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/refcount.h> 21 #include <linux/types.h> 22 #include <linux/blkdev.h> 23 #include <linux/magic.h> 24 #include <linux/jbd2.h> 25 #include <linux/quota.h> 26 #include <linux/rwsem.h> 27 #include <linux/rbtree.h> 28 #include <linux/seqlock.h> 29 #include <linux/mutex.h> 30 #include <linux/timer.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 44 #include <linux/fscrypt.h> 45 #include <linux/fsverity.h> 46 47 #include <linux/compiler.h> 48 49 /* 50 * The fourth extended filesystem constants/structures 51 */ 52 53 /* 54 * with AGGRESSIVE_CHECK allocator runs consistency checks over 55 * structures. these checks slow things down a lot 56 */ 57 #define AGGRESSIVE_CHECK__ 58 59 /* 60 * with DOUBLE_CHECK defined mballoc creates persistent in-core 61 * bitmaps, maintains and uses them to check for double allocations 62 */ 63 #define DOUBLE_CHECK__ 64 65 /* 66 * Define EXT4FS_DEBUG to produce debug messages 67 */ 68 #undef EXT4FS_DEBUG 69 70 /* 71 * Debug code 72 */ 73 #ifdef EXT4FS_DEBUG 74 #define ext4_debug(f, a...) \ 75 do { \ 76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 77 __FILE__, __LINE__, __func__); \ 78 printk(KERN_DEBUG f, ## a); \ 79 } while (0) 80 #else 81 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 82 #endif 83 84 /* 85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 86 */ 87 #define EXT_DEBUG__ 88 89 /* 90 * Dynamic printk for controlled extents debugging. 91 */ 92 #ifdef CONFIG_EXT4_DEBUG 93 #define ext_debug(ino, fmt, ...) \ 94 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 95 current->comm, task_pid_nr(current), \ 96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 97 __func__, ##__VA_ARGS__) 98 #else 99 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 100 #endif 101 102 #define ASSERT(assert) \ 103 do { \ 104 if (unlikely(!(assert))) { \ 105 printk(KERN_EMERG \ 106 "Assertion failure in %s() at %s:%d: '%s'\n", \ 107 __func__, __FILE__, __LINE__, #assert); \ 108 BUG(); \ 109 } \ 110 } while (0) 111 112 /* data type for block offset of block group */ 113 typedef int ext4_grpblk_t; 114 115 /* data type for filesystem-wide blocks number */ 116 typedef unsigned long long ext4_fsblk_t; 117 118 /* data type for file logical block number */ 119 typedef __u32 ext4_lblk_t; 120 121 /* data type for block group number */ 122 typedef unsigned int ext4_group_t; 123 124 enum SHIFT_DIRECTION { 125 SHIFT_LEFT = 0, 126 SHIFT_RIGHT, 127 }; 128 129 /* 130 * Flags used in mballoc's allocation_context flags field. 131 * 132 * Also used to show what's going on for debugging purposes when the 133 * flag field is exported via the traceport interface 134 */ 135 136 /* prefer goal again. length */ 137 #define EXT4_MB_HINT_MERGE 0x0001 138 /* blocks already reserved */ 139 #define EXT4_MB_HINT_RESERVED 0x0002 140 /* metadata is being allocated */ 141 #define EXT4_MB_HINT_METADATA 0x0004 142 /* first blocks in the file */ 143 #define EXT4_MB_HINT_FIRST 0x0008 144 /* search for the best chunk */ 145 #define EXT4_MB_HINT_BEST 0x0010 146 /* data is being allocated */ 147 #define EXT4_MB_HINT_DATA 0x0020 148 /* don't preallocate (for tails) */ 149 #define EXT4_MB_HINT_NOPREALLOC 0x0040 150 /* allocate for locality group */ 151 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 152 /* allocate goal blocks or none */ 153 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 154 /* goal is meaningful */ 155 #define EXT4_MB_HINT_TRY_GOAL 0x0200 156 /* blocks already pre-reserved by delayed allocation */ 157 #define EXT4_MB_DELALLOC_RESERVED 0x0400 158 /* We are doing stream allocation */ 159 #define EXT4_MB_STREAM_ALLOC 0x0800 160 /* Use reserved root blocks if needed */ 161 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 162 /* Use blocks from reserved pool */ 163 #define EXT4_MB_USE_RESERVED 0x2000 164 /* Do strict check for free blocks while retrying block allocation */ 165 #define EXT4_MB_STRICT_CHECK 0x4000 166 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 167 #define EXT4_MB_CR0_OPTIMIZED 0x8000 168 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 169 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 170 struct ext4_allocation_request { 171 /* target inode for block we're allocating */ 172 struct inode *inode; 173 /* how many blocks we want to allocate */ 174 unsigned int len; 175 /* logical block in target inode */ 176 ext4_lblk_t logical; 177 /* the closest logical allocated block to the left */ 178 ext4_lblk_t lleft; 179 /* the closest logical allocated block to the right */ 180 ext4_lblk_t lright; 181 /* phys. target (a hint) */ 182 ext4_fsblk_t goal; 183 /* phys. block for the closest logical allocated block to the left */ 184 ext4_fsblk_t pleft; 185 /* phys. block for the closest logical allocated block to the right */ 186 ext4_fsblk_t pright; 187 /* flags. see above EXT4_MB_HINT_* */ 188 unsigned int flags; 189 }; 190 191 /* 192 * Logical to physical block mapping, used by ext4_map_blocks() 193 * 194 * This structure is used to pass requests into ext4_map_blocks() as 195 * well as to store the information returned by ext4_map_blocks(). It 196 * takes less room on the stack than a struct buffer_head. 197 */ 198 #define EXT4_MAP_NEW BIT(BH_New) 199 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 200 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 201 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 202 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 203 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 204 205 struct ext4_map_blocks { 206 ext4_fsblk_t m_pblk; 207 ext4_lblk_t m_lblk; 208 unsigned int m_len; 209 unsigned int m_flags; 210 }; 211 212 /* 213 * Block validity checking, system zone rbtree. 214 */ 215 struct ext4_system_blocks { 216 struct rb_root root; 217 struct rcu_head rcu; 218 }; 219 220 /* 221 * Flags for ext4_io_end->flags 222 */ 223 #define EXT4_IO_END_UNWRITTEN 0x0001 224 225 struct ext4_io_end_vec { 226 struct list_head list; /* list of io_end_vec */ 227 loff_t offset; /* offset in the file */ 228 ssize_t size; /* size of the extent */ 229 }; 230 231 /* 232 * For converting unwritten extents on a work queue. 'handle' is used for 233 * buffered writeback. 234 */ 235 typedef struct ext4_io_end { 236 struct list_head list; /* per-file finished IO list */ 237 handle_t *handle; /* handle reserved for extent 238 * conversion */ 239 struct inode *inode; /* file being written to */ 240 struct bio *bio; /* Linked list of completed 241 * bios covering the extent */ 242 unsigned int flag; /* unwritten or not */ 243 refcount_t count; /* reference counter */ 244 struct list_head list_vec; /* list of ext4_io_end_vec */ 245 } ext4_io_end_t; 246 247 struct ext4_io_submit { 248 struct writeback_control *io_wbc; 249 struct bio *io_bio; 250 ext4_io_end_t *io_end; 251 sector_t io_next_block; 252 }; 253 254 /* 255 * Special inodes numbers 256 */ 257 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 258 #define EXT4_ROOT_INO 2 /* Root inode */ 259 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 260 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 261 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 262 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 263 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 264 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 265 266 /* First non-reserved inode for old ext4 filesystems */ 267 #define EXT4_GOOD_OLD_FIRST_INO 11 268 269 /* 270 * Maximal count of links to a file 271 */ 272 #define EXT4_LINK_MAX 65000 273 274 /* 275 * Macro-instructions used to manage several block sizes 276 */ 277 #define EXT4_MIN_BLOCK_SIZE 1024 278 #define EXT4_MAX_BLOCK_SIZE 65536 279 #define EXT4_MIN_BLOCK_LOG_SIZE 10 280 #define EXT4_MAX_BLOCK_LOG_SIZE 16 281 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 282 #ifdef __KERNEL__ 283 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 284 #else 285 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 286 #endif 287 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 288 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 289 EXT4_SB(s)->s_cluster_bits) 290 #ifdef __KERNEL__ 291 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 292 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 293 #else 294 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 295 #endif 296 #ifdef __KERNEL__ 297 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 298 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 299 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 300 #else 301 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 302 EXT4_GOOD_OLD_INODE_SIZE : \ 303 (s)->s_inode_size) 304 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 305 EXT4_GOOD_OLD_FIRST_INO : \ 306 (s)->s_first_ino) 307 #endif 308 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 309 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 310 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 311 blkbits)) 312 313 /* Translate a block number to a cluster number */ 314 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 315 /* Translate a cluster number to a block number */ 316 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 317 /* Translate # of blks to # of clusters */ 318 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 319 (sbi)->s_cluster_bits) 320 /* Mask out the low bits to get the starting block of the cluster */ 321 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 322 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 323 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 324 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 325 /* Fill in the low bits to get the last block of the cluster */ 326 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 327 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 328 /* Get the cluster offset */ 329 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 330 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 331 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 332 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 333 334 /* 335 * Structure of a blocks group descriptor 336 */ 337 struct ext4_group_desc 338 { 339 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 340 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 341 __le32 bg_inode_table_lo; /* Inodes table block */ 342 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 343 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 344 __le16 bg_used_dirs_count_lo; /* Directories count */ 345 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 346 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 347 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 348 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 349 __le16 bg_itable_unused_lo; /* Unused inodes count */ 350 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 351 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 352 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 353 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 354 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 355 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 356 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 357 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 358 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 359 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 360 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 361 __u32 bg_reserved; 362 }; 363 364 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 365 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 366 sizeof(__le16)) 367 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 368 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 369 sizeof(__le16)) 370 371 /* 372 * Structure of a flex block group info 373 */ 374 375 struct flex_groups { 376 atomic64_t free_clusters; 377 atomic_t free_inodes; 378 atomic_t used_dirs; 379 }; 380 381 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 382 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 383 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 384 385 /* 386 * Macro-instructions used to manage group descriptors 387 */ 388 #define EXT4_MIN_DESC_SIZE 32 389 #define EXT4_MIN_DESC_SIZE_64BIT 64 390 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 391 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 392 #ifdef __KERNEL__ 393 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 394 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 395 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 396 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 397 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 398 #else 399 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 400 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 401 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 402 #endif 403 404 /* 405 * Constants relative to the data blocks 406 */ 407 #define EXT4_NDIR_BLOCKS 12 408 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 409 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 410 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 411 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 412 413 /* 414 * Inode flags 415 */ 416 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 417 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 418 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 419 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 420 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 421 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 422 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 423 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 424 /* Reserved for compression usage... */ 425 #define EXT4_DIRTY_FL 0x00000100 426 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 427 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 428 /* nb: was previously EXT2_ECOMPR_FL */ 429 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 430 /* End compression flags --- maybe not all used */ 431 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 432 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 433 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 434 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 435 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 436 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 437 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 438 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 439 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 440 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 441 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 442 443 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 444 445 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 446 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 447 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 448 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 449 450 /* User modifiable flags */ 451 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 452 EXT4_UNRM_FL | \ 453 EXT4_COMPR_FL | \ 454 EXT4_SYNC_FL | \ 455 EXT4_IMMUTABLE_FL | \ 456 EXT4_APPEND_FL | \ 457 EXT4_NODUMP_FL | \ 458 EXT4_NOATIME_FL | \ 459 EXT4_JOURNAL_DATA_FL | \ 460 EXT4_NOTAIL_FL | \ 461 EXT4_DIRSYNC_FL | \ 462 EXT4_TOPDIR_FL | \ 463 EXT4_EXTENTS_FL | \ 464 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 465 EXT4_DAX_FL | \ 466 EXT4_PROJINHERIT_FL | \ 467 EXT4_CASEFOLD_FL) 468 469 /* User visible flags */ 470 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 471 EXT4_DIRTY_FL | \ 472 EXT4_COMPRBLK_FL | \ 473 EXT4_NOCOMPR_FL | \ 474 EXT4_ENCRYPT_FL | \ 475 EXT4_INDEX_FL | \ 476 EXT4_VERITY_FL | \ 477 EXT4_INLINE_DATA_FL) 478 479 /* Flags that should be inherited by new inodes from their parent. */ 480 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 481 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 482 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 483 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 484 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 485 EXT4_DAX_FL) 486 487 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 488 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 489 EXT4_PROJINHERIT_FL)) 490 491 /* Flags that are appropriate for non-directories/regular files. */ 492 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 493 494 /* The only flags that should be swapped */ 495 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 496 497 /* Flags which are mutually exclusive to DAX */ 498 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 499 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 500 501 /* Mask out flags that are inappropriate for the given type of inode. */ 502 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 503 { 504 if (S_ISDIR(mode)) 505 return flags; 506 else if (S_ISREG(mode)) 507 return flags & EXT4_REG_FLMASK; 508 else 509 return flags & EXT4_OTHER_FLMASK; 510 } 511 512 /* 513 * Inode flags used for atomic set/get 514 */ 515 enum { 516 EXT4_INODE_SECRM = 0, /* Secure deletion */ 517 EXT4_INODE_UNRM = 1, /* Undelete */ 518 EXT4_INODE_COMPR = 2, /* Compress file */ 519 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 520 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 521 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 522 EXT4_INODE_NODUMP = 6, /* do not dump file */ 523 EXT4_INODE_NOATIME = 7, /* do not update atime */ 524 /* Reserved for compression usage... */ 525 EXT4_INODE_DIRTY = 8, 526 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 527 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 528 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 529 /* End compression flags --- maybe not all used */ 530 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 531 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 532 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 533 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 534 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 535 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 536 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 537 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 538 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 539 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 540 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 541 EXT4_INODE_DAX = 25, /* Inode is DAX */ 542 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 543 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 544 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 545 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 546 }; 547 548 /* 549 * Since it's pretty easy to mix up bit numbers and hex values, we use a 550 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 551 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 552 * any extra space in the compiled kernel image, otherwise, the build will fail. 553 * It's important that these values are the same, since we are using 554 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 555 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 556 * values found in ext2, ext3 and ext4 filesystems, and of course the values 557 * defined in e2fsprogs. 558 * 559 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 560 */ 561 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 562 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 563 564 static inline void ext4_check_flag_values(void) 565 { 566 CHECK_FLAG_VALUE(SECRM); 567 CHECK_FLAG_VALUE(UNRM); 568 CHECK_FLAG_VALUE(COMPR); 569 CHECK_FLAG_VALUE(SYNC); 570 CHECK_FLAG_VALUE(IMMUTABLE); 571 CHECK_FLAG_VALUE(APPEND); 572 CHECK_FLAG_VALUE(NODUMP); 573 CHECK_FLAG_VALUE(NOATIME); 574 CHECK_FLAG_VALUE(DIRTY); 575 CHECK_FLAG_VALUE(COMPRBLK); 576 CHECK_FLAG_VALUE(NOCOMPR); 577 CHECK_FLAG_VALUE(ENCRYPT); 578 CHECK_FLAG_VALUE(INDEX); 579 CHECK_FLAG_VALUE(IMAGIC); 580 CHECK_FLAG_VALUE(JOURNAL_DATA); 581 CHECK_FLAG_VALUE(NOTAIL); 582 CHECK_FLAG_VALUE(DIRSYNC); 583 CHECK_FLAG_VALUE(TOPDIR); 584 CHECK_FLAG_VALUE(HUGE_FILE); 585 CHECK_FLAG_VALUE(EXTENTS); 586 CHECK_FLAG_VALUE(VERITY); 587 CHECK_FLAG_VALUE(EA_INODE); 588 CHECK_FLAG_VALUE(INLINE_DATA); 589 CHECK_FLAG_VALUE(PROJINHERIT); 590 CHECK_FLAG_VALUE(CASEFOLD); 591 CHECK_FLAG_VALUE(RESERVED); 592 } 593 594 /* Used to pass group descriptor data when online resize is done */ 595 struct ext4_new_group_input { 596 __u32 group; /* Group number for this data */ 597 __u64 block_bitmap; /* Absolute block number of block bitmap */ 598 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 599 __u64 inode_table; /* Absolute block number of inode table start */ 600 __u32 blocks_count; /* Total number of blocks in this group */ 601 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 602 __u16 unused; 603 }; 604 605 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 606 struct compat_ext4_new_group_input { 607 u32 group; 608 compat_u64 block_bitmap; 609 compat_u64 inode_bitmap; 610 compat_u64 inode_table; 611 u32 blocks_count; 612 u16 reserved_blocks; 613 u16 unused; 614 }; 615 #endif 616 617 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 618 struct ext4_new_group_data { 619 __u32 group; 620 __u64 block_bitmap; 621 __u64 inode_bitmap; 622 __u64 inode_table; 623 __u32 blocks_count; 624 __u16 reserved_blocks; 625 __u16 mdata_blocks; 626 __u32 free_clusters_count; 627 }; 628 629 /* Indexes used to index group tables in ext4_new_group_data */ 630 enum { 631 BLOCK_BITMAP = 0, /* block bitmap */ 632 INODE_BITMAP, /* inode bitmap */ 633 INODE_TABLE, /* inode tables */ 634 GROUP_TABLE_COUNT, 635 }; 636 637 /* 638 * Flags used by ext4_map_blocks() 639 */ 640 /* Allocate any needed blocks and/or convert an unwritten 641 extent to be an initialized ext4 */ 642 #define EXT4_GET_BLOCKS_CREATE 0x0001 643 /* Request the creation of an unwritten extent */ 644 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 645 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 646 EXT4_GET_BLOCKS_CREATE) 647 /* Caller is from the delayed allocation writeout path 648 * finally doing the actual allocation of delayed blocks */ 649 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 650 /* caller is from the direct IO path, request to creation of an 651 unwritten extents if not allocated, split the unwritten 652 extent if blocks has been preallocated already*/ 653 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 654 #define EXT4_GET_BLOCKS_CONVERT 0x0010 655 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 656 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 657 /* Convert extent to initialized after IO complete */ 658 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 659 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 660 /* Eventual metadata allocation (due to growing extent tree) 661 * should not fail, so try to use reserved blocks for that.*/ 662 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 663 /* Don't normalize allocation size (used for fallocate) */ 664 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 665 /* Convert written extents to unwritten */ 666 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 667 /* Write zeros to newly created written extents */ 668 #define EXT4_GET_BLOCKS_ZERO 0x0200 669 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 670 EXT4_GET_BLOCKS_ZERO) 671 /* Caller will submit data before dropping transaction handle. This 672 * allows jbd2 to avoid submitting data before commit. */ 673 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 674 /* Caller is in the atomic contex, find extent if it has been cached */ 675 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 676 677 /* 678 * The bit position of these flags must not overlap with any of the 679 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 680 * read_extent_tree_block(), ext4_split_extent_at(), 681 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 682 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 683 * caching the extents when reading from the extent tree while a 684 * truncate or punch hole operation is in progress. 685 */ 686 #define EXT4_EX_NOCACHE 0x40000000 687 #define EXT4_EX_FORCE_CACHE 0x20000000 688 #define EXT4_EX_NOFAIL 0x10000000 689 690 /* 691 * Flags used by ext4_free_blocks 692 */ 693 #define EXT4_FREE_BLOCKS_METADATA 0x0001 694 #define EXT4_FREE_BLOCKS_FORGET 0x0002 695 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 696 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 697 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 698 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 699 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 700 701 /* 702 * ioctl commands 703 */ 704 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 705 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 706 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 707 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 708 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 709 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 710 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 711 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 712 #define EXT4_IOC_MIGRATE _IO('f', 9) 713 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 714 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 715 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 716 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 717 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 718 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 719 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 720 /* ioctl codes 19--39 are reserved for fscrypt */ 721 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 722 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 723 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 724 #define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32) 725 #define EXT4_IOC_GETFSUUID _IOR('f', 44, struct fsuuid) 726 #define EXT4_IOC_SETFSUUID _IOW('f', 44, struct fsuuid) 727 728 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 729 730 /* 731 * Flags for going down operation 732 */ 733 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 734 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 735 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 736 737 /* 738 * Flags returned by EXT4_IOC_GETSTATE 739 * 740 * We only expose to userspace a subset of the state flags in 741 * i_state_flags 742 */ 743 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 744 #define EXT4_STATE_FLAG_NEW 0x00000002 745 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 746 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 747 748 /* flags for ioctl EXT4_IOC_CHECKPOINT */ 749 #define EXT4_IOC_CHECKPOINT_FLAG_DISCARD 0x1 750 #define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT 0x2 751 #define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN 0x4 752 #define EXT4_IOC_CHECKPOINT_FLAG_VALID (EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \ 753 EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \ 754 EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN) 755 756 /* 757 * Structure for EXT4_IOC_GETFSUUID/EXT4_IOC_SETFSUUID 758 */ 759 struct fsuuid { 760 __u32 fsu_len; 761 __u32 fsu_flags; 762 __u8 fsu_uuid[]; 763 }; 764 765 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 766 /* 767 * ioctl commands in 32 bit emulation 768 */ 769 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 770 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 771 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 772 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 773 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 774 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 775 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 776 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 777 #endif 778 779 /* 780 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 781 * It indicates that the entry in extent status cache is for a hole. 782 */ 783 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 784 785 /* Max physical block we can address w/o extents */ 786 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 787 788 /* Max logical block we can support */ 789 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 790 791 /* 792 * Structure of an inode on the disk 793 */ 794 struct ext4_inode { 795 __le16 i_mode; /* File mode */ 796 __le16 i_uid; /* Low 16 bits of Owner Uid */ 797 __le32 i_size_lo; /* Size in bytes */ 798 __le32 i_atime; /* Access time */ 799 __le32 i_ctime; /* Inode Change time */ 800 __le32 i_mtime; /* Modification time */ 801 __le32 i_dtime; /* Deletion Time */ 802 __le16 i_gid; /* Low 16 bits of Group Id */ 803 __le16 i_links_count; /* Links count */ 804 __le32 i_blocks_lo; /* Blocks count */ 805 __le32 i_flags; /* File flags */ 806 union { 807 struct { 808 __le32 l_i_version; 809 } linux1; 810 struct { 811 __u32 h_i_translator; 812 } hurd1; 813 struct { 814 __u32 m_i_reserved1; 815 } masix1; 816 } osd1; /* OS dependent 1 */ 817 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 818 __le32 i_generation; /* File version (for NFS) */ 819 __le32 i_file_acl_lo; /* File ACL */ 820 __le32 i_size_high; 821 __le32 i_obso_faddr; /* Obsoleted fragment address */ 822 union { 823 struct { 824 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 825 __le16 l_i_file_acl_high; 826 __le16 l_i_uid_high; /* these 2 fields */ 827 __le16 l_i_gid_high; /* were reserved2[0] */ 828 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 829 __le16 l_i_reserved; 830 } linux2; 831 struct { 832 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 833 __u16 h_i_mode_high; 834 __u16 h_i_uid_high; 835 __u16 h_i_gid_high; 836 __u32 h_i_author; 837 } hurd2; 838 struct { 839 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 840 __le16 m_i_file_acl_high; 841 __u32 m_i_reserved2[2]; 842 } masix2; 843 } osd2; /* OS dependent 2 */ 844 __le16 i_extra_isize; 845 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 846 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 847 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 848 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 849 __le32 i_crtime; /* File Creation time */ 850 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 851 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 852 __le32 i_projid; /* Project ID */ 853 }; 854 855 struct move_extent { 856 __u32 reserved; /* should be zero */ 857 __u32 donor_fd; /* donor file descriptor */ 858 __u64 orig_start; /* logical start offset in block for orig */ 859 __u64 donor_start; /* logical start offset in block for donor */ 860 __u64 len; /* block length to be moved */ 861 __u64 moved_len; /* moved block length */ 862 }; 863 864 #define EXT4_EPOCH_BITS 2 865 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 866 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 867 868 /* 869 * Extended fields will fit into an inode if the filesystem was formatted 870 * with large inodes (-I 256 or larger) and there are not currently any EAs 871 * consuming all of the available space. For new inodes we always reserve 872 * enough space for the kernel's known extended fields, but for inodes 873 * created with an old kernel this might not have been the case. None of 874 * the extended inode fields is critical for correct filesystem operation. 875 * This macro checks if a certain field fits in the inode. Note that 876 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 877 */ 878 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 879 ((offsetof(typeof(*ext4_inode), field) + \ 880 sizeof((ext4_inode)->field)) \ 881 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 882 (einode)->i_extra_isize)) \ 883 884 /* 885 * We use an encoding that preserves the times for extra epoch "00": 886 * 887 * extra msb of adjust for signed 888 * epoch 32-bit 32-bit tv_sec to 889 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 890 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 891 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 892 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 893 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 894 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 895 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 896 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 897 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 898 * 899 * Note that previous versions of the kernel on 64-bit systems would 900 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 901 * 1970. e2fsck will correct this, assuming that it is run on the 902 * affected filesystem before 2242. 903 */ 904 905 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 906 { 907 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 908 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 909 } 910 911 static inline void ext4_decode_extra_time(struct timespec64 *time, 912 __le32 extra) 913 { 914 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 915 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 916 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 917 } 918 919 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 920 do { \ 921 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 922 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 923 (raw_inode)->xtime ## _extra = \ 924 ext4_encode_extra_time(&(inode)->xtime); \ 925 } \ 926 else \ 927 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 928 } while (0) 929 930 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 931 do { \ 932 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 933 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 934 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 935 (raw_inode)->xtime ## _extra = \ 936 ext4_encode_extra_time(&(einode)->xtime); \ 937 } while (0) 938 939 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 940 do { \ 941 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 942 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 943 ext4_decode_extra_time(&(inode)->xtime, \ 944 raw_inode->xtime ## _extra); \ 945 } \ 946 else \ 947 (inode)->xtime.tv_nsec = 0; \ 948 } while (0) 949 950 951 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 952 do { \ 953 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 954 (einode)->xtime.tv_sec = \ 955 (signed)le32_to_cpu((raw_inode)->xtime); \ 956 else \ 957 (einode)->xtime.tv_sec = 0; \ 958 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 959 ext4_decode_extra_time(&(einode)->xtime, \ 960 raw_inode->xtime ## _extra); \ 961 else \ 962 (einode)->xtime.tv_nsec = 0; \ 963 } while (0) 964 965 #define i_disk_version osd1.linux1.l_i_version 966 967 #if defined(__KERNEL__) || defined(__linux__) 968 #define i_reserved1 osd1.linux1.l_i_reserved1 969 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 970 #define i_blocks_high osd2.linux2.l_i_blocks_high 971 #define i_uid_low i_uid 972 #define i_gid_low i_gid 973 #define i_uid_high osd2.linux2.l_i_uid_high 974 #define i_gid_high osd2.linux2.l_i_gid_high 975 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 976 977 #elif defined(__GNU__) 978 979 #define i_translator osd1.hurd1.h_i_translator 980 #define i_uid_high osd2.hurd2.h_i_uid_high 981 #define i_gid_high osd2.hurd2.h_i_gid_high 982 #define i_author osd2.hurd2.h_i_author 983 984 #elif defined(__masix__) 985 986 #define i_reserved1 osd1.masix1.m_i_reserved1 987 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 988 #define i_reserved2 osd2.masix2.m_i_reserved2 989 990 #endif /* defined(__KERNEL__) || defined(__linux__) */ 991 992 #include "extents_status.h" 993 #include "fast_commit.h" 994 995 /* 996 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 997 * 998 * These are needed to avoid lockdep false positives when we need to 999 * allocate blocks to the quota inode during ext4_map_blocks(), while 1000 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1001 * do quota tracking for the quota inode, this avoids deadlock (as 1002 * well as infinite recursion, since it isn't turtles all the way 1003 * down...) 1004 * 1005 * I_DATA_SEM_NORMAL - Used for most inodes 1006 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1007 * where the second inode has larger inode number 1008 * than the first 1009 * I_DATA_SEM_QUOTA - Used for quota inodes only 1010 */ 1011 enum { 1012 I_DATA_SEM_NORMAL = 0, 1013 I_DATA_SEM_OTHER, 1014 I_DATA_SEM_QUOTA, 1015 }; 1016 1017 1018 /* 1019 * fourth extended file system inode data in memory 1020 */ 1021 struct ext4_inode_info { 1022 __le32 i_data[15]; /* unconverted */ 1023 __u32 i_dtime; 1024 ext4_fsblk_t i_file_acl; 1025 1026 /* 1027 * i_block_group is the number of the block group which contains 1028 * this file's inode. Constant across the lifetime of the inode, 1029 * it is used for making block allocation decisions - we try to 1030 * place a file's data blocks near its inode block, and new inodes 1031 * near to their parent directory's inode. 1032 */ 1033 ext4_group_t i_block_group; 1034 ext4_lblk_t i_dir_start_lookup; 1035 #if (BITS_PER_LONG < 64) 1036 unsigned long i_state_flags; /* Dynamic state flags */ 1037 #endif 1038 unsigned long i_flags; 1039 1040 /* 1041 * Extended attributes can be read independently of the main file 1042 * data. Taking i_rwsem even when reading would cause contention 1043 * between readers of EAs and writers of regular file data, so 1044 * instead we synchronize on xattr_sem when reading or changing 1045 * EAs. 1046 */ 1047 struct rw_semaphore xattr_sem; 1048 1049 /* 1050 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1051 * i_orphan is used. 1052 */ 1053 union { 1054 struct list_head i_orphan; /* unlinked but open inodes */ 1055 unsigned int i_orphan_idx; /* Index in orphan file */ 1056 }; 1057 1058 /* Fast commit related info */ 1059 1060 /* For tracking dentry create updates */ 1061 struct list_head i_fc_dilist; 1062 struct list_head i_fc_list; /* 1063 * inodes that need fast commit 1064 * protected by sbi->s_fc_lock. 1065 */ 1066 1067 /* Start of lblk range that needs to be committed in this fast commit */ 1068 ext4_lblk_t i_fc_lblk_start; 1069 1070 /* End of lblk range that needs to be committed in this fast commit */ 1071 ext4_lblk_t i_fc_lblk_len; 1072 1073 /* Number of ongoing updates on this inode */ 1074 atomic_t i_fc_updates; 1075 1076 /* Fast commit wait queue for this inode */ 1077 wait_queue_head_t i_fc_wait; 1078 1079 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1080 struct mutex i_fc_lock; 1081 1082 /* 1083 * i_disksize keeps track of what the inode size is ON DISK, not 1084 * in memory. During truncate, i_size is set to the new size by 1085 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1086 * set i_disksize to 0 until the truncate is actually under way. 1087 * 1088 * The intent is that i_disksize always represents the blocks which 1089 * are used by this file. This allows recovery to restart truncate 1090 * on orphans if we crash during truncate. We actually write i_disksize 1091 * into the on-disk inode when writing inodes out, instead of i_size. 1092 * 1093 * The only time when i_disksize and i_size may be different is when 1094 * a truncate is in progress. The only things which change i_disksize 1095 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1096 */ 1097 loff_t i_disksize; 1098 1099 /* 1100 * i_data_sem is for serialising ext4_truncate() against 1101 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1102 * data tree are chopped off during truncate. We can't do that in 1103 * ext4 because whenever we perform intermediate commits during 1104 * truncate, the inode and all the metadata blocks *must* be in a 1105 * consistent state which allows truncation of the orphans to restart 1106 * during recovery. Hence we must fix the get_block-vs-truncate race 1107 * by other means, so we have i_data_sem. 1108 */ 1109 struct rw_semaphore i_data_sem; 1110 struct inode vfs_inode; 1111 struct jbd2_inode *jinode; 1112 1113 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1114 1115 /* 1116 * File creation time. Its function is same as that of 1117 * struct timespec64 i_{a,c,m}time in the generic inode. 1118 */ 1119 struct timespec64 i_crtime; 1120 1121 /* mballoc */ 1122 atomic_t i_prealloc_active; 1123 struct rb_root i_prealloc_node; 1124 rwlock_t i_prealloc_lock; 1125 1126 /* extents status tree */ 1127 struct ext4_es_tree i_es_tree; 1128 rwlock_t i_es_lock; 1129 struct list_head i_es_list; 1130 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1131 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1132 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1133 extents to shrink. Protected by 1134 i_es_lock */ 1135 1136 /* ialloc */ 1137 ext4_group_t i_last_alloc_group; 1138 1139 /* allocation reservation info for delalloc */ 1140 /* In case of bigalloc, this refer to clusters rather than blocks */ 1141 unsigned int i_reserved_data_blocks; 1142 1143 /* pending cluster reservations for bigalloc file systems */ 1144 struct ext4_pending_tree i_pending_tree; 1145 1146 /* on-disk additional length */ 1147 __u16 i_extra_isize; 1148 1149 /* Indicate the inline data space. */ 1150 u16 i_inline_off; 1151 u16 i_inline_size; 1152 1153 #ifdef CONFIG_QUOTA 1154 /* quota space reservation, managed internally by quota code */ 1155 qsize_t i_reserved_quota; 1156 #endif 1157 1158 /* Lock protecting lists below */ 1159 spinlock_t i_completed_io_lock; 1160 /* 1161 * Completed IOs that need unwritten extents handling and have 1162 * transaction reserved 1163 */ 1164 struct list_head i_rsv_conversion_list; 1165 struct work_struct i_rsv_conversion_work; 1166 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1167 1168 spinlock_t i_block_reservation_lock; 1169 1170 /* 1171 * Transactions that contain inode's metadata needed to complete 1172 * fsync and fdatasync, respectively. 1173 */ 1174 tid_t i_sync_tid; 1175 tid_t i_datasync_tid; 1176 1177 #ifdef CONFIG_QUOTA 1178 struct dquot *i_dquot[MAXQUOTAS]; 1179 #endif 1180 1181 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1182 __u32 i_csum_seed; 1183 1184 kprojid_t i_projid; 1185 }; 1186 1187 /* 1188 * File system states 1189 */ 1190 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1191 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1192 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1193 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1194 1195 /* 1196 * Misc. filesystem flags 1197 */ 1198 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1199 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1200 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1201 1202 /* 1203 * Mount flags set via mount options or defaults 1204 */ 1205 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1206 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1207 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1208 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1209 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1210 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1211 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1212 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1213 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1214 #ifdef CONFIG_FS_DAX 1215 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1216 #else 1217 #define EXT4_MOUNT_DAX_ALWAYS 0 1218 #endif 1219 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1220 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1221 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1222 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1223 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1224 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1225 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1226 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1227 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1228 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1229 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1230 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1231 * enable enforcement for hidden 1232 * quota files */ 1233 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1234 * enforcement for hidden quota 1235 * files */ 1236 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1237 * enforcement */ 1238 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1239 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1240 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1241 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1242 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1243 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1244 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1245 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1246 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1247 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1248 1249 /* 1250 * Mount flags set either automatically (could not be set by mount option) 1251 * based on per file system feature or property or in special cases such as 1252 * distinguishing between explicit mount option definition and default. 1253 */ 1254 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1255 specified delalloc */ 1256 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1257 size of blocksize * 8 1258 blocks */ 1259 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1260 file systems */ 1261 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1262 specified journal checksum */ 1263 1264 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1265 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1266 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1267 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1268 * scanning in mballoc 1269 */ 1270 1271 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1272 ~EXT4_MOUNT_##opt 1273 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1274 EXT4_MOUNT_##opt 1275 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1276 EXT4_MOUNT_##opt) 1277 1278 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1279 ~EXT4_MOUNT2_##opt 1280 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1281 EXT4_MOUNT2_##opt 1282 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1283 EXT4_MOUNT2_##opt) 1284 1285 #define ext4_test_and_set_bit __test_and_set_bit_le 1286 #define ext4_set_bit __set_bit_le 1287 #define ext4_set_bit_atomic ext2_set_bit_atomic 1288 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1289 #define ext4_clear_bit __clear_bit_le 1290 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1291 #define ext4_test_bit test_bit_le 1292 #define ext4_find_next_zero_bit find_next_zero_bit_le 1293 #define ext4_find_next_bit find_next_bit_le 1294 1295 extern void mb_set_bits(void *bm, int cur, int len); 1296 1297 /* 1298 * Maximal mount counts between two filesystem checks 1299 */ 1300 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1301 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1302 1303 /* 1304 * Behaviour when detecting errors 1305 */ 1306 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1307 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1308 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1309 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1310 1311 /* Metadata checksum algorithm codes */ 1312 #define EXT4_CRC32C_CHKSUM 1 1313 1314 #define EXT4_LABEL_MAX 16 1315 1316 /* 1317 * Structure of the super block 1318 */ 1319 struct ext4_super_block { 1320 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1321 __le32 s_blocks_count_lo; /* Blocks count */ 1322 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1323 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1324 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1325 __le32 s_first_data_block; /* First Data Block */ 1326 __le32 s_log_block_size; /* Block size */ 1327 __le32 s_log_cluster_size; /* Allocation cluster size */ 1328 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1329 __le32 s_clusters_per_group; /* # Clusters per group */ 1330 __le32 s_inodes_per_group; /* # Inodes per group */ 1331 __le32 s_mtime; /* Mount time */ 1332 /*30*/ __le32 s_wtime; /* Write time */ 1333 __le16 s_mnt_count; /* Mount count */ 1334 __le16 s_max_mnt_count; /* Maximal mount count */ 1335 __le16 s_magic; /* Magic signature */ 1336 __le16 s_state; /* File system state */ 1337 __le16 s_errors; /* Behaviour when detecting errors */ 1338 __le16 s_minor_rev_level; /* minor revision level */ 1339 /*40*/ __le32 s_lastcheck; /* time of last check */ 1340 __le32 s_checkinterval; /* max. time between checks */ 1341 __le32 s_creator_os; /* OS */ 1342 __le32 s_rev_level; /* Revision level */ 1343 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1344 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1345 /* 1346 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1347 * 1348 * Note: the difference between the compatible feature set and 1349 * the incompatible feature set is that if there is a bit set 1350 * in the incompatible feature set that the kernel doesn't 1351 * know about, it should refuse to mount the filesystem. 1352 * 1353 * e2fsck's requirements are more strict; if it doesn't know 1354 * about a feature in either the compatible or incompatible 1355 * feature set, it must abort and not try to meddle with 1356 * things it doesn't understand... 1357 */ 1358 __le32 s_first_ino; /* First non-reserved inode */ 1359 __le16 s_inode_size; /* size of inode structure */ 1360 __le16 s_block_group_nr; /* block group # of this superblock */ 1361 __le32 s_feature_compat; /* compatible feature set */ 1362 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1363 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1364 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1365 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1366 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1367 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1368 /* 1369 * Performance hints. Directory preallocation should only 1370 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1371 */ 1372 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1373 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1374 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1375 /* 1376 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1377 */ 1378 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1379 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1380 __le32 s_journal_dev; /* device number of journal file */ 1381 __le32 s_last_orphan; /* start of list of inodes to delete */ 1382 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1383 __u8 s_def_hash_version; /* Default hash version to use */ 1384 __u8 s_jnl_backup_type; 1385 __le16 s_desc_size; /* size of group descriptor */ 1386 /*100*/ __le32 s_default_mount_opts; 1387 __le32 s_first_meta_bg; /* First metablock block group */ 1388 __le32 s_mkfs_time; /* When the filesystem was created */ 1389 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1390 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1391 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1392 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1393 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1394 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1395 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1396 __le32 s_flags; /* Miscellaneous flags */ 1397 __le16 s_raid_stride; /* RAID stride */ 1398 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1399 __le64 s_mmp_block; /* Block for multi-mount protection */ 1400 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1401 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1402 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1403 __u8 s_encryption_level; /* versioning level for encryption */ 1404 __u8 s_reserved_pad; /* Padding to next 32bits */ 1405 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1406 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1407 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1408 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1409 snapshot's future use */ 1410 __le32 s_snapshot_list; /* inode number of the head of the 1411 on-disk snapshot list */ 1412 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1413 __le32 s_error_count; /* number of fs errors */ 1414 __le32 s_first_error_time; /* first time an error happened */ 1415 __le32 s_first_error_ino; /* inode involved in first error */ 1416 __le64 s_first_error_block; /* block involved of first error */ 1417 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1418 __le32 s_first_error_line; /* line number where error happened */ 1419 __le32 s_last_error_time; /* most recent time of an error */ 1420 __le32 s_last_error_ino; /* inode involved in last error */ 1421 __le32 s_last_error_line; /* line number where error happened */ 1422 __le64 s_last_error_block; /* block involved of last error */ 1423 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1424 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1425 __u8 s_mount_opts[64]; 1426 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1427 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1428 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1429 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1430 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1431 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1432 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1433 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1434 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1435 __u8 s_wtime_hi; 1436 __u8 s_mtime_hi; 1437 __u8 s_mkfs_time_hi; 1438 __u8 s_lastcheck_hi; 1439 __u8 s_first_error_time_hi; 1440 __u8 s_last_error_time_hi; 1441 __u8 s_first_error_errcode; 1442 __u8 s_last_error_errcode; 1443 __le16 s_encoding; /* Filename charset encoding */ 1444 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1445 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1446 __le32 s_reserved[94]; /* Padding to the end of the block */ 1447 __le32 s_checksum; /* crc32c(superblock) */ 1448 }; 1449 1450 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1451 1452 #ifdef __KERNEL__ 1453 1454 /* Number of quota types we support */ 1455 #define EXT4_MAXQUOTAS 3 1456 1457 #define EXT4_ENC_UTF8_12_1 1 1458 1459 /* Types of ext4 journal triggers */ 1460 enum ext4_journal_trigger_type { 1461 EXT4_JTR_ORPHAN_FILE, 1462 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1463 }; 1464 1465 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1466 1467 struct ext4_journal_trigger { 1468 struct jbd2_buffer_trigger_type tr_triggers; 1469 struct super_block *sb; 1470 }; 1471 1472 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1473 struct jbd2_buffer_trigger_type *trigger) 1474 { 1475 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1476 } 1477 1478 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1479 1480 /* Structure at the tail of orphan block */ 1481 struct ext4_orphan_block_tail { 1482 __le32 ob_magic; 1483 __le32 ob_checksum; 1484 }; 1485 1486 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1487 { 1488 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1489 sizeof(u32); 1490 } 1491 1492 struct ext4_orphan_block { 1493 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1494 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1495 }; 1496 1497 /* 1498 * Info about orphan file. 1499 */ 1500 struct ext4_orphan_info { 1501 int of_blocks; /* Number of orphan blocks in a file */ 1502 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1503 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1504 * file blocks */ 1505 }; 1506 1507 /* 1508 * fourth extended-fs super-block data in memory 1509 */ 1510 struct ext4_sb_info { 1511 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1512 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1513 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1514 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1515 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1516 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1517 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1518 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1519 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1520 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1521 unsigned long s_overhead; /* # of fs overhead clusters */ 1522 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1523 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1524 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1525 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1526 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1527 struct buffer_head * __rcu *s_group_desc; 1528 unsigned int s_mount_opt; 1529 unsigned int s_mount_opt2; 1530 unsigned long s_mount_flags; 1531 unsigned int s_def_mount_opt; 1532 unsigned int s_def_mount_opt2; 1533 ext4_fsblk_t s_sb_block; 1534 atomic64_t s_resv_clusters; 1535 kuid_t s_resuid; 1536 kgid_t s_resgid; 1537 unsigned short s_mount_state; 1538 unsigned short s_pad; 1539 int s_addr_per_block_bits; 1540 int s_desc_per_block_bits; 1541 int s_inode_size; 1542 int s_first_ino; 1543 unsigned int s_inode_readahead_blks; 1544 unsigned int s_inode_goal; 1545 u32 s_hash_seed[4]; 1546 int s_def_hash_version; 1547 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1548 struct percpu_counter s_freeclusters_counter; 1549 struct percpu_counter s_freeinodes_counter; 1550 struct percpu_counter s_dirs_counter; 1551 struct percpu_counter s_dirtyclusters_counter; 1552 struct percpu_counter s_sra_exceeded_retry_limit; 1553 struct blockgroup_lock *s_blockgroup_lock; 1554 struct proc_dir_entry *s_proc; 1555 struct kobject s_kobj; 1556 struct completion s_kobj_unregister; 1557 struct super_block *s_sb; 1558 struct buffer_head *s_mmp_bh; 1559 1560 /* Journaling */ 1561 struct journal_s *s_journal; 1562 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1563 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1564 struct list_head s_orphan; /* List of orphaned inodes in on disk 1565 list */ 1566 struct ext4_orphan_info s_orphan_info; 1567 unsigned long s_commit_interval; 1568 u32 s_max_batch_time; 1569 u32 s_min_batch_time; 1570 struct block_device *s_journal_bdev; 1571 #ifdef CONFIG_QUOTA 1572 /* Names of quota files with journalled quota */ 1573 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1574 int s_jquota_fmt; /* Format of quota to use */ 1575 #endif 1576 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1577 struct ext4_system_blocks __rcu *s_system_blks; 1578 1579 #ifdef EXTENTS_STATS 1580 /* ext4 extents stats */ 1581 unsigned long s_ext_min; 1582 unsigned long s_ext_max; 1583 unsigned long s_depth_max; 1584 spinlock_t s_ext_stats_lock; 1585 unsigned long s_ext_blocks; 1586 unsigned long s_ext_extents; 1587 #endif 1588 1589 /* for buddy allocator */ 1590 struct ext4_group_info ** __rcu *s_group_info; 1591 struct inode *s_buddy_cache; 1592 spinlock_t s_md_lock; 1593 unsigned short *s_mb_offsets; 1594 unsigned int *s_mb_maxs; 1595 unsigned int s_group_info_size; 1596 unsigned int s_mb_free_pending; 1597 struct list_head s_freed_data_list; /* List of blocks to be freed 1598 after commit completed */ 1599 struct list_head s_discard_list; 1600 struct work_struct s_discard_work; 1601 atomic_t s_retry_alloc_pending; 1602 struct list_head *s_mb_avg_fragment_size; 1603 rwlock_t *s_mb_avg_fragment_size_locks; 1604 struct list_head *s_mb_largest_free_orders; 1605 rwlock_t *s_mb_largest_free_orders_locks; 1606 1607 /* tunables */ 1608 unsigned long s_stripe; 1609 unsigned int s_mb_max_linear_groups; 1610 unsigned int s_mb_stream_request; 1611 unsigned int s_mb_max_to_scan; 1612 unsigned int s_mb_min_to_scan; 1613 unsigned int s_mb_stats; 1614 unsigned int s_mb_order2_reqs; 1615 unsigned int s_mb_group_prealloc; 1616 unsigned int s_max_dir_size_kb; 1617 /* where last allocation was done - for stream allocation */ 1618 unsigned long s_mb_last_group; 1619 unsigned long s_mb_last_start; 1620 unsigned int s_mb_prefetch; 1621 unsigned int s_mb_prefetch_limit; 1622 1623 /* stats for buddy allocator */ 1624 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1625 atomic_t s_bal_success; /* we found long enough chunks */ 1626 atomic_t s_bal_allocated; /* in blocks */ 1627 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1628 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1629 atomic_t s_bal_goals; /* goal hits */ 1630 atomic_t s_bal_breaks; /* too long searches */ 1631 atomic_t s_bal_2orders; /* 2^order hits */ 1632 atomic_t s_bal_cr0_bad_suggestions; 1633 atomic_t s_bal_cr1_bad_suggestions; 1634 atomic64_t s_bal_cX_groups_considered[4]; 1635 atomic64_t s_bal_cX_hits[4]; 1636 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1637 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1638 atomic64_t s_mb_generation_time; 1639 atomic_t s_mb_lost_chunks; 1640 atomic_t s_mb_preallocated; 1641 atomic_t s_mb_discarded; 1642 atomic_t s_lock_busy; 1643 1644 /* locality groups */ 1645 struct ext4_locality_group __percpu *s_locality_groups; 1646 1647 /* for write statistics */ 1648 unsigned long s_sectors_written_start; 1649 u64 s_kbytes_written; 1650 1651 /* the size of zero-out chunk */ 1652 unsigned int s_extent_max_zeroout_kb; 1653 1654 unsigned int s_log_groups_per_flex; 1655 struct flex_groups * __rcu *s_flex_groups; 1656 ext4_group_t s_flex_groups_allocated; 1657 1658 /* workqueue for reserved extent conversions (buffered io) */ 1659 struct workqueue_struct *rsv_conversion_wq; 1660 1661 /* timer for periodic error stats printing */ 1662 struct timer_list s_err_report; 1663 1664 /* Lazy inode table initialization info */ 1665 struct ext4_li_request *s_li_request; 1666 /* Wait multiplier for lazy initialization thread */ 1667 unsigned int s_li_wait_mult; 1668 1669 /* Kernel thread for multiple mount protection */ 1670 struct task_struct *s_mmp_tsk; 1671 1672 /* record the last minlen when FITRIM is called. */ 1673 unsigned long s_last_trim_minblks; 1674 1675 /* Reference to checksum algorithm driver via cryptoapi */ 1676 struct crypto_shash *s_chksum_driver; 1677 1678 /* Precomputed FS UUID checksum for seeding other checksums */ 1679 __u32 s_csum_seed; 1680 1681 /* Reclaim extents from extent status tree */ 1682 struct shrinker s_es_shrinker; 1683 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1684 long s_es_nr_inode; 1685 struct ext4_es_stats s_es_stats; 1686 struct mb_cache *s_ea_block_cache; 1687 struct mb_cache *s_ea_inode_cache; 1688 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1689 1690 /* Journal triggers for checksum computation */ 1691 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1692 1693 /* Ratelimit ext4 messages. */ 1694 struct ratelimit_state s_err_ratelimit_state; 1695 struct ratelimit_state s_warning_ratelimit_state; 1696 struct ratelimit_state s_msg_ratelimit_state; 1697 atomic_t s_warning_count; 1698 atomic_t s_msg_count; 1699 1700 /* Encryption policy for '-o test_dummy_encryption' */ 1701 struct fscrypt_dummy_policy s_dummy_enc_policy; 1702 1703 /* 1704 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1705 * or EXTENTS flag. 1706 */ 1707 struct percpu_rw_semaphore s_writepages_rwsem; 1708 struct dax_device *s_daxdev; 1709 u64 s_dax_part_off; 1710 #ifdef CONFIG_EXT4_DEBUG 1711 unsigned long s_simulate_fail; 1712 #endif 1713 /* Record the errseq of the backing block device */ 1714 errseq_t s_bdev_wb_err; 1715 spinlock_t s_bdev_wb_lock; 1716 1717 /* Information about errors that happened during this mount */ 1718 spinlock_t s_error_lock; 1719 int s_add_error_count; 1720 int s_first_error_code; 1721 __u32 s_first_error_line; 1722 __u32 s_first_error_ino; 1723 __u64 s_first_error_block; 1724 const char *s_first_error_func; 1725 time64_t s_first_error_time; 1726 int s_last_error_code; 1727 __u32 s_last_error_line; 1728 __u32 s_last_error_ino; 1729 __u64 s_last_error_block; 1730 const char *s_last_error_func; 1731 time64_t s_last_error_time; 1732 /* 1733 * If we are in a context where we cannot update error information in 1734 * the on-disk superblock, we queue this work to do it. 1735 */ 1736 struct work_struct s_error_work; 1737 1738 /* Ext4 fast commit sub transaction ID */ 1739 atomic_t s_fc_subtid; 1740 1741 /* 1742 * After commit starts, the main queue gets locked, and the further 1743 * updates get added in the staging queue. 1744 */ 1745 #define FC_Q_MAIN 0 1746 #define FC_Q_STAGING 1 1747 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1748 * that have data changes in them. 1749 */ 1750 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1751 unsigned int s_fc_bytes; 1752 /* 1753 * Main fast commit lock. This lock protects accesses to the 1754 * following fields: 1755 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1756 */ 1757 spinlock_t s_fc_lock; 1758 struct buffer_head *s_fc_bh; 1759 struct ext4_fc_stats s_fc_stats; 1760 tid_t s_fc_ineligible_tid; 1761 #ifdef CONFIG_EXT4_DEBUG 1762 int s_fc_debug_max_replay; 1763 #endif 1764 struct ext4_fc_replay_state s_fc_replay_state; 1765 }; 1766 1767 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1768 { 1769 return sb->s_fs_info; 1770 } 1771 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1772 { 1773 return container_of(inode, struct ext4_inode_info, vfs_inode); 1774 } 1775 1776 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1777 { 1778 return ino == EXT4_ROOT_INO || 1779 (ino >= EXT4_FIRST_INO(sb) && 1780 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1781 } 1782 1783 /* 1784 * Returns: sbi->field[index] 1785 * Used to access an array element from the following sbi fields which require 1786 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1787 * - s_group_desc 1788 * - s_group_info 1789 * - s_flex_group 1790 */ 1791 #define sbi_array_rcu_deref(sbi, field, index) \ 1792 ({ \ 1793 typeof(*((sbi)->field)) _v; \ 1794 rcu_read_lock(); \ 1795 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1796 rcu_read_unlock(); \ 1797 _v; \ 1798 }) 1799 1800 /* 1801 * run-time mount flags 1802 */ 1803 enum { 1804 EXT4_MF_MNTDIR_SAMPLED, 1805 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1806 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1807 }; 1808 1809 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1810 { 1811 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1812 } 1813 1814 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1815 { 1816 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1817 } 1818 1819 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1820 { 1821 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1822 } 1823 1824 1825 /* 1826 * Simulate_fail codes 1827 */ 1828 #define EXT4_SIM_BBITMAP_EIO 1 1829 #define EXT4_SIM_BBITMAP_CRC 2 1830 #define EXT4_SIM_IBITMAP_EIO 3 1831 #define EXT4_SIM_IBITMAP_CRC 4 1832 #define EXT4_SIM_INODE_EIO 5 1833 #define EXT4_SIM_INODE_CRC 6 1834 #define EXT4_SIM_DIRBLOCK_EIO 7 1835 #define EXT4_SIM_DIRBLOCK_CRC 8 1836 1837 static inline bool ext4_simulate_fail(struct super_block *sb, 1838 unsigned long code) 1839 { 1840 #ifdef CONFIG_EXT4_DEBUG 1841 struct ext4_sb_info *sbi = EXT4_SB(sb); 1842 1843 if (unlikely(sbi->s_simulate_fail == code)) { 1844 sbi->s_simulate_fail = 0; 1845 return true; 1846 } 1847 #endif 1848 return false; 1849 } 1850 1851 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1852 struct buffer_head *bh, 1853 unsigned long code) 1854 { 1855 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1856 clear_buffer_uptodate(bh); 1857 } 1858 1859 /* 1860 * Error number codes for s_{first,last}_error_errno 1861 * 1862 * Linux errno numbers are architecture specific, so we need to translate 1863 * them into something which is architecture independent. We don't define 1864 * codes for all errno's; just the ones which are most likely to be the cause 1865 * of an ext4_error() call. 1866 */ 1867 #define EXT4_ERR_UNKNOWN 1 1868 #define EXT4_ERR_EIO 2 1869 #define EXT4_ERR_ENOMEM 3 1870 #define EXT4_ERR_EFSBADCRC 4 1871 #define EXT4_ERR_EFSCORRUPTED 5 1872 #define EXT4_ERR_ENOSPC 6 1873 #define EXT4_ERR_ENOKEY 7 1874 #define EXT4_ERR_EROFS 8 1875 #define EXT4_ERR_EFBIG 9 1876 #define EXT4_ERR_EEXIST 10 1877 #define EXT4_ERR_ERANGE 11 1878 #define EXT4_ERR_EOVERFLOW 12 1879 #define EXT4_ERR_EBUSY 13 1880 #define EXT4_ERR_ENOTDIR 14 1881 #define EXT4_ERR_ENOTEMPTY 15 1882 #define EXT4_ERR_ESHUTDOWN 16 1883 #define EXT4_ERR_EFAULT 17 1884 1885 /* 1886 * Inode dynamic state flags 1887 */ 1888 enum { 1889 EXT4_STATE_JDATA, /* journaled data exists */ 1890 EXT4_STATE_NEW, /* inode is newly created */ 1891 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1892 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1893 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1894 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1895 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1896 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1897 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1898 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1899 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1900 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1901 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1902 }; 1903 1904 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1905 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1906 { \ 1907 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1908 } \ 1909 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1910 { \ 1911 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1912 } \ 1913 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1914 { \ 1915 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1916 } 1917 1918 /* Add these declarations here only so that these functions can be 1919 * found by name. Otherwise, they are very hard to locate. */ 1920 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1921 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1922 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1923 EXT4_INODE_BIT_FNS(flag, flags, 0) 1924 1925 /* Add these declarations here only so that these functions can be 1926 * found by name. Otherwise, they are very hard to locate. */ 1927 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1928 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1929 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1930 #if (BITS_PER_LONG < 64) 1931 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1932 1933 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1934 { 1935 (ei)->i_state_flags = 0; 1936 } 1937 #else 1938 EXT4_INODE_BIT_FNS(state, flags, 32) 1939 1940 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1941 { 1942 /* We depend on the fact that callers will set i_flags */ 1943 } 1944 #endif 1945 #else 1946 /* Assume that user mode programs are passing in an ext4fs superblock, not 1947 * a kernel struct super_block. This will allow us to call the feature-test 1948 * macros from user land. */ 1949 #define EXT4_SB(sb) (sb) 1950 #endif 1951 1952 static inline bool ext4_verity_in_progress(struct inode *inode) 1953 { 1954 return IS_ENABLED(CONFIG_FS_VERITY) && 1955 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1956 } 1957 1958 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1959 1960 /* 1961 * Codes for operating systems 1962 */ 1963 #define EXT4_OS_LINUX 0 1964 #define EXT4_OS_HURD 1 1965 #define EXT4_OS_MASIX 2 1966 #define EXT4_OS_FREEBSD 3 1967 #define EXT4_OS_LITES 4 1968 1969 /* 1970 * Revision levels 1971 */ 1972 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1973 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1974 1975 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1976 1977 #define EXT4_GOOD_OLD_INODE_SIZE 128 1978 1979 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1980 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1981 #define EXT4_TIMESTAMP_MIN S32_MIN 1982 1983 /* 1984 * Feature set definitions 1985 */ 1986 1987 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1988 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1989 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1990 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1991 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1992 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1993 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1994 /* 1995 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1996 * incompatible only if fast commit blocks are present in the FS. Since we 1997 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1998 * incompatible. We also have a JBD2 incompat feature, which gets set when 1999 * there are fast commit blocks present in the journal. 2000 */ 2001 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2002 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2003 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2004 2005 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2006 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2007 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2008 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2009 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2010 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2011 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2012 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2013 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2014 /* 2015 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2016 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2017 * all other data structures' checksums. However, the METADATA_CSUM and 2018 * GDT_CSUM bits are mutually exclusive. 2019 */ 2020 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2021 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2022 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2023 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2024 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2025 non-empty */ 2026 2027 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2028 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2029 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2030 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2031 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2032 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2033 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2034 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2035 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2036 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2037 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2038 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2039 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2040 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2041 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2042 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2043 2044 extern void ext4_update_dynamic_rev(struct super_block *sb); 2045 2046 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2047 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2048 { \ 2049 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2050 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2051 } \ 2052 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2053 { \ 2054 ext4_update_dynamic_rev(sb); \ 2055 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2056 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2057 } \ 2058 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2059 { \ 2060 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2061 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2062 } 2063 2064 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2065 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2066 { \ 2067 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2068 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2069 } \ 2070 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2071 { \ 2072 ext4_update_dynamic_rev(sb); \ 2073 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2074 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2075 } \ 2076 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2077 { \ 2078 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2079 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2080 } 2081 2082 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2083 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2084 { \ 2085 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2086 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2087 } \ 2088 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2089 { \ 2090 ext4_update_dynamic_rev(sb); \ 2091 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2092 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2093 } \ 2094 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2095 { \ 2096 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2097 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2098 } 2099 2100 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2101 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2102 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2103 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2104 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2105 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2106 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2107 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2108 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2109 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2110 2111 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2112 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2113 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2114 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2115 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2116 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2117 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2118 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2119 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2120 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2121 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2122 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2123 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2124 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2125 2126 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2127 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2128 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2129 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2130 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2131 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2132 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2133 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2134 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2135 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2136 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2137 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2138 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2139 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2140 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2141 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2142 2143 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2144 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2145 EXT4_FEATURE_INCOMPAT_META_BG) 2146 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2147 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2148 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2149 2150 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2151 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2152 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2153 EXT4_FEATURE_INCOMPAT_META_BG) 2154 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2155 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2156 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2157 2158 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2159 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2160 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2161 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2162 EXT4_FEATURE_INCOMPAT_META_BG| \ 2163 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2164 EXT4_FEATURE_INCOMPAT_64BIT| \ 2165 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2166 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2167 EXT4_FEATURE_INCOMPAT_MMP | \ 2168 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2169 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2170 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2171 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2172 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2173 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2174 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2175 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2176 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2177 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2178 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2179 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2180 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2181 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2182 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2183 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2184 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2185 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2186 2187 #define EXTN_FEATURE_FUNCS(ver) \ 2188 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2189 { \ 2190 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2191 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2192 } \ 2193 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2194 { \ 2195 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2196 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2197 } \ 2198 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2199 { \ 2200 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2201 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2202 } 2203 2204 EXTN_FEATURE_FUNCS(2) 2205 EXTN_FEATURE_FUNCS(3) 2206 EXTN_FEATURE_FUNCS(4) 2207 2208 static inline bool ext4_has_compat_features(struct super_block *sb) 2209 { 2210 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2211 } 2212 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2213 { 2214 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2215 } 2216 static inline bool ext4_has_incompat_features(struct super_block *sb) 2217 { 2218 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2219 } 2220 2221 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2222 2223 /* 2224 * Superblock flags 2225 */ 2226 #define EXT4_FLAGS_RESIZING 0 2227 #define EXT4_FLAGS_SHUTDOWN 1 2228 #define EXT4_FLAGS_BDEV_IS_DAX 2 2229 2230 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2231 { 2232 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2233 } 2234 2235 /* 2236 * Default values for user and/or group using reserved blocks 2237 */ 2238 #define EXT4_DEF_RESUID 0 2239 #define EXT4_DEF_RESGID 0 2240 2241 /* 2242 * Default project ID 2243 */ 2244 #define EXT4_DEF_PROJID 0 2245 2246 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2247 2248 /* 2249 * Default mount options 2250 */ 2251 #define EXT4_DEFM_DEBUG 0x0001 2252 #define EXT4_DEFM_BSDGROUPS 0x0002 2253 #define EXT4_DEFM_XATTR_USER 0x0004 2254 #define EXT4_DEFM_ACL 0x0008 2255 #define EXT4_DEFM_UID16 0x0010 2256 #define EXT4_DEFM_JMODE 0x0060 2257 #define EXT4_DEFM_JMODE_DATA 0x0020 2258 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2259 #define EXT4_DEFM_JMODE_WBACK 0x0060 2260 #define EXT4_DEFM_NOBARRIER 0x0100 2261 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2262 #define EXT4_DEFM_DISCARD 0x0400 2263 #define EXT4_DEFM_NODELALLOC 0x0800 2264 2265 /* 2266 * Default journal batch times 2267 */ 2268 #define EXT4_DEF_MIN_BATCH_TIME 0 2269 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2270 2271 /* 2272 * Minimum number of groups in a flexgroup before we separate out 2273 * directories into the first block group of a flexgroup 2274 */ 2275 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2276 2277 /* 2278 * Structure of a directory entry 2279 */ 2280 #define EXT4_NAME_LEN 255 2281 /* 2282 * Base length of the ext4 directory entry excluding the name length 2283 */ 2284 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2285 2286 struct ext4_dir_entry { 2287 __le32 inode; /* Inode number */ 2288 __le16 rec_len; /* Directory entry length */ 2289 __le16 name_len; /* Name length */ 2290 char name[EXT4_NAME_LEN]; /* File name */ 2291 }; 2292 2293 2294 /* 2295 * Encrypted Casefolded entries require saving the hash on disk. This structure 2296 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2297 * boundary. 2298 */ 2299 struct ext4_dir_entry_hash { 2300 __le32 hash; 2301 __le32 minor_hash; 2302 }; 2303 2304 /* 2305 * The new version of the directory entry. Since EXT4 structures are 2306 * stored in intel byte order, and the name_len field could never be 2307 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2308 * file_type field. 2309 */ 2310 struct ext4_dir_entry_2 { 2311 __le32 inode; /* Inode number */ 2312 __le16 rec_len; /* Directory entry length */ 2313 __u8 name_len; /* Name length */ 2314 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2315 char name[EXT4_NAME_LEN]; /* File name */ 2316 }; 2317 2318 /* 2319 * Access the hashes at the end of ext4_dir_entry_2 2320 */ 2321 #define EXT4_DIRENT_HASHES(entry) \ 2322 ((struct ext4_dir_entry_hash *) \ 2323 (((void *)(entry)) + \ 2324 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2325 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2326 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2327 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2328 2329 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2330 { 2331 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2332 } 2333 2334 /* 2335 * This is a bogus directory entry at the end of each leaf block that 2336 * records checksums. 2337 */ 2338 struct ext4_dir_entry_tail { 2339 __le32 det_reserved_zero1; /* Pretend to be unused */ 2340 __le16 det_rec_len; /* 12 */ 2341 __u8 det_reserved_zero2; /* Zero name length */ 2342 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2343 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2344 }; 2345 2346 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2347 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2348 ((blocksize) - \ 2349 sizeof(struct ext4_dir_entry_tail)))) 2350 2351 /* 2352 * Ext4 directory file types. Only the low 3 bits are used. The 2353 * other bits are reserved for now. 2354 */ 2355 #define EXT4_FT_UNKNOWN 0 2356 #define EXT4_FT_REG_FILE 1 2357 #define EXT4_FT_DIR 2 2358 #define EXT4_FT_CHRDEV 3 2359 #define EXT4_FT_BLKDEV 4 2360 #define EXT4_FT_FIFO 5 2361 #define EXT4_FT_SOCK 6 2362 #define EXT4_FT_SYMLINK 7 2363 2364 #define EXT4_FT_MAX 8 2365 2366 #define EXT4_FT_DIR_CSUM 0xDE 2367 2368 /* 2369 * EXT4_DIR_PAD defines the directory entries boundaries 2370 * 2371 * NOTE: It must be a multiple of 4 2372 */ 2373 #define EXT4_DIR_PAD 4 2374 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2375 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2376 2377 /* 2378 * The rec_len is dependent on the type of directory. Directories that are 2379 * casefolded and encrypted need to store the hash as well, so we add room for 2380 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2381 * pass NULL for dir, as those entries do not use the extra fields. 2382 */ 2383 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2384 const struct inode *dir) 2385 { 2386 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2387 2388 if (dir && ext4_hash_in_dirent(dir)) 2389 rec_len += sizeof(struct ext4_dir_entry_hash); 2390 return (rec_len & ~EXT4_DIR_ROUND); 2391 } 2392 2393 /* 2394 * If we ever get support for fs block sizes > page_size, we'll need 2395 * to remove the #if statements in the next two functions... 2396 */ 2397 static inline unsigned int 2398 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2399 { 2400 unsigned len = le16_to_cpu(dlen); 2401 2402 #if (PAGE_SIZE >= 65536) 2403 if (len == EXT4_MAX_REC_LEN || len == 0) 2404 return blocksize; 2405 return (len & 65532) | ((len & 3) << 16); 2406 #else 2407 return len; 2408 #endif 2409 } 2410 2411 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2412 { 2413 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2414 #if (PAGE_SIZE >= 65536) 2415 if (len < 65536) 2416 return cpu_to_le16(len); 2417 if (len == blocksize) { 2418 if (blocksize == 65536) 2419 return cpu_to_le16(EXT4_MAX_REC_LEN); 2420 else 2421 return cpu_to_le16(0); 2422 } 2423 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2424 #else 2425 return cpu_to_le16(len); 2426 #endif 2427 } 2428 2429 /* 2430 * Hash Tree Directory indexing 2431 * (c) Daniel Phillips, 2001 2432 */ 2433 2434 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2435 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2436 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2437 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2438 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2439 2440 /* Legal values for the dx_root hash_version field: */ 2441 2442 #define DX_HASH_LEGACY 0 2443 #define DX_HASH_HALF_MD4 1 2444 #define DX_HASH_TEA 2 2445 #define DX_HASH_LEGACY_UNSIGNED 3 2446 #define DX_HASH_HALF_MD4_UNSIGNED 4 2447 #define DX_HASH_TEA_UNSIGNED 5 2448 #define DX_HASH_SIPHASH 6 2449 2450 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2451 const void *address, unsigned int length) 2452 { 2453 struct { 2454 struct shash_desc shash; 2455 char ctx[4]; 2456 } desc; 2457 2458 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2459 2460 desc.shash.tfm = sbi->s_chksum_driver; 2461 *(u32 *)desc.ctx = crc; 2462 2463 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2464 2465 return *(u32 *)desc.ctx; 2466 } 2467 2468 #ifdef __KERNEL__ 2469 2470 /* hash info structure used by the directory hash */ 2471 struct dx_hash_info 2472 { 2473 u32 hash; 2474 u32 minor_hash; 2475 int hash_version; 2476 u32 *seed; 2477 }; 2478 2479 2480 /* 32 and 64 bit signed EOF for dx directories */ 2481 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2482 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2483 2484 2485 /* 2486 * Control parameters used by ext4_htree_next_block 2487 */ 2488 #define HASH_NB_ALWAYS 1 2489 2490 struct ext4_filename { 2491 const struct qstr *usr_fname; 2492 struct fscrypt_str disk_name; 2493 struct dx_hash_info hinfo; 2494 #ifdef CONFIG_FS_ENCRYPTION 2495 struct fscrypt_str crypto_buf; 2496 #endif 2497 #if IS_ENABLED(CONFIG_UNICODE) 2498 struct fscrypt_str cf_name; 2499 #endif 2500 }; 2501 2502 #define fname_name(p) ((p)->disk_name.name) 2503 #define fname_usr_name(p) ((p)->usr_fname->name) 2504 #define fname_len(p) ((p)->disk_name.len) 2505 2506 /* 2507 * Describe an inode's exact location on disk and in memory 2508 */ 2509 struct ext4_iloc 2510 { 2511 struct buffer_head *bh; 2512 unsigned long offset; 2513 ext4_group_t block_group; 2514 }; 2515 2516 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2517 { 2518 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2519 } 2520 2521 static inline bool ext4_is_quota_file(struct inode *inode) 2522 { 2523 return IS_NOQUOTA(inode) && 2524 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2525 } 2526 2527 /* 2528 * This structure is stuffed into the struct file's private_data field 2529 * for directories. It is where we put information so that we can do 2530 * readdir operations in hash tree order. 2531 */ 2532 struct dir_private_info { 2533 struct rb_root root; 2534 struct rb_node *curr_node; 2535 struct fname *extra_fname; 2536 loff_t last_pos; 2537 __u32 curr_hash; 2538 __u32 curr_minor_hash; 2539 __u32 next_hash; 2540 }; 2541 2542 /* calculate the first block number of the group */ 2543 static inline ext4_fsblk_t 2544 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2545 { 2546 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2547 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2548 } 2549 2550 /* 2551 * Special error return code only used by dx_probe() and its callers. 2552 */ 2553 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2554 2555 /* htree levels for ext4 */ 2556 #define EXT4_HTREE_LEVEL_COMPAT 2 2557 #define EXT4_HTREE_LEVEL 3 2558 2559 static inline int ext4_dir_htree_level(struct super_block *sb) 2560 { 2561 return ext4_has_feature_largedir(sb) ? 2562 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2563 } 2564 2565 /* 2566 * Timeout and state flag for lazy initialization inode thread. 2567 */ 2568 #define EXT4_DEF_LI_WAIT_MULT 10 2569 #define EXT4_DEF_LI_MAX_START_DELAY 5 2570 #define EXT4_LAZYINIT_QUIT 0x0001 2571 #define EXT4_LAZYINIT_RUNNING 0x0002 2572 2573 /* 2574 * Lazy inode table initialization info 2575 */ 2576 struct ext4_lazy_init { 2577 unsigned long li_state; 2578 struct list_head li_request_list; 2579 struct mutex li_list_mtx; 2580 }; 2581 2582 enum ext4_li_mode { 2583 EXT4_LI_MODE_PREFETCH_BBITMAP, 2584 EXT4_LI_MODE_ITABLE, 2585 }; 2586 2587 struct ext4_li_request { 2588 struct super_block *lr_super; 2589 enum ext4_li_mode lr_mode; 2590 ext4_group_t lr_first_not_zeroed; 2591 ext4_group_t lr_next_group; 2592 struct list_head lr_request; 2593 unsigned long lr_next_sched; 2594 unsigned long lr_timeout; 2595 }; 2596 2597 struct ext4_features { 2598 struct kobject f_kobj; 2599 struct completion f_kobj_unregister; 2600 }; 2601 2602 /* 2603 * This structure will be used for multiple mount protection. It will be 2604 * written into the block number saved in the s_mmp_block field in the 2605 * superblock. Programs that check MMP should assume that if 2606 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2607 * to use the filesystem, regardless of how old the timestamp is. 2608 */ 2609 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2610 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2611 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2612 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2613 2614 struct mmp_struct { 2615 __le32 mmp_magic; /* Magic number for MMP */ 2616 __le32 mmp_seq; /* Sequence no. updated periodically */ 2617 2618 /* 2619 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2620 * purposes and do not affect the correctness of the algorithm 2621 */ 2622 __le64 mmp_time; /* Time last updated */ 2623 char mmp_nodename[64]; /* Node which last updated MMP block */ 2624 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2625 2626 /* 2627 * mmp_check_interval is used to verify if the MMP block has been 2628 * updated on the block device. The value is updated based on the 2629 * maximum time to write the MMP block during an update cycle. 2630 */ 2631 __le16 mmp_check_interval; 2632 2633 __le16 mmp_pad1; 2634 __le32 mmp_pad2[226]; 2635 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2636 }; 2637 2638 /* arguments passed to the mmp thread */ 2639 struct mmpd_data { 2640 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2641 struct super_block *sb; /* super block of the fs */ 2642 }; 2643 2644 /* 2645 * Check interval multiplier 2646 * The MMP block is written every update interval and initially checked every 2647 * update interval x the multiplier (the value is then adapted based on the 2648 * write latency). The reason is that writes can be delayed under load and we 2649 * don't want readers to incorrectly assume that the filesystem is no longer 2650 * in use. 2651 */ 2652 #define EXT4_MMP_CHECK_MULT 2UL 2653 2654 /* 2655 * Minimum interval for MMP checking in seconds. 2656 */ 2657 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2658 2659 /* 2660 * Maximum interval for MMP checking in seconds. 2661 */ 2662 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2663 2664 /* 2665 * Function prototypes 2666 */ 2667 2668 /* 2669 * Ok, these declarations are also in <linux/kernel.h> but none of the 2670 * ext4 source programs needs to include it so they are duplicated here. 2671 */ 2672 # define NORET_TYPE /**/ 2673 # define ATTRIB_NORET __attribute__((noreturn)) 2674 # define NORET_AND noreturn, 2675 2676 /* bitmap.c */ 2677 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2678 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2679 struct ext4_group_desc *gdp, 2680 struct buffer_head *bh, int sz); 2681 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2682 struct ext4_group_desc *gdp, 2683 struct buffer_head *bh, int sz); 2684 void ext4_block_bitmap_csum_set(struct super_block *sb, 2685 struct ext4_group_desc *gdp, 2686 struct buffer_head *bh); 2687 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2688 struct ext4_group_desc *gdp, 2689 struct buffer_head *bh); 2690 2691 /* balloc.c */ 2692 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2693 ext4_fsblk_t blocknr, 2694 ext4_group_t *blockgrpp, 2695 ext4_grpblk_t *offsetp); 2696 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2697 ext4_fsblk_t block); 2698 2699 extern unsigned int ext4_block_group(struct super_block *sb, 2700 ext4_fsblk_t blocknr); 2701 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2702 ext4_fsblk_t blocknr); 2703 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2704 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2705 ext4_group_t group); 2706 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2707 ext4_fsblk_t goal, 2708 unsigned int flags, 2709 unsigned long *count, 2710 int *errp); 2711 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2712 s64 nclusters, unsigned int flags); 2713 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2714 extern void ext4_check_blocks_bitmap(struct super_block *); 2715 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2716 ext4_group_t block_group, 2717 struct buffer_head ** bh); 2718 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2719 2720 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2721 ext4_group_t block_group, 2722 bool ignore_locked); 2723 extern int ext4_wait_block_bitmap(struct super_block *sb, 2724 ext4_group_t block_group, 2725 struct buffer_head *bh); 2726 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2727 ext4_group_t block_group); 2728 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2729 ext4_group_t block_group, 2730 struct ext4_group_desc *gdp); 2731 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2732 2733 #if IS_ENABLED(CONFIG_UNICODE) 2734 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2735 const struct qstr *iname, 2736 struct ext4_filename *fname); 2737 #endif 2738 2739 /* ext4 encryption related stuff goes here crypto.c */ 2740 #ifdef CONFIG_FS_ENCRYPTION 2741 extern const struct fscrypt_operations ext4_cryptops; 2742 2743 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2744 int lookup, struct ext4_filename *fname); 2745 2746 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2747 struct ext4_filename *fname); 2748 2749 void ext4_fname_free_filename(struct ext4_filename *fname); 2750 2751 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2752 2753 #else /* !CONFIG_FS_ENCRYPTION */ 2754 static inline int ext4_fname_setup_filename(struct inode *dir, 2755 const struct qstr *iname, 2756 int lookup, 2757 struct ext4_filename *fname) 2758 { 2759 int err = 0; 2760 fname->usr_fname = iname; 2761 fname->disk_name.name = (unsigned char *) iname->name; 2762 fname->disk_name.len = iname->len; 2763 2764 #if IS_ENABLED(CONFIG_UNICODE) 2765 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2766 #endif 2767 2768 return err; 2769 } 2770 2771 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2772 struct dentry *dentry, 2773 struct ext4_filename *fname) 2774 { 2775 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2776 } 2777 2778 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2779 { 2780 #if IS_ENABLED(CONFIG_UNICODE) 2781 kfree(fname->cf_name.name); 2782 fname->cf_name.name = NULL; 2783 #endif 2784 } 2785 2786 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2787 void __user *arg) 2788 { 2789 return -EOPNOTSUPP; 2790 } 2791 #endif /* !CONFIG_FS_ENCRYPTION */ 2792 2793 /* dir.c */ 2794 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2795 struct file *, 2796 struct ext4_dir_entry_2 *, 2797 struct buffer_head *, char *, int, 2798 unsigned int); 2799 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2800 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2801 (de), (bh), (buf), (size), (offset))) 2802 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2803 __u32 minor_hash, 2804 struct ext4_dir_entry_2 *dirent, 2805 struct fscrypt_str *ent_name); 2806 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2807 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2808 struct buffer_head *bh, 2809 void *buf, int buf_size, 2810 struct ext4_filename *fname, 2811 struct ext4_dir_entry_2 **dest_de); 2812 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2813 struct ext4_dir_entry_2 *de, 2814 int buf_size, 2815 struct ext4_filename *fname); 2816 static inline void ext4_update_dx_flag(struct inode *inode) 2817 { 2818 if (!ext4_has_feature_dir_index(inode->i_sb) && 2819 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2820 /* ext4_iget() should have caught this... */ 2821 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2822 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2823 } 2824 } 2825 static const unsigned char ext4_filetype_table[] = { 2826 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2827 }; 2828 2829 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2830 { 2831 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2832 return DT_UNKNOWN; 2833 2834 return ext4_filetype_table[filetype]; 2835 } 2836 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2837 void *buf, int buf_size); 2838 2839 /* fsync.c */ 2840 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2841 2842 /* hash.c */ 2843 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2844 struct dx_hash_info *hinfo); 2845 2846 /* ialloc.c */ 2847 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2848 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2849 struct inode *, umode_t, 2850 const struct qstr *qstr, __u32 goal, 2851 uid_t *owner, __u32 i_flags, 2852 int handle_type, unsigned int line_no, 2853 int nblocks); 2854 2855 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2856 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2857 (goal), (owner), i_flags, 0, 0, 0) 2858 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2859 type, nblocks) \ 2860 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2861 0, (type), __LINE__, (nblocks)) 2862 2863 2864 extern void ext4_free_inode(handle_t *, struct inode *); 2865 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2866 extern unsigned long ext4_count_free_inodes(struct super_block *); 2867 extern unsigned long ext4_count_dirs(struct super_block *); 2868 extern void ext4_check_inodes_bitmap(struct super_block *); 2869 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2870 extern int ext4_init_inode_table(struct super_block *sb, 2871 ext4_group_t group, int barrier); 2872 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2873 2874 /* fast_commit.c */ 2875 int ext4_fc_info_show(struct seq_file *seq, void *v); 2876 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2877 void ext4_fc_init_inode(struct inode *inode); 2878 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2879 ext4_lblk_t end); 2880 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2881 struct dentry *dentry); 2882 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2883 struct dentry *dentry); 2884 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2885 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2886 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2887 struct dentry *dentry); 2888 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2889 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2890 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2891 void ext4_fc_start_update(struct inode *inode); 2892 void ext4_fc_stop_update(struct inode *inode); 2893 void ext4_fc_del(struct inode *inode); 2894 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2895 void ext4_fc_replay_cleanup(struct super_block *sb); 2896 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2897 int __init ext4_fc_init_dentry_cache(void); 2898 void ext4_fc_destroy_dentry_cache(void); 2899 int ext4_fc_record_regions(struct super_block *sb, int ino, 2900 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2901 int len, int replay); 2902 2903 /* mballoc.c */ 2904 extern const struct seq_operations ext4_mb_seq_groups_ops; 2905 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2906 extern long ext4_mb_stats; 2907 extern long ext4_mb_max_to_scan; 2908 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2909 extern int ext4_mb_init(struct super_block *); 2910 extern int ext4_mb_release(struct super_block *); 2911 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2912 struct ext4_allocation_request *, int *); 2913 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2914 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2915 extern int __init ext4_init_mballoc(void); 2916 extern void ext4_exit_mballoc(void); 2917 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2918 ext4_group_t group, 2919 unsigned int nr, int *cnt); 2920 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2921 unsigned int nr); 2922 2923 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2924 struct buffer_head *bh, ext4_fsblk_t block, 2925 unsigned long count, int flags); 2926 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2927 ext4_group_t ngroups); 2928 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2929 ext4_group_t i, struct ext4_group_desc *desc); 2930 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2931 ext4_fsblk_t block, unsigned long count); 2932 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2933 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2934 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2935 int len, int state); 2936 2937 /* inode.c */ 2938 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2939 struct ext4_inode_info *ei); 2940 int ext4_inode_is_fast_symlink(struct inode *inode); 2941 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2942 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2943 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2944 bool wait, struct buffer_head **bhs); 2945 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2946 struct buffer_head *bh_result, int create); 2947 int ext4_get_block(struct inode *inode, sector_t iblock, 2948 struct buffer_head *bh_result, int create); 2949 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2950 struct buffer_head *bh, int create); 2951 int ext4_walk_page_buffers(handle_t *handle, 2952 struct inode *inode, 2953 struct buffer_head *head, 2954 unsigned from, 2955 unsigned to, 2956 int *partial, 2957 int (*fn)(handle_t *handle, struct inode *inode, 2958 struct buffer_head *bh)); 2959 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2960 struct buffer_head *bh); 2961 #define FALL_BACK_TO_NONDELALLOC 1 2962 #define CONVERT_INLINE_DATA 2 2963 2964 typedef enum { 2965 EXT4_IGET_NORMAL = 0, 2966 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2967 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2968 EXT4_IGET_BAD = 0x0004 /* Allow to iget a bad inode */ 2969 } ext4_iget_flags; 2970 2971 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2972 ext4_iget_flags flags, const char *function, 2973 unsigned int line); 2974 2975 #define ext4_iget(sb, ino, flags) \ 2976 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2977 2978 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2979 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2980 struct iattr *); 2981 extern u32 ext4_dio_alignment(struct inode *inode); 2982 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2983 struct kstat *, u32, unsigned int); 2984 extern void ext4_evict_inode(struct inode *); 2985 extern void ext4_clear_inode(struct inode *); 2986 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2987 struct kstat *, u32, unsigned int); 2988 extern int ext4_sync_inode(handle_t *, struct inode *); 2989 extern void ext4_dirty_inode(struct inode *, int); 2990 extern int ext4_change_inode_journal_flag(struct inode *, int); 2991 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2992 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2993 struct ext4_iloc *iloc); 2994 extern int ext4_inode_attach_jinode(struct inode *inode); 2995 extern int ext4_can_truncate(struct inode *inode); 2996 extern int ext4_truncate(struct inode *); 2997 extern int ext4_break_layouts(struct inode *); 2998 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2999 extern void ext4_set_inode_flags(struct inode *, bool init); 3000 extern int ext4_alloc_da_blocks(struct inode *inode); 3001 extern void ext4_set_aops(struct inode *inode); 3002 extern int ext4_writepage_trans_blocks(struct inode *); 3003 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3004 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3005 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3006 loff_t lstart, loff_t lend); 3007 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3008 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3009 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3010 extern void ext4_da_release_space(struct inode *inode, int to_free); 3011 extern void ext4_da_update_reserve_space(struct inode *inode, 3012 int used, int quota_claim); 3013 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3014 ext4_fsblk_t pblk, ext4_lblk_t len); 3015 3016 /* indirect.c */ 3017 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3018 struct ext4_map_blocks *map, int flags); 3019 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3020 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3021 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3022 ext4_lblk_t start, ext4_lblk_t end); 3023 3024 /* ioctl.c */ 3025 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3026 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3027 int ext4_fileattr_set(struct mnt_idmap *idmap, 3028 struct dentry *dentry, struct fileattr *fa); 3029 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3030 extern void ext4_reset_inode_seed(struct inode *inode); 3031 int ext4_update_overhead(struct super_block *sb, bool force); 3032 3033 /* migrate.c */ 3034 extern int ext4_ext_migrate(struct inode *); 3035 extern int ext4_ind_migrate(struct inode *inode); 3036 3037 /* namei.c */ 3038 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3039 struct inode *inode); 3040 extern int ext4_dirblock_csum_verify(struct inode *inode, 3041 struct buffer_head *bh); 3042 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3043 __u32 start_minor_hash, __u32 *next_hash); 3044 extern int ext4_search_dir(struct buffer_head *bh, 3045 char *search_buf, 3046 int buf_size, 3047 struct inode *dir, 3048 struct ext4_filename *fname, 3049 unsigned int offset, 3050 struct ext4_dir_entry_2 **res_dir); 3051 extern int ext4_generic_delete_entry(struct inode *dir, 3052 struct ext4_dir_entry_2 *de_del, 3053 struct buffer_head *bh, 3054 void *entry_buf, 3055 int buf_size, 3056 int csum_size); 3057 extern bool ext4_empty_dir(struct inode *inode); 3058 3059 /* resize.c */ 3060 extern void ext4_kvfree_array_rcu(void *to_free); 3061 extern int ext4_group_add(struct super_block *sb, 3062 struct ext4_new_group_data *input); 3063 extern int ext4_group_extend(struct super_block *sb, 3064 struct ext4_super_block *es, 3065 ext4_fsblk_t n_blocks_count); 3066 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3067 extern unsigned int ext4_list_backups(struct super_block *sb, 3068 unsigned int *three, unsigned int *five, 3069 unsigned int *seven); 3070 3071 /* super.c */ 3072 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3073 sector_t block, blk_opf_t op_flags); 3074 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3075 sector_t block); 3076 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3077 bh_end_io_t *end_io); 3078 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3079 bh_end_io_t *end_io); 3080 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3081 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3082 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3083 extern int ext4_calculate_overhead(struct super_block *sb); 3084 extern __le32 ext4_superblock_csum(struct super_block *sb, 3085 struct ext4_super_block *es); 3086 extern void ext4_superblock_csum_set(struct super_block *sb); 3087 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3088 ext4_group_t ngroup); 3089 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3090 char nbuf[16]); 3091 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3092 ext4_group_t block_group, 3093 unsigned int flags); 3094 3095 extern __printf(7, 8) 3096 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3097 int, __u64, const char *, ...); 3098 extern __printf(6, 7) 3099 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3100 ext4_fsblk_t, int, const char *, ...); 3101 extern __printf(5, 6) 3102 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3103 const char *, ...); 3104 extern void __ext4_std_error(struct super_block *, const char *, 3105 unsigned int, int); 3106 extern __printf(4, 5) 3107 void __ext4_warning(struct super_block *, const char *, unsigned int, 3108 const char *, ...); 3109 extern __printf(4, 5) 3110 void __ext4_warning_inode(const struct inode *inode, const char *function, 3111 unsigned int line, const char *fmt, ...); 3112 extern __printf(3, 4) 3113 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3114 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3115 const char *, unsigned int, const char *); 3116 extern __printf(7, 8) 3117 void __ext4_grp_locked_error(const char *, unsigned int, 3118 struct super_block *, ext4_group_t, 3119 unsigned long, ext4_fsblk_t, 3120 const char *, ...); 3121 3122 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3123 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3124 3125 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3126 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3127 3128 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3129 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3130 (fmt), ## a) 3131 3132 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3133 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3134 3135 #define ext4_abort(sb, err, fmt, a...) \ 3136 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3137 3138 #ifdef CONFIG_PRINTK 3139 3140 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3141 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3142 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3143 __ext4_error_inode((inode), (func), (line), (block), \ 3144 (err), (fmt), ##__VA_ARGS__) 3145 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3146 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3147 #define ext4_error(sb, fmt, ...) \ 3148 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3149 ##__VA_ARGS__) 3150 #define ext4_error_err(sb, err, fmt, ...) \ 3151 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3152 ##__VA_ARGS__) 3153 #define ext4_warning(sb, fmt, ...) \ 3154 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3155 #define ext4_warning_inode(inode, fmt, ...) \ 3156 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3157 #define ext4_msg(sb, level, fmt, ...) \ 3158 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3159 #define dump_mmp_msg(sb, mmp, msg) \ 3160 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3161 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3162 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3163 fmt, ##__VA_ARGS__) 3164 3165 #else 3166 3167 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3168 do { \ 3169 no_printk(fmt, ##__VA_ARGS__); \ 3170 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3171 } while (0) 3172 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3173 do { \ 3174 no_printk(fmt, ##__VA_ARGS__); \ 3175 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3176 } while (0) 3177 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3178 do { \ 3179 no_printk(fmt, ##__VA_ARGS__); \ 3180 __ext4_error_file(file, "", 0, block, " "); \ 3181 } while (0) 3182 #define ext4_error(sb, fmt, ...) \ 3183 do { \ 3184 no_printk(fmt, ##__VA_ARGS__); \ 3185 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3186 } while (0) 3187 #define ext4_error_err(sb, err, fmt, ...) \ 3188 do { \ 3189 no_printk(fmt, ##__VA_ARGS__); \ 3190 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3191 } while (0) 3192 #define ext4_warning(sb, fmt, ...) \ 3193 do { \ 3194 no_printk(fmt, ##__VA_ARGS__); \ 3195 __ext4_warning(sb, "", 0, " "); \ 3196 } while (0) 3197 #define ext4_warning_inode(inode, fmt, ...) \ 3198 do { \ 3199 no_printk(fmt, ##__VA_ARGS__); \ 3200 __ext4_warning_inode(inode, "", 0, " "); \ 3201 } while (0) 3202 #define ext4_msg(sb, level, fmt, ...) \ 3203 do { \ 3204 no_printk(fmt, ##__VA_ARGS__); \ 3205 __ext4_msg(sb, "", " "); \ 3206 } while (0) 3207 #define dump_mmp_msg(sb, mmp, msg) \ 3208 __dump_mmp_msg(sb, mmp, "", 0, "") 3209 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3210 do { \ 3211 no_printk(fmt, ##__VA_ARGS__); \ 3212 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3213 } while (0) 3214 3215 #endif 3216 3217 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3218 struct ext4_group_desc *bg); 3219 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3220 struct ext4_group_desc *bg); 3221 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3222 struct ext4_group_desc *bg); 3223 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3224 struct ext4_group_desc *bg); 3225 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3226 struct ext4_group_desc *bg); 3227 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3228 struct ext4_group_desc *bg); 3229 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3230 struct ext4_group_desc *bg); 3231 extern void ext4_block_bitmap_set(struct super_block *sb, 3232 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3233 extern void ext4_inode_bitmap_set(struct super_block *sb, 3234 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3235 extern void ext4_inode_table_set(struct super_block *sb, 3236 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3237 extern void ext4_free_group_clusters_set(struct super_block *sb, 3238 struct ext4_group_desc *bg, 3239 __u32 count); 3240 extern void ext4_free_inodes_set(struct super_block *sb, 3241 struct ext4_group_desc *bg, __u32 count); 3242 extern void ext4_used_dirs_set(struct super_block *sb, 3243 struct ext4_group_desc *bg, __u32 count); 3244 extern void ext4_itable_unused_set(struct super_block *sb, 3245 struct ext4_group_desc *bg, __u32 count); 3246 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3247 struct ext4_group_desc *gdp); 3248 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3249 struct ext4_group_desc *gdp); 3250 extern int ext4_register_li_request(struct super_block *sb, 3251 ext4_group_t first_not_zeroed); 3252 3253 static inline int ext4_has_metadata_csum(struct super_block *sb) 3254 { 3255 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3256 !EXT4_SB(sb)->s_chksum_driver); 3257 3258 return ext4_has_feature_metadata_csum(sb) && 3259 (EXT4_SB(sb)->s_chksum_driver != NULL); 3260 } 3261 3262 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3263 { 3264 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3265 } 3266 3267 #define ext4_read_incompat_64bit_val(es, name) \ 3268 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3269 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3270 le32_to_cpu(es->name##_lo)) 3271 3272 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3273 { 3274 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3275 } 3276 3277 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3278 { 3279 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3280 } 3281 3282 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3283 { 3284 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3285 } 3286 3287 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3288 ext4_fsblk_t blk) 3289 { 3290 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3291 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3292 } 3293 3294 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3295 ext4_fsblk_t blk) 3296 { 3297 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3298 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3299 } 3300 3301 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3302 ext4_fsblk_t blk) 3303 { 3304 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3305 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3306 } 3307 3308 static inline loff_t ext4_isize(struct super_block *sb, 3309 struct ext4_inode *raw_inode) 3310 { 3311 if (ext4_has_feature_largedir(sb) || 3312 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3313 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3314 le32_to_cpu(raw_inode->i_size_lo); 3315 3316 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3317 } 3318 3319 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3320 { 3321 raw_inode->i_size_lo = cpu_to_le32(i_size); 3322 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3323 } 3324 3325 static inline 3326 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3327 ext4_group_t group) 3328 { 3329 struct ext4_group_info **grp_info; 3330 long indexv, indexh; 3331 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3332 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3333 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3334 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3335 return grp_info[indexh]; 3336 } 3337 3338 /* 3339 * Reading s_groups_count requires using smp_rmb() afterwards. See 3340 * the locking protocol documented in the comments of ext4_group_add() 3341 * in resize.c 3342 */ 3343 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3344 { 3345 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3346 3347 smp_rmb(); 3348 return ngroups; 3349 } 3350 3351 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3352 ext4_group_t block_group) 3353 { 3354 return block_group >> sbi->s_log_groups_per_flex; 3355 } 3356 3357 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3358 { 3359 return 1 << sbi->s_log_groups_per_flex; 3360 } 3361 3362 #define ext4_std_error(sb, errno) \ 3363 do { \ 3364 if ((errno)) \ 3365 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3366 } while (0) 3367 3368 #ifdef CONFIG_SMP 3369 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3370 * counters. So we need to make sure we have free clusters more 3371 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3372 */ 3373 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3374 #else 3375 #define EXT4_FREECLUSTERS_WATERMARK 0 3376 #endif 3377 3378 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3379 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3380 { 3381 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3382 !inode_is_locked(inode)); 3383 down_write(&EXT4_I(inode)->i_data_sem); 3384 if (newsize > EXT4_I(inode)->i_disksize) 3385 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3386 up_write(&EXT4_I(inode)->i_data_sem); 3387 } 3388 3389 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3390 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3391 { 3392 int changed = 0; 3393 3394 if (newsize > inode->i_size) { 3395 i_size_write(inode, newsize); 3396 changed = 1; 3397 } 3398 if (newsize > EXT4_I(inode)->i_disksize) { 3399 ext4_update_i_disksize(inode, newsize); 3400 changed |= 2; 3401 } 3402 return changed; 3403 } 3404 3405 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3406 loff_t len); 3407 3408 struct ext4_group_info { 3409 unsigned long bb_state; 3410 #ifdef AGGRESSIVE_CHECK 3411 unsigned long bb_check_counter; 3412 #endif 3413 struct rb_root bb_free_root; 3414 ext4_grpblk_t bb_first_free; /* first free block */ 3415 ext4_grpblk_t bb_free; /* total free blocks */ 3416 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3417 int bb_avg_fragment_size_order; /* order of average 3418 fragment in BG */ 3419 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3420 ext4_group_t bb_group; /* Group number */ 3421 struct list_head bb_prealloc_list; 3422 #ifdef DOUBLE_CHECK 3423 void *bb_bitmap; 3424 #endif 3425 struct rw_semaphore alloc_sem; 3426 struct list_head bb_avg_fragment_size_node; 3427 struct list_head bb_largest_free_order_node; 3428 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3429 * regions, index is order. 3430 * bb_counters[3] = 5 means 3431 * 5 free 8-block regions. */ 3432 }; 3433 3434 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3435 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3436 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3437 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3438 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3439 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3440 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3441 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3442 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3443 3444 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3445 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3446 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3447 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3448 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3449 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3450 3451 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3452 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3453 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3454 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3455 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3456 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3457 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3458 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3459 3460 #define EXT4_MAX_CONTENTION 8 3461 #define EXT4_CONTENTION_THRESHOLD 2 3462 3463 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3464 ext4_group_t group) 3465 { 3466 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3467 } 3468 3469 /* 3470 * Returns true if the filesystem is busy enough that attempts to 3471 * access the block group locks has run into contention. 3472 */ 3473 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3474 { 3475 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3476 } 3477 3478 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3479 { 3480 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3481 if (spin_trylock(lock)) 3482 /* 3483 * We're able to grab the lock right away, so drop the 3484 * lock contention counter. 3485 */ 3486 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3487 else { 3488 /* 3489 * The lock is busy, so bump the contention counter, 3490 * and then wait on the spin lock. 3491 */ 3492 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3493 EXT4_MAX_CONTENTION); 3494 spin_lock(lock); 3495 } 3496 } 3497 3498 static inline void ext4_unlock_group(struct super_block *sb, 3499 ext4_group_t group) 3500 { 3501 spin_unlock(ext4_group_lock_ptr(sb, group)); 3502 } 3503 3504 #ifdef CONFIG_QUOTA 3505 static inline bool ext4_quota_capable(struct super_block *sb) 3506 { 3507 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3508 } 3509 3510 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3511 { 3512 struct ext4_sb_info *sbi = EXT4_SB(sb); 3513 3514 return (ext4_has_feature_quota(sb) || 3515 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3516 } 3517 int ext4_enable_quotas(struct super_block *sb); 3518 #endif 3519 3520 /* 3521 * Block validity checking 3522 */ 3523 #define ext4_check_indirect_blockref(inode, bh) \ 3524 ext4_check_blockref(__func__, __LINE__, inode, \ 3525 (__le32 *)(bh)->b_data, \ 3526 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3527 3528 #define ext4_ind_check_inode(inode) \ 3529 ext4_check_blockref(__func__, __LINE__, inode, \ 3530 EXT4_I(inode)->i_data, \ 3531 EXT4_NDIR_BLOCKS) 3532 3533 /* 3534 * Inodes and files operations 3535 */ 3536 3537 /* dir.c */ 3538 extern const struct file_operations ext4_dir_operations; 3539 3540 /* file.c */ 3541 extern const struct inode_operations ext4_file_inode_operations; 3542 extern const struct file_operations ext4_file_operations; 3543 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3544 3545 /* inline.c */ 3546 extern int ext4_get_max_inline_size(struct inode *inode); 3547 extern int ext4_find_inline_data_nolock(struct inode *inode); 3548 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3549 unsigned int len); 3550 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3551 3552 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3553 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3554 struct inode *inode, 3555 loff_t pos, unsigned len, 3556 struct page **pagep); 3557 extern int ext4_write_inline_data_end(struct inode *inode, 3558 loff_t pos, unsigned len, 3559 unsigned copied, 3560 struct page *page); 3561 extern struct buffer_head * 3562 ext4_journalled_write_inline_data(struct inode *inode, 3563 unsigned len, 3564 struct page *page); 3565 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3566 struct inode *inode, 3567 loff_t pos, unsigned len, 3568 struct page **pagep, 3569 void **fsdata); 3570 extern int ext4_try_add_inline_entry(handle_t *handle, 3571 struct ext4_filename *fname, 3572 struct inode *dir, struct inode *inode); 3573 extern int ext4_try_create_inline_dir(handle_t *handle, 3574 struct inode *parent, 3575 struct inode *inode); 3576 extern int ext4_read_inline_dir(struct file *filp, 3577 struct dir_context *ctx, 3578 int *has_inline_data); 3579 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3580 struct inode *dir, ext4_lblk_t block, 3581 struct dx_hash_info *hinfo, 3582 __u32 start_hash, __u32 start_minor_hash, 3583 int *has_inline_data); 3584 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3585 struct ext4_filename *fname, 3586 struct ext4_dir_entry_2 **res_dir, 3587 int *has_inline_data); 3588 extern int ext4_delete_inline_entry(handle_t *handle, 3589 struct inode *dir, 3590 struct ext4_dir_entry_2 *de_del, 3591 struct buffer_head *bh, 3592 int *has_inline_data); 3593 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3594 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3595 struct ext4_dir_entry_2 **parent_de, 3596 int *retval); 3597 extern void *ext4_read_inline_link(struct inode *inode); 3598 3599 struct iomap; 3600 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3601 3602 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3603 3604 extern int ext4_convert_inline_data(struct inode *inode); 3605 3606 static inline int ext4_has_inline_data(struct inode *inode) 3607 { 3608 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3609 EXT4_I(inode)->i_inline_off; 3610 } 3611 3612 /* namei.c */ 3613 extern const struct inode_operations ext4_dir_inode_operations; 3614 extern const struct inode_operations ext4_special_inode_operations; 3615 extern struct dentry *ext4_get_parent(struct dentry *child); 3616 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3617 struct ext4_dir_entry_2 *de, 3618 int blocksize, int csum_size, 3619 unsigned int parent_ino, int dotdot_real_len); 3620 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3621 unsigned int blocksize); 3622 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3623 struct buffer_head *bh); 3624 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3625 struct inode *inode, struct dentry *dentry); 3626 extern int __ext4_link(struct inode *dir, struct inode *inode, 3627 struct dentry *dentry); 3628 3629 #define S_SHIFT 12 3630 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3631 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3632 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3633 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3634 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3635 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3636 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3637 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3638 }; 3639 3640 static inline void ext4_set_de_type(struct super_block *sb, 3641 struct ext4_dir_entry_2 *de, 3642 umode_t mode) { 3643 if (ext4_has_feature_filetype(sb)) 3644 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3645 } 3646 3647 /* readpages.c */ 3648 extern int ext4_mpage_readpages(struct inode *inode, 3649 struct readahead_control *rac, struct page *page); 3650 extern int __init ext4_init_post_read_processing(void); 3651 extern void ext4_exit_post_read_processing(void); 3652 3653 /* symlink.c */ 3654 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3655 extern const struct inode_operations ext4_symlink_inode_operations; 3656 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3657 3658 /* sysfs.c */ 3659 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3660 extern int ext4_register_sysfs(struct super_block *sb); 3661 extern void ext4_unregister_sysfs(struct super_block *sb); 3662 extern int __init ext4_init_sysfs(void); 3663 extern void ext4_exit_sysfs(void); 3664 3665 /* block_validity */ 3666 extern void ext4_release_system_zone(struct super_block *sb); 3667 extern int ext4_setup_system_zone(struct super_block *sb); 3668 extern int __init ext4_init_system_zone(void); 3669 extern void ext4_exit_system_zone(void); 3670 extern int ext4_inode_block_valid(struct inode *inode, 3671 ext4_fsblk_t start_blk, 3672 unsigned int count); 3673 extern int ext4_check_blockref(const char *, unsigned int, 3674 struct inode *, __le32 *, unsigned int); 3675 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3676 ext4_fsblk_t start_blk, unsigned int count); 3677 3678 3679 /* extents.c */ 3680 struct ext4_ext_path; 3681 struct ext4_extent; 3682 3683 /* 3684 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3685 * __le32. 3686 */ 3687 #define EXT_MAX_BLOCKS 0xffffffff 3688 3689 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3690 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3691 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3692 struct ext4_map_blocks *map, int flags); 3693 extern int ext4_ext_truncate(handle_t *, struct inode *); 3694 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3695 ext4_lblk_t end); 3696 extern void ext4_ext_init(struct super_block *); 3697 extern void ext4_ext_release(struct super_block *); 3698 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3699 loff_t len); 3700 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3701 loff_t offset, ssize_t len); 3702 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3703 ext4_io_end_t *io_end); 3704 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3705 struct ext4_map_blocks *map, int flags); 3706 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3707 int num, 3708 struct ext4_ext_path *path); 3709 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3710 struct ext4_ext_path **, 3711 struct ext4_extent *, int); 3712 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3713 struct ext4_ext_path **, 3714 int flags); 3715 extern void ext4_free_ext_path(struct ext4_ext_path *); 3716 extern int ext4_ext_check_inode(struct inode *inode); 3717 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3718 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3719 __u64 start, __u64 len); 3720 extern int ext4_get_es_cache(struct inode *inode, 3721 struct fiemap_extent_info *fieinfo, 3722 __u64 start, __u64 len); 3723 extern int ext4_ext_precache(struct inode *inode); 3724 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3725 struct inode *inode2, ext4_lblk_t lblk1, 3726 ext4_lblk_t lblk2, ext4_lblk_t count, 3727 int mark_unwritten,int *err); 3728 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3729 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3730 int check_cred, int restart_cred, 3731 int revoke_cred); 3732 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3733 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3734 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3735 int len, int unwritten, ext4_fsblk_t pblk); 3736 extern int ext4_ext_clear_bb(struct inode *inode); 3737 3738 3739 /* move_extent.c */ 3740 extern void ext4_double_down_write_data_sem(struct inode *first, 3741 struct inode *second); 3742 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3743 struct inode *donor_inode); 3744 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3745 __u64 start_orig, __u64 start_donor, 3746 __u64 len, __u64 *moved_len); 3747 3748 /* page-io.c */ 3749 extern int __init ext4_init_pageio(void); 3750 extern void ext4_exit_pageio(void); 3751 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3752 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3753 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3754 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3755 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3756 struct writeback_control *wbc); 3757 extern void ext4_end_io_rsv_work(struct work_struct *work); 3758 extern void ext4_io_submit(struct ext4_io_submit *io); 3759 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3760 struct page *page, 3761 int len); 3762 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3763 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3764 3765 /* mmp.c */ 3766 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3767 3768 /* mmp.c */ 3769 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3770 3771 /* verity.c */ 3772 extern const struct fsverity_operations ext4_verityops; 3773 3774 /* orphan.c */ 3775 extern int ext4_orphan_add(handle_t *, struct inode *); 3776 extern int ext4_orphan_del(handle_t *, struct inode *); 3777 extern void ext4_orphan_cleanup(struct super_block *sb, 3778 struct ext4_super_block *es); 3779 extern void ext4_release_orphan_info(struct super_block *sb); 3780 extern int ext4_init_orphan_info(struct super_block *sb); 3781 extern int ext4_orphan_file_empty(struct super_block *sb); 3782 extern void ext4_orphan_file_block_trigger( 3783 struct jbd2_buffer_trigger_type *triggers, 3784 struct buffer_head *bh, 3785 void *data, size_t size); 3786 3787 /* 3788 * Add new method to test whether block and inode bitmaps are properly 3789 * initialized. With uninit_bg reading the block from disk is not enough 3790 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3791 */ 3792 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3793 3794 static inline int bitmap_uptodate(struct buffer_head *bh) 3795 { 3796 return (buffer_uptodate(bh) && 3797 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3798 } 3799 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3800 { 3801 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3802 } 3803 3804 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3805 3806 /* For ioend & aio unwritten conversion wait queues */ 3807 #define EXT4_WQ_HASH_SZ 37 3808 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3809 EXT4_WQ_HASH_SZ]) 3810 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3811 3812 extern int ext4_resize_begin(struct super_block *sb); 3813 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3814 3815 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3816 struct ext4_io_end *io_end) 3817 { 3818 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3819 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3820 atomic_inc(&EXT4_I(inode)->i_unwritten); 3821 } 3822 } 3823 3824 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3825 { 3826 struct inode *inode = io_end->inode; 3827 3828 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3829 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3830 /* Wake up anyone waiting on unwritten extent conversion */ 3831 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3832 wake_up_all(ext4_ioend_wq(inode)); 3833 } 3834 } 3835 3836 extern const struct iomap_ops ext4_iomap_ops; 3837 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3838 extern const struct iomap_ops ext4_iomap_report_ops; 3839 3840 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3841 { 3842 /* 3843 * If the buffer has the write error flag, we have failed 3844 * to write out data in the block. In this case, we don't 3845 * have to read the block because we may read the old data 3846 * successfully. 3847 */ 3848 if (buffer_write_io_error(bh)) 3849 set_buffer_uptodate(bh); 3850 return buffer_uptodate(bh); 3851 } 3852 3853 #endif /* __KERNEL__ */ 3854 3855 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3856 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3857 3858 #endif /* _EXT4_H */ 3859