 /* fsync.c */ 2874 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2875 2876 /* hash.c */ 2877 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2878 struct dx_hash_info *hinfo); 2879 2880 /* ialloc.c */ 2881 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2882 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *, 2883 struct inode *, umode_t, 2884 const struct qstr *qstr, __u32 goal, 2885 uid_t *owner, __u32 i_flags, 2886 int handle_type, unsigned int line_no, 2887 int nblocks); 2888 2889 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2890 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \ 2891 (goal), (owner), i_flags, 0, 0, 0) 2892 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \ 2893 type, nblocks) \ 2894 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2895 0, (type), __LINE__, (nblocks)) 2896 2897 2898 extern void ext4_free_inode(handle_t *, struct inode *); 2899 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2900 extern unsigned long ext4_count_free_inodes(struct super_block *); 2901 extern unsigned long ext4_count_dirs(struct super_block *); 2902 extern void ext4_check_inodes_bitmap(struct super_block *); 2903 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2904 extern int ext4_init_inode_table(struct super_block *sb, 2905 ext4_group_t group, int barrier); 2906 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2907 2908 /* fast_commit.c */ 2909 int ext4_fc_info_show(struct seq_file *seq, void *v); 2910 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2911 void ext4_fc_init_inode(struct inode *inode); 2912 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2913 ext4_lblk_t end); 2914 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2915 struct dentry *dentry); 2916 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2917 struct dentry *dentry); 2918 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2919 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2920 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2921 struct dentry *dentry); 2922 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2923 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2924 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2925 void ext4_fc_start_update(struct inode *inode); 2926 void ext4_fc_stop_update(struct inode *inode); 2927 void ext4_fc_del(struct inode *inode); 2928 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2929 void ext4_fc_replay_cleanup(struct super_block *sb); 2930 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2931 int __init ext4_fc_init_dentry_cache(void); 2932 void ext4_fc_destroy_dentry_cache(void); 2933 int ext4_fc_record_regions(struct super_block *sb, int ino, 2934 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2935 int len, int replay); 2936 2937 /* mballoc.c */ 2938 extern const struct seq_operations ext4_mb_seq_groups_ops; 2939 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2940 extern long ext4_mb_stats; 2941 extern long ext4_mb_max_to_scan; 2942 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2943 extern int ext4_mb_init(struct super_block *); 2944 extern int ext4_mb_release(struct super_block *); 2945 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2946 struct ext4_allocation_request *, int *); 2947 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2948 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2949 extern int __init ext4_init_mballoc(void); 2950 extern void ext4_exit_mballoc(void); 2951 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2952 ext4_group_t group, 2953 unsigned int nr, int *cnt); 2954 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2955 unsigned int nr); 2956 2957 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2958 struct buffer_head *bh, ext4_fsblk_t block, 2959 unsigned long count, int flags); 2960 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2961 ext4_group_t ngroups); 2962 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2963 ext4_group_t i, struct ext4_group_desc *desc); 2964 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2965 ext4_fsblk_t block, unsigned long count); 2966 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2967 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2968 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2969 int len, int state); 2970 2971 /* inode.c */ 2972 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2973 struct ext4_inode_info *ei); 2974 int ext4_inode_is_fast_symlink(struct inode *inode); 2975 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2976 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2977 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2978 bool wait, struct buffer_head **bhs); 2979 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2980 struct buffer_head *bh_result, int create); 2981 int ext4_get_block(struct inode *inode, sector_t iblock, 2982 struct buffer_head *bh_result, int create); 2983 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2984 struct buffer_head *bh, int create); 2985 int ext4_walk_page_buffers(handle_t *handle, 2986 struct inode *inode, 2987 struct buffer_head *head, 2988 unsigned from, 2989 unsigned to, 2990 int *partial, 2991 int (*fn)(handle_t *handle, struct inode *inode, 2992 struct buffer_head *bh)); 2993 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2994 struct buffer_head *bh); 2995 #define FALL_BACK_TO_NONDELALLOC 1 2996 #define CONVERT_INLINE_DATA 2 2997 2998 typedef enum { 2999 EXT4_IGET_NORMAL = 0, 3000 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3001 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 3002 } ext4_iget_flags; 3003 3004 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3005 ext4_iget_flags flags, const char *function, 3006 unsigned int line); 3007 3008 #define ext4_iget(sb, ino, flags) \ 3009 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3010 3011 extern int ext4_write_inode(struct inode *, struct writeback_control *); 3012 extern int ext4_setattr(struct user_namespace *, struct dentry *, 3013 struct iattr *); 3014 extern int ext4_getattr(struct user_namespace *, const struct path *, 3015 struct kstat *, u32, unsigned int); 3016 extern void ext4_evict_inode(struct inode *); 3017 extern void ext4_clear_inode(struct inode *); 3018 extern int ext4_file_getattr(struct user_namespace *, const struct path *, 3019 struct kstat *, u32, unsigned int); 3020 extern int ext4_sync_inode(handle_t *, struct inode *); 3021 extern void ext4_dirty_inode(struct inode *, int); 3022 extern int ext4_change_inode_journal_flag(struct inode *, int); 3023 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3024 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3025 struct ext4_iloc *iloc); 3026 extern int ext4_inode_attach_jinode(struct inode *inode); 3027 extern int ext4_can_truncate(struct inode *inode); 3028 extern int ext4_truncate(struct inode *); 3029 extern int ext4_break_layouts(struct inode *); 3030 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 3031 extern void ext4_set_inode_flags(struct inode *, bool init); 3032 extern int ext4_alloc_da_blocks(struct inode *inode); 3033 extern void ext4_set_aops(struct inode *inode); 3034 extern int ext4_writepage_trans_blocks(struct inode *); 3035 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3036 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3037 loff_t lstart, loff_t lend); 3038 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3039 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3040 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3041 extern void ext4_da_release_space(struct inode *inode, int to_free); 3042 extern void ext4_da_update_reserve_space(struct inode *inode, 3043 int used, int quota_claim); 3044 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3045 ext4_fsblk_t pblk, ext4_lblk_t len); 3046 3047 /* indirect.c */ 3048 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3049 struct ext4_map_blocks *map, int flags); 3050 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3051 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3052 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3053 ext4_lblk_t start, ext4_lblk_t end); 3054 3055 /* ioctl.c */ 3056 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3057 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3058 int ext4_fileattr_set(struct user_namespace *mnt_userns, 3059 struct dentry *dentry, struct fileattr *fa); 3060 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3061 extern void ext4_reset_inode_seed(struct inode *inode); 3062 3063 /* migrate.c */ 3064 extern int ext4_ext_migrate(struct inode *); 3065 extern int ext4_ind_migrate(struct inode *inode); 3066 3067 /* namei.c */ 3068 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3069 struct inode *inode); 3070 extern int ext4_dirblock_csum_verify(struct inode *inode, 3071 struct buffer_head *bh); 3072 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3073 __u32 start_minor_hash, __u32 *next_hash); 3074 extern int ext4_search_dir(struct buffer_head *bh, 3075 char *search_buf, 3076 int buf_size, 3077 struct inode *dir, 3078 struct ext4_filename *fname, 3079 unsigned int offset, 3080 struct ext4_dir_entry_2 **res_dir); 3081 extern int ext4_generic_delete_entry(struct inode *dir, 3082 struct ext4_dir_entry_2 *de_del, 3083 struct buffer_head *bh, 3084 void *entry_buf, 3085 int buf_size, 3086 int csum_size); 3087 extern bool ext4_empty_dir(struct inode *inode); 3088 3089 /* resize.c */ 3090 extern void ext4_kvfree_array_rcu(void *to_free); 3091 extern int ext4_group_add(struct super_block *sb, 3092 struct ext4_new_group_data *input); 3093 extern int ext4_group_extend(struct super_block *sb, 3094 struct ext4_super_block *es, 3095 ext4_fsblk_t n_blocks_count); 3096 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3097 3098 /* super.c */ 3099 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3100 sector_t block, int op_flags); 3101 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3102 sector_t block); 3103 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags, 3104 bh_end_io_t *end_io); 3105 extern int ext4_read_bh(struct buffer_head *bh, int op_flags, 3106 bh_end_io_t *end_io); 3107 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait); 3108 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3109 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3110 extern int ext4_calculate_overhead(struct super_block *sb); 3111 extern void ext4_superblock_csum_set(struct super_block *sb); 3112 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3113 ext4_group_t ngroup); 3114 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3115 char nbuf[16]); 3116 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3117 ext4_group_t block_group, 3118 unsigned int flags); 3119 3120 extern __printf(7, 8) 3121 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3122 int, __u64, const char *, ...); 3123 extern __printf(6, 7) 3124 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3125 ext4_fsblk_t, int, const char *, ...); 3126 extern __printf(5, 6) 3127 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3128 const char *, ...); 3129 extern void __ext4_std_error(struct super_block *, const char *, 3130 unsigned int, int); 3131 extern __printf(4, 5) 3132 void __ext4_warning(struct super_block *, const char *, unsigned int, 3133 const char *, ...); 3134 extern __printf(4, 5) 3135 void __ext4_warning_inode(const struct inode *inode, const char *function, 3136 unsigned int line, const char *fmt, ...); 3137 extern __printf(3, 4) 3138 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3139 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3140 const char *, unsigned int, const char *); 3141 extern __printf(7, 8) 3142 void __ext4_grp_locked_error(const char *, unsigned int, 3143 struct super_block *, ext4_group_t, 3144 unsigned long, ext4_fsblk_t, 3145 const char *, ...); 3146 3147 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3148 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3149 3150 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3151 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3152 3153 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3154 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3155 (fmt), ## a) 3156 3157 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3158 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3159 3160 #define ext4_abort(sb, err, fmt, a...) \ 3161 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3162 3163 #ifdef CONFIG_PRINTK 3164 3165 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3166 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3167 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3168 __ext4_error_inode((inode), (func), (line), (block), \ 3169 (err), (fmt), ##__VA_ARGS__) 3170 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3171 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3172 #define ext4_error(sb, fmt, ...) \ 3173 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3174 ##__VA_ARGS__) 3175 #define ext4_error_err(sb, err, fmt, ...) \ 3176 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3177 ##__VA_ARGS__) 3178 #define ext4_warning(sb, fmt, ...) \ 3179 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3180 #define ext4_warning_inode(inode, fmt, ...) \ 3181 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3182 #define ext4_msg(sb, level, fmt, ...) \ 3183 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3184 #define dump_mmp_msg(sb, mmp, msg) \ 3185 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3186 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3187 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3188 fmt, ##__VA_ARGS__) 3189 3190 #else 3191 3192 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3193 do { \ 3194 no_printk(fmt, ##__VA_ARGS__); \ 3195 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3196 } while (0) 3197 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3198 do { \ 3199 no_printk(fmt, ##__VA_ARGS__); \ 3200 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3201 } while (0) 3202 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3203 do { \ 3204 no_printk(fmt, ##__VA_ARGS__); \ 3205 __ext4_error_file(file, "", 0, block, " "); \ 3206 } while (0) 3207 #define ext4_error(sb, fmt, ...) \ 3208 do { \ 3209 no_printk(fmt, ##__VA_ARGS__); \ 3210 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3211 } while (0) 3212 #define ext4_error_err(sb, err, fmt, ...) \ 3213 do { \ 3214 no_printk(fmt, ##__VA_ARGS__); \ 3215 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3216 } while (0) 3217 #define ext4_warning(sb, fmt, ...) \ 3218 do { \ 3219 no_printk(fmt, ##__VA_ARGS__); \ 3220 __ext4_warning(sb, "", 0, " "); \ 3221 } while (0) 3222 #define ext4_warning_inode(inode, fmt, ...) \ 3223 do { \ 3224 no_printk(fmt, ##__VA_ARGS__); \ 3225 __ext4_warning_inode(inode, "", 0, " "); \ 3226 } while (0) 3227 #define ext4_msg(sb, level, fmt, ...) \ 3228 do { \ 3229 no_printk(fmt, ##__VA_ARGS__); \ 3230 __ext4_msg(sb, "", " "); \ 3231 } while (0) 3232 #define dump_mmp_msg(sb, mmp, msg) \ 3233 __dump_mmp_msg(sb, mmp, "", 0, "") 3234 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3235 do { \ 3236 no_printk(fmt, ##__VA_ARGS__); \ 3237 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3238 } while (0) 3239 3240 #endif 3241 3242 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3243 struct ext4_group_desc *bg); 3244 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3245 struct ext4_group_desc *bg); 3246 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3247 struct ext4_group_desc *bg); 3248 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3249 struct ext4_group_desc *bg); 3250 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3251 struct ext4_group_desc *bg); 3252 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3253 struct ext4_group_desc *bg); 3254 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3255 struct ext4_group_desc *bg); 3256 extern void ext4_block_bitmap_set(struct super_block *sb, 3257 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3258 extern void ext4_inode_bitmap_set(struct super_block *sb, 3259 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3260 extern void ext4_inode_table_set(struct super_block *sb, 3261 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3262 extern void ext4_free_group_clusters_set(struct super_block *sb, 3263 struct ext4_group_desc *bg, 3264 __u32 count); 3265 extern void ext4_free_inodes_set(struct super_block *sb, 3266 struct ext4_group_desc *bg, __u32 count); 3267 extern void ext4_used_dirs_set(struct super_block *sb, 3268 struct ext4_group_desc *bg, __u32 count); 3269 extern void ext4_itable_unused_set(struct super_block *sb, 3270 struct ext4_group_desc *bg, __u32 count); 3271 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3272 struct ext4_group_desc *gdp); 3273 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3274 struct ext4_group_desc *gdp); 3275 extern int ext4_register_li_request(struct super_block *sb, 3276 ext4_group_t first_not_zeroed); 3277 3278 static inline int ext4_has_metadata_csum(struct super_block *sb) 3279 { 3280 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3281 !EXT4_SB(sb)->s_chksum_driver); 3282 3283 return ext4_has_feature_metadata_csum(sb) && 3284 (EXT4_SB(sb)->s_chksum_driver != NULL); 3285 } 3286 3287 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3288 { 3289 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3290 } 3291 3292 #define ext4_read_incompat_64bit_val(es, name) \ 3293 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3294 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3295 le32_to_cpu(es->name##_lo)) 3296 3297 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3298 { 3299 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3300 } 3301 3302 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3303 { 3304 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3305 } 3306 3307 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3308 { 3309 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3310 } 3311 3312 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3313 ext4_fsblk_t blk) 3314 { 3315 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3316 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3317 } 3318 3319 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3320 ext4_fsblk_t blk) 3321 { 3322 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3323 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3324 } 3325 3326 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3327 ext4_fsblk_t blk) 3328 { 3329 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3330 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3331 } 3332 3333 static inline loff_t ext4_isize(struct super_block *sb, 3334 struct ext4_inode *raw_inode) 3335 { 3336 if (ext4_has_feature_largedir(sb) || 3337 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3338 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3339 le32_to_cpu(raw_inode->i_size_lo); 3340 3341 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3342 } 3343 3344 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3345 { 3346 raw_inode->i_size_lo = cpu_to_le32(i_size); 3347 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3348 } 3349 3350 static inline 3351 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3352 ext4_group_t group) 3353 { 3354 struct ext4_group_info **grp_info; 3355 long indexv, indexh; 3356 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3357 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3358 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3359 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3360 return grp_info[indexh]; 3361 } 3362 3363 /* 3364 * Reading s_groups_count requires using smp_rmb() afterwards. See 3365 * the locking protocol documented in the comments of ext4_group_add() 3366 * in resize.c 3367 */ 3368 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3369 { 3370 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3371 3372 smp_rmb(); 3373 return ngroups; 3374 } 3375 3376 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3377 ext4_group_t block_group) 3378 { 3379 return block_group >> sbi->s_log_groups_per_flex; 3380 } 3381 3382 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3383 { 3384 return 1 << sbi->s_log_groups_per_flex; 3385 } 3386 3387 #define ext4_std_error(sb, errno) \ 3388 do { \ 3389 if ((errno)) \ 3390 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3391 } while (0) 3392 3393 #ifdef CONFIG_SMP 3394 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3395 * counters. So we need to make sure we have free clusters more 3396 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3397 */ 3398 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3399 #else 3400 #define EXT4_FREECLUSTERS_WATERMARK 0 3401 #endif 3402 3403 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 3404 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3405 { 3406 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3407 !inode_is_locked(inode)); 3408 down_write(&EXT4_I(inode)->i_data_sem); 3409 if (newsize > EXT4_I(inode)->i_disksize) 3410 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3411 up_write(&EXT4_I(inode)->i_data_sem); 3412 } 3413 3414 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 3415 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3416 { 3417 int changed = 0; 3418 3419 if (newsize > inode->i_size) { 3420 i_size_write(inode, newsize); 3421 changed = 1; 3422 } 3423 if (newsize > EXT4_I(inode)->i_disksize) { 3424 ext4_update_i_disksize(inode, newsize); 3425 changed |= 2; 3426 } 3427 return changed; 3428 } 3429 3430 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3431 loff_t len); 3432 3433 struct ext4_group_info { 3434 unsigned long bb_state; 3435 #ifdef AGGRESSIVE_CHECK 3436 unsigned long bb_check_counter; 3437 #endif 3438 struct rb_root bb_free_root; 3439 ext4_grpblk_t bb_first_free; /* first free block */ 3440 ext4_grpblk_t bb_free; /* total free blocks */ 3441 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3442 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3443 ext4_group_t bb_group; /* Group number */ 3444 struct list_head bb_prealloc_list; 3445 #ifdef DOUBLE_CHECK 3446 void *bb_bitmap; 3447 #endif 3448 struct rw_semaphore alloc_sem; 3449 struct rb_node bb_avg_fragment_size_rb; 3450 struct list_head bb_largest_free_order_node; 3451 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3452 * regions, index is order. 3453 * bb_counters[3] = 5 means 3454 * 5 free 8-block regions. */ 3455 }; 3456 3457 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3458 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3459 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3460 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3461 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3462 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3463 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3464 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3465 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3466 3467 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3468 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3469 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3470 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3471 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3472 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3473 3474 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3475 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3476 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3477 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3478 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3479 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3480 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3481 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3482 3483 #define EXT4_MAX_CONTENTION 8 3484 #define EXT4_CONTENTION_THRESHOLD 2 3485 3486 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3487 ext4_group_t group) 3488 { 3489 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3490 } 3491 3492 /* 3493 * Returns true if the filesystem is busy enough that attempts to 3494 * access the block group locks has run into contention. 3495 */ 3496 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3497 { 3498 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3499 } 3500 3501 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3502 { 3503 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3504 if (spin_trylock(lock)) 3505 /* 3506 * We're able to grab the lock right away, so drop the 3507 * lock contention counter. 3508 */ 3509 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3510 else { 3511 /* 3512 * The lock is busy, so bump the contention counter, 3513 * and then wait on the spin lock. 3514 */ 3515 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3516 EXT4_MAX_CONTENTION); 3517 spin_lock(lock); 3518 } 3519 } 3520 3521 static inline void ext4_unlock_group(struct super_block *sb, 3522 ext4_group_t group) 3523 { 3524 spin_unlock(ext4_group_lock_ptr(sb, group)); 3525 } 3526 3527 #ifdef CONFIG_QUOTA 3528 static inline bool ext4_quota_capable(struct super_block *sb) 3529 { 3530 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3531 } 3532 3533 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3534 { 3535 struct ext4_sb_info *sbi = EXT4_SB(sb); 3536 3537 return (ext4_has_feature_quota(sb) || 3538 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3539 } 3540 int ext4_enable_quotas(struct super_block *sb); 3541 #endif 3542 3543 /* 3544 * Block validity checking 3545 */ 3546 #define ext4_check_indirect_blockref(inode, bh) \ 3547 ext4_check_blockref(__func__, __LINE__, inode, \ 3548 (__le32 *)(bh)->b_data, \ 3549 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3550 3551 #define ext4_ind_check_inode(inode) \ 3552 ext4_check_blockref(__func__, __LINE__, inode, \ 3553 EXT4_I(inode)->i_data, \ 3554 EXT4_NDIR_BLOCKS) 3555 3556 /* 3557 * Inodes and files operations 3558 */ 3559 3560 /* dir.c */ 3561 extern const struct file_operations ext4_dir_operations; 3562 3563 /* file.c */ 3564 extern const struct inode_operations ext4_file_inode_operations; 3565 extern const struct file_operations ext4_file_operations; 3566 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3567 3568 /* inline.c */ 3569 extern int ext4_get_max_inline_size(struct inode *inode); 3570 extern int ext4_find_inline_data_nolock(struct inode *inode); 3571 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3572 unsigned int len); 3573 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3574 3575 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3576 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3577 struct inode *inode, 3578 loff_t pos, unsigned len, 3579 unsigned flags, 3580 struct page **pagep); 3581 extern int ext4_write_inline_data_end(struct inode *inode, 3582 loff_t pos, unsigned len, 3583 unsigned copied, 3584 struct page *page); 3585 extern struct buffer_head * 3586 ext4_journalled_write_inline_data(struct inode *inode, 3587 unsigned len, 3588 struct page *page); 3589 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3590 struct inode *inode, 3591 loff_t pos, unsigned len, 3592 unsigned flags, 3593 struct page **pagep, 3594 void **fsdata); 3595 extern int ext4_try_add_inline_entry(handle_t *handle, 3596 struct ext4_filename *fname, 3597 struct inode *dir, struct inode *inode); 3598 extern int ext4_try_create_inline_dir(handle_t *handle, 3599 struct inode *parent, 3600 struct inode *inode); 3601 extern int ext4_read_inline_dir(struct file *filp, 3602 struct dir_context *ctx, 3603 int *has_inline_data); 3604 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3605 struct inode *dir, ext4_lblk_t block, 3606 struct dx_hash_info *hinfo, 3607 __u32 start_hash, __u32 start_minor_hash, 3608 int *has_inline_data); 3609 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3610 struct ext4_filename *fname, 3611 struct ext4_dir_entry_2 **res_dir, 3612 int *has_inline_data); 3613 extern int ext4_delete_inline_entry(handle_t *handle, 3614 struct inode *dir, 3615 struct ext4_dir_entry_2 *de_del, 3616 struct buffer_head *bh, 3617 int *has_inline_data); 3618 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3619 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3620 struct ext4_dir_entry_2 **parent_de, 3621 int *retval); 3622 extern int ext4_inline_data_fiemap(struct inode *inode, 3623 struct fiemap_extent_info *fieinfo, 3624 int *has_inline, __u64 start, __u64 len); 3625 3626 struct iomap; 3627 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3628 3629 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3630 3631 extern int ext4_convert_inline_data(struct inode *inode); 3632 3633 static inline int ext4_has_inline_data(struct inode *inode) 3634 { 3635 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3636 EXT4_I(inode)->i_inline_off; 3637 } 3638 3639 /* namei.c */ 3640 extern const struct inode_operations ext4_dir_inode_operations; 3641 extern const struct inode_operations ext4_special_inode_operations; 3642 extern struct dentry *ext4_get_parent(struct dentry *child); 3643 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3644 struct ext4_dir_entry_2 *de, 3645 int blocksize, int csum_size, 3646 unsigned int parent_ino, int dotdot_real_len); 3647 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3648 unsigned int blocksize); 3649 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3650 struct buffer_head *bh); 3651 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name, 3652 struct inode *inode); 3653 extern int __ext4_link(struct inode *dir, struct inode *inode, 3654 struct dentry *dentry); 3655 3656 #define S_SHIFT 12 3657 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3658 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3659 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3660 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3661 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3662 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3663 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3664 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3665 }; 3666 3667 static inline void ext4_set_de_type(struct super_block *sb, 3668 struct ext4_dir_entry_2 *de, 3669 umode_t mode) { 3670 if (ext4_has_feature_filetype(sb)) 3671 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3672 } 3673 3674 /* readpages.c */ 3675 extern int ext4_mpage_readpages(struct inode *inode, 3676 struct readahead_control *rac, struct page *page); 3677 extern int __init ext4_init_post_read_processing(void); 3678 extern void ext4_exit_post_read_processing(void); 3679 3680 /* symlink.c */ 3681 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3682 extern const struct inode_operations ext4_symlink_inode_operations; 3683 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3684 3685 /* sysfs.c */ 3686 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3687 extern int ext4_register_sysfs(struct super_block *sb); 3688 extern void ext4_unregister_sysfs(struct super_block *sb); 3689 extern int __init ext4_init_sysfs(void); 3690 extern void ext4_exit_sysfs(void); 3691 3692 /* block_validity */ 3693 extern void ext4_release_system_zone(struct super_block *sb); 3694 extern int ext4_setup_system_zone(struct super_block *sb); 3695 extern int __init ext4_init_system_zone(void); 3696 extern void ext4_exit_system_zone(void); 3697 extern int ext4_inode_block_valid(struct inode *inode, 3698 ext4_fsblk_t start_blk, 3699 unsigned int count); 3700 extern int ext4_check_blockref(const char *, unsigned int, 3701 struct inode *, __le32 *, unsigned int); 3702 3703 /* extents.c */ 3704 struct ext4_ext_path; 3705 struct ext4_extent; 3706 3707 /* 3708 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3709 * __le32. 3710 */ 3711 #define EXT_MAX_BLOCKS 0xffffffff 3712 3713 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3714 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3715 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3716 struct ext4_map_blocks *map, int flags); 3717 extern int ext4_ext_truncate(handle_t *, struct inode *); 3718 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3719 ext4_lblk_t end); 3720 extern void ext4_ext_init(struct super_block *); 3721 extern void ext4_ext_release(struct super_block *); 3722 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3723 loff_t len); 3724 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3725 loff_t offset, ssize_t len); 3726 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3727 ext4_io_end_t *io_end); 3728 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3729 struct ext4_map_blocks *map, int flags); 3730 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3731 int num, 3732 struct ext4_ext_path *path); 3733 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3734 struct ext4_ext_path **, 3735 struct ext4_extent *, int); 3736 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3737 struct ext4_ext_path **, 3738 int flags); 3739 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3740 extern int ext4_ext_check_inode(struct inode *inode); 3741 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3742 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3743 __u64 start, __u64 len); 3744 extern int ext4_get_es_cache(struct inode *inode, 3745 struct fiemap_extent_info *fieinfo, 3746 __u64 start, __u64 len); 3747 extern int ext4_ext_precache(struct inode *inode); 3748 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3749 struct inode *inode2, ext4_lblk_t lblk1, 3750 ext4_lblk_t lblk2, ext4_lblk_t count, 3751 int mark_unwritten,int *err); 3752 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3753 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3754 int check_cred, int restart_cred, 3755 int revoke_cred); 3756 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3757 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3758 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3759 int len, int unwritten, ext4_fsblk_t pblk); 3760 extern int ext4_ext_clear_bb(struct inode *inode); 3761 3762 3763 /* move_extent.c */ 3764 extern void ext4_double_down_write_data_sem(struct inode *first, 3765 struct inode *second); 3766 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3767 struct inode *donor_inode); 3768 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3769 __u64 start_orig, __u64 start_donor, 3770 __u64 len, __u64 *moved_len); 3771 3772 /* page-io.c */ 3773 extern int __init ext4_init_pageio(void); 3774 extern void ext4_exit_pageio(void); 3775 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3776 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3777 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3778 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3779 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3780 struct writeback_control *wbc); 3781 extern void ext4_end_io_rsv_work(struct work_struct *work); 3782 extern void ext4_io_submit(struct ext4_io_submit *io); 3783 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3784 struct page *page, 3785 int len, 3786 bool keep_towrite); 3787 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3788 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3789 3790 /* mmp.c */ 3791 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3792 3793 /* mmp.c */ 3794 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3795 3796 /* verity.c */ 3797 extern const struct fsverity_operations ext4_verityops; 3798 3799 /* orphan.c */ 3800 extern int ext4_orphan_add(handle_t *, struct inode *); 3801 extern int ext4_orphan_del(handle_t *, struct inode *); 3802 extern void ext4_orphan_cleanup(struct super_block *sb, 3803 struct ext4_super_block *es); 3804 extern void ext4_release_orphan_info(struct super_block *sb); 3805 extern int ext4_init_orphan_info(struct super_block *sb); 3806 extern int ext4_orphan_file_empty(struct super_block *sb); 3807 extern void ext4_orphan_file_block_trigger( 3808 struct jbd2_buffer_trigger_type *triggers, 3809 struct buffer_head *bh, 3810 void *data, size_t size); 3811 3812 /* 3813 * Add new method to test whether block and inode bitmaps are properly 3814 * initialized. With uninit_bg reading the block from disk is not enough 3815 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3816 */ 3817 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3818 3819 static inline int bitmap_uptodate(struct buffer_head *bh) 3820 { 3821 return (buffer_uptodate(bh) && 3822 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3823 } 3824 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3825 { 3826 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3827 } 3828 3829 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3830 3831 /* For ioend & aio unwritten conversion wait queues */ 3832 #define EXT4_WQ_HASH_SZ 37 3833 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3834 EXT4_WQ_HASH_SZ]) 3835 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3836 3837 extern int ext4_resize_begin(struct super_block *sb); 3838 extern void ext4_resize_end(struct super_block *sb); 3839 3840 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3841 struct ext4_io_end *io_end) 3842 { 3843 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3844 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3845 atomic_inc(&EXT4_I(inode)->i_unwritten); 3846 } 3847 } 3848 3849 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3850 { 3851 struct inode *inode = io_end->inode; 3852 3853 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3854 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3855 /* Wake up anyone waiting on unwritten extent conversion */ 3856 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3857 wake_up_all(ext4_ioend_wq(inode)); 3858 } 3859 } 3860 3861 extern const struct iomap_ops ext4_iomap_ops; 3862 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3863 extern const struct iomap_ops ext4_iomap_report_ops; 3864 3865 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3866 { 3867 /* 3868 * If the buffer has the write error flag, we have failed 3869 * to write out data in the block. In this case, we don't 3870 * have to read the block because we may read the old data 3871 * successfully. 3872 */ 3873 if (buffer_write_io_error(bh)) 3874 set_buffer_uptodate(bh); 3875 return buffer_uptodate(bh); 3876 } 3877 3878 #endif /* __KERNEL__ */ 3879 3880 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3881 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3882 3883 #endif /* _EXT4_H */ 3884