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 /* Perform linear traversal for one group */ 171 #define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000 172 struct ext4_allocation_request { 173 /* target inode for block we're allocating */ 174 struct inode *inode; 175 /* how many blocks we want to allocate */ 176 unsigned int len; 177 /* logical block in target inode */ 178 ext4_lblk_t logical; 179 /* the closest logical allocated block to the left */ 180 ext4_lblk_t lleft; 181 /* the closest logical allocated block to the right */ 182 ext4_lblk_t lright; 183 /* phys. target (a hint) */ 184 ext4_fsblk_t goal; 185 /* phys. block for the closest logical allocated block to the left */ 186 ext4_fsblk_t pleft; 187 /* phys. block for the closest logical allocated block to the right */ 188 ext4_fsblk_t pright; 189 /* flags. see above EXT4_MB_HINT_* */ 190 unsigned int flags; 191 }; 192 193 /* 194 * Logical to physical block mapping, used by ext4_map_blocks() 195 * 196 * This structure is used to pass requests into ext4_map_blocks() as 197 * well as to store the information returned by ext4_map_blocks(). It 198 * takes less room on the stack than a struct buffer_head. 199 */ 200 #define EXT4_MAP_NEW BIT(BH_New) 201 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 202 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 203 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 204 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 205 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 206 207 struct ext4_map_blocks { 208 ext4_fsblk_t m_pblk; 209 ext4_lblk_t m_lblk; 210 unsigned int m_len; 211 unsigned int m_flags; 212 }; 213 214 /* 215 * Block validity checking, system zone rbtree. 216 */ 217 struct ext4_system_blocks { 218 struct rb_root root; 219 struct rcu_head rcu; 220 }; 221 222 /* 223 * Flags for ext4_io_end->flags 224 */ 225 #define EXT4_IO_END_UNWRITTEN 0x0001 226 227 struct ext4_io_end_vec { 228 struct list_head list; /* list of io_end_vec */ 229 loff_t offset; /* offset in the file */ 230 ssize_t size; /* size of the extent */ 231 }; 232 233 /* 234 * For converting unwritten extents on a work queue. 'handle' is used for 235 * buffered writeback. 236 */ 237 typedef struct ext4_io_end { 238 struct list_head list; /* per-file finished IO list */ 239 handle_t *handle; /* handle reserved for extent 240 * conversion */ 241 struct inode *inode; /* file being written to */ 242 struct bio *bio; /* Linked list of completed 243 * bios covering the extent */ 244 unsigned int flag; /* unwritten or not */ 245 refcount_t count; /* reference counter */ 246 struct list_head list_vec; /* list of ext4_io_end_vec */ 247 } ext4_io_end_t; 248 249 struct ext4_io_submit { 250 struct writeback_control *io_wbc; 251 struct bio *io_bio; 252 ext4_io_end_t *io_end; 253 sector_t io_next_block; 254 }; 255 256 /* 257 * Special inodes numbers 258 */ 259 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 260 #define EXT4_ROOT_INO 2 /* Root inode */ 261 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 262 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 263 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 264 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 265 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 266 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 267 268 /* First non-reserved inode for old ext4 filesystems */ 269 #define EXT4_GOOD_OLD_FIRST_INO 11 270 271 /* 272 * Maximal count of links to a file 273 */ 274 #define EXT4_LINK_MAX 65000 275 276 /* 277 * Macro-instructions used to manage several block sizes 278 */ 279 #define EXT4_MIN_BLOCK_SIZE 1024 280 #define EXT4_MAX_BLOCK_SIZE 65536 281 #define EXT4_MIN_BLOCK_LOG_SIZE 10 282 #define EXT4_MAX_BLOCK_LOG_SIZE 16 283 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 284 #ifdef __KERNEL__ 285 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 286 #else 287 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 288 #endif 289 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 290 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 291 EXT4_SB(s)->s_cluster_bits) 292 #ifdef __KERNEL__ 293 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 294 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 295 #else 296 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 297 #endif 298 #ifdef __KERNEL__ 299 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 300 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 301 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 302 #else 303 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 304 EXT4_GOOD_OLD_INODE_SIZE : \ 305 (s)->s_inode_size) 306 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 307 EXT4_GOOD_OLD_FIRST_INO : \ 308 (s)->s_first_ino) 309 #endif 310 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 311 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 312 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 313 blkbits)) 314 315 /* Translate a block number to a cluster number */ 316 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 317 /* Translate a cluster number to a block number */ 318 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 319 /* Translate # of blks to # of clusters */ 320 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 321 (sbi)->s_cluster_bits) 322 /* Mask out the low bits to get the starting block of the cluster */ 323 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 324 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 325 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 326 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 327 /* Fill in the low bits to get the last block of the cluster */ 328 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 329 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 330 /* Get the cluster offset */ 331 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 332 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 333 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 334 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 335 336 /* 337 * Structure of a blocks group descriptor 338 */ 339 struct ext4_group_desc 340 { 341 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 342 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 343 __le32 bg_inode_table_lo; /* Inodes table block */ 344 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 345 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 346 __le16 bg_used_dirs_count_lo; /* Directories count */ 347 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 348 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 349 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 350 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 351 __le16 bg_itable_unused_lo; /* Unused inodes count */ 352 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 353 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 354 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 355 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 356 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 357 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 358 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 359 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 360 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 361 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 362 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 363 __u32 bg_reserved; 364 }; 365 366 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 367 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 368 sizeof(__le16)) 369 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 370 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 371 sizeof(__le16)) 372 373 /* 374 * Structure of a flex block group info 375 */ 376 377 struct flex_groups { 378 atomic64_t free_clusters; 379 atomic_t free_inodes; 380 atomic_t used_dirs; 381 }; 382 383 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 384 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 385 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 386 387 /* 388 * Macro-instructions used to manage group descriptors 389 */ 390 #define EXT4_MIN_DESC_SIZE 32 391 #define EXT4_MIN_DESC_SIZE_64BIT 64 392 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 393 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 394 #ifdef __KERNEL__ 395 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 396 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 397 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 398 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 399 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 400 #else 401 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 402 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 403 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 404 #endif 405 406 /* 407 * Constants relative to the data blocks 408 */ 409 #define EXT4_NDIR_BLOCKS 12 410 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 411 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 412 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 413 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 414 415 /* 416 * Inode flags 417 */ 418 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 419 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 420 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 421 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 422 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 423 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 424 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 425 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 426 /* Reserved for compression usage... */ 427 #define EXT4_DIRTY_FL 0x00000100 428 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 429 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 430 /* nb: was previously EXT2_ECOMPR_FL */ 431 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 432 /* End compression flags --- maybe not all used */ 433 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 434 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 435 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 436 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 437 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 438 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 439 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 440 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 441 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 442 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 443 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 444 445 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 446 447 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 448 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 449 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 450 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 451 452 /* User modifiable flags */ 453 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 454 EXT4_UNRM_FL | \ 455 EXT4_COMPR_FL | \ 456 EXT4_SYNC_FL | \ 457 EXT4_IMMUTABLE_FL | \ 458 EXT4_APPEND_FL | \ 459 EXT4_NODUMP_FL | \ 460 EXT4_NOATIME_FL | \ 461 EXT4_JOURNAL_DATA_FL | \ 462 EXT4_NOTAIL_FL | \ 463 EXT4_DIRSYNC_FL | \ 464 EXT4_TOPDIR_FL | \ 465 EXT4_EXTENTS_FL | \ 466 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 467 EXT4_DAX_FL | \ 468 EXT4_PROJINHERIT_FL | \ 469 EXT4_CASEFOLD_FL) 470 471 /* User visible flags */ 472 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 473 EXT4_DIRTY_FL | \ 474 EXT4_COMPRBLK_FL | \ 475 EXT4_NOCOMPR_FL | \ 476 EXT4_ENCRYPT_FL | \ 477 EXT4_INDEX_FL | \ 478 EXT4_VERITY_FL | \ 479 EXT4_INLINE_DATA_FL) 480 481 /* Flags that should be inherited by new inodes from their parent. */ 482 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 483 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 484 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 485 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 486 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 487 EXT4_DAX_FL) 488 489 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 490 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 491 EXT4_PROJINHERIT_FL)) 492 493 /* Flags that are appropriate for non-directories/regular files. */ 494 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 495 496 /* The only flags that should be swapped */ 497 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 498 499 /* Flags which are mutually exclusive to DAX */ 500 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 501 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 502 503 /* Mask out flags that are inappropriate for the given type of inode. */ 504 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 505 { 506 if (S_ISDIR(mode)) 507 return flags; 508 else if (S_ISREG(mode)) 509 return flags & EXT4_REG_FLMASK; 510 else 511 return flags & EXT4_OTHER_FLMASK; 512 } 513 514 /* 515 * Inode flags used for atomic set/get 516 */ 517 enum { 518 EXT4_INODE_SECRM = 0, /* Secure deletion */ 519 EXT4_INODE_UNRM = 1, /* Undelete */ 520 EXT4_INODE_COMPR = 2, /* Compress file */ 521 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 522 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 523 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 524 EXT4_INODE_NODUMP = 6, /* do not dump file */ 525 EXT4_INODE_NOATIME = 7, /* do not update atime */ 526 /* Reserved for compression usage... */ 527 EXT4_INODE_DIRTY = 8, 528 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 529 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 530 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 531 /* End compression flags --- maybe not all used */ 532 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 533 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 534 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 535 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 536 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 537 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 538 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 539 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 540 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 541 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 542 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 543 EXT4_INODE_DAX = 25, /* Inode is DAX */ 544 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 545 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 546 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 547 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 548 }; 549 550 /* 551 * Since it's pretty easy to mix up bit numbers and hex values, we use a 552 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 553 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 554 * any extra space in the compiled kernel image, otherwise, the build will fail. 555 * It's important that these values are the same, since we are using 556 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 557 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 558 * values found in ext2, ext3 and ext4 filesystems, and of course the values 559 * defined in e2fsprogs. 560 * 561 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 562 */ 563 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 564 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 565 566 static inline void ext4_check_flag_values(void) 567 { 568 CHECK_FLAG_VALUE(SECRM); 569 CHECK_FLAG_VALUE(UNRM); 570 CHECK_FLAG_VALUE(COMPR); 571 CHECK_FLAG_VALUE(SYNC); 572 CHECK_FLAG_VALUE(IMMUTABLE); 573 CHECK_FLAG_VALUE(APPEND); 574 CHECK_FLAG_VALUE(NODUMP); 575 CHECK_FLAG_VALUE(NOATIME); 576 CHECK_FLAG_VALUE(DIRTY); 577 CHECK_FLAG_VALUE(COMPRBLK); 578 CHECK_FLAG_VALUE(NOCOMPR); 579 CHECK_FLAG_VALUE(ENCRYPT); 580 CHECK_FLAG_VALUE(INDEX); 581 CHECK_FLAG_VALUE(IMAGIC); 582 CHECK_FLAG_VALUE(JOURNAL_DATA); 583 CHECK_FLAG_VALUE(NOTAIL); 584 CHECK_FLAG_VALUE(DIRSYNC); 585 CHECK_FLAG_VALUE(TOPDIR); 586 CHECK_FLAG_VALUE(HUGE_FILE); 587 CHECK_FLAG_VALUE(EXTENTS); 588 CHECK_FLAG_VALUE(VERITY); 589 CHECK_FLAG_VALUE(EA_INODE); 590 CHECK_FLAG_VALUE(INLINE_DATA); 591 CHECK_FLAG_VALUE(PROJINHERIT); 592 CHECK_FLAG_VALUE(CASEFOLD); 593 CHECK_FLAG_VALUE(RESERVED); 594 } 595 596 /* Used to pass group descriptor data when online resize is done */ 597 struct ext4_new_group_input { 598 __u32 group; /* Group number for this data */ 599 __u64 block_bitmap; /* Absolute block number of block bitmap */ 600 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 601 __u64 inode_table; /* Absolute block number of inode table start */ 602 __u32 blocks_count; /* Total number of blocks in this group */ 603 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 604 __u16 unused; 605 }; 606 607 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 608 struct compat_ext4_new_group_input { 609 u32 group; 610 compat_u64 block_bitmap; 611 compat_u64 inode_bitmap; 612 compat_u64 inode_table; 613 u32 blocks_count; 614 u16 reserved_blocks; 615 u16 unused; 616 }; 617 #endif 618 619 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 620 struct ext4_new_group_data { 621 __u32 group; 622 __u64 block_bitmap; 623 __u64 inode_bitmap; 624 __u64 inode_table; 625 __u32 blocks_count; 626 __u16 reserved_blocks; 627 __u16 mdata_blocks; 628 __u32 free_clusters_count; 629 }; 630 631 /* Indexes used to index group tables in ext4_new_group_data */ 632 enum { 633 BLOCK_BITMAP = 0, /* block bitmap */ 634 INODE_BITMAP, /* inode bitmap */ 635 INODE_TABLE, /* inode tables */ 636 GROUP_TABLE_COUNT, 637 }; 638 639 /* 640 * Flags used by ext4_map_blocks() 641 */ 642 /* Allocate any needed blocks and/or convert an unwritten 643 extent to be an initialized ext4 */ 644 #define EXT4_GET_BLOCKS_CREATE 0x0001 645 /* Request the creation of an unwritten extent */ 646 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 647 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 648 EXT4_GET_BLOCKS_CREATE) 649 /* Caller is from the delayed allocation writeout path 650 * finally doing the actual allocation of delayed blocks */ 651 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 652 /* caller is from the direct IO path, request to creation of an 653 unwritten extents if not allocated, split the unwritten 654 extent if blocks has been preallocated already*/ 655 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 656 #define EXT4_GET_BLOCKS_CONVERT 0x0010 657 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 658 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 659 /* Convert extent to initialized after IO complete */ 660 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 661 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 662 /* Eventual metadata allocation (due to growing extent tree) 663 * should not fail, so try to use reserved blocks for that.*/ 664 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 665 /* Don't normalize allocation size (used for fallocate) */ 666 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 667 /* Convert written extents to unwritten */ 668 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 669 /* Write zeros to newly created written extents */ 670 #define EXT4_GET_BLOCKS_ZERO 0x0200 671 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 672 EXT4_GET_BLOCKS_ZERO) 673 /* Caller will submit data before dropping transaction handle. This 674 * allows jbd2 to avoid submitting data before commit. */ 675 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 676 /* Caller is in the atomic contex, find extent if it has been cached */ 677 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 678 679 /* 680 * The bit position of these flags must not overlap with any of the 681 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 682 * read_extent_tree_block(), ext4_split_extent_at(), 683 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 684 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 685 * caching the extents when reading from the extent tree while a 686 * truncate or punch hole operation is in progress. 687 */ 688 #define EXT4_EX_NOCACHE 0x40000000 689 #define EXT4_EX_FORCE_CACHE 0x20000000 690 #define EXT4_EX_NOFAIL 0x10000000 691 692 /* 693 * Flags used by ext4_free_blocks 694 */ 695 #define EXT4_FREE_BLOCKS_METADATA 0x0001 696 #define EXT4_FREE_BLOCKS_FORGET 0x0002 697 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 698 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 699 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 700 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 701 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 702 703 /* 704 * ioctl commands 705 */ 706 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 707 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 708 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 709 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 710 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 711 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 712 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 713 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 714 #define EXT4_IOC_MIGRATE _IO('f', 9) 715 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 716 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 717 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 718 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 719 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 720 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 721 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 722 /* ioctl codes 19--39 are reserved for fscrypt */ 723 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 724 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 725 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 726 #define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32) 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 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 757 /* 758 * ioctl commands in 32 bit emulation 759 */ 760 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 761 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 762 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 763 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 764 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 765 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 766 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 767 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 768 #endif 769 770 /* 771 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 772 * It indicates that the entry in extent status cache is for a hole. 773 */ 774 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 775 776 /* Max physical block we can address w/o extents */ 777 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 778 779 /* Max logical block we can support */ 780 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 781 782 /* 783 * Structure of an inode on the disk 784 */ 785 struct ext4_inode { 786 __le16 i_mode; /* File mode */ 787 __le16 i_uid; /* Low 16 bits of Owner Uid */ 788 __le32 i_size_lo; /* Size in bytes */ 789 __le32 i_atime; /* Access time */ 790 __le32 i_ctime; /* Inode Change time */ 791 __le32 i_mtime; /* Modification time */ 792 __le32 i_dtime; /* Deletion Time */ 793 __le16 i_gid; /* Low 16 bits of Group Id */ 794 __le16 i_links_count; /* Links count */ 795 __le32 i_blocks_lo; /* Blocks count */ 796 __le32 i_flags; /* File flags */ 797 union { 798 struct { 799 __le32 l_i_version; 800 } linux1; 801 struct { 802 __u32 h_i_translator; 803 } hurd1; 804 struct { 805 __u32 m_i_reserved1; 806 } masix1; 807 } osd1; /* OS dependent 1 */ 808 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 809 __le32 i_generation; /* File version (for NFS) */ 810 __le32 i_file_acl_lo; /* File ACL */ 811 __le32 i_size_high; 812 __le32 i_obso_faddr; /* Obsoleted fragment address */ 813 union { 814 struct { 815 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 816 __le16 l_i_file_acl_high; 817 __le16 l_i_uid_high; /* these 2 fields */ 818 __le16 l_i_gid_high; /* were reserved2[0] */ 819 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 820 __le16 l_i_reserved; 821 } linux2; 822 struct { 823 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 824 __u16 h_i_mode_high; 825 __u16 h_i_uid_high; 826 __u16 h_i_gid_high; 827 __u32 h_i_author; 828 } hurd2; 829 struct { 830 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 831 __le16 m_i_file_acl_high; 832 __u32 m_i_reserved2[2]; 833 } masix2; 834 } osd2; /* OS dependent 2 */ 835 __le16 i_extra_isize; 836 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 837 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 838 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 839 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 840 __le32 i_crtime; /* File Creation time */ 841 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 842 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 843 __le32 i_projid; /* Project ID */ 844 }; 845 846 struct move_extent { 847 __u32 reserved; /* should be zero */ 848 __u32 donor_fd; /* donor file descriptor */ 849 __u64 orig_start; /* logical start offset in block for orig */ 850 __u64 donor_start; /* logical start offset in block for donor */ 851 __u64 len; /* block length to be moved */ 852 __u64 moved_len; /* moved block length */ 853 }; 854 855 #define EXT4_EPOCH_BITS 2 856 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 857 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 858 859 /* 860 * Extended fields will fit into an inode if the filesystem was formatted 861 * with large inodes (-I 256 or larger) and there are not currently any EAs 862 * consuming all of the available space. For new inodes we always reserve 863 * enough space for the kernel's known extended fields, but for inodes 864 * created with an old kernel this might not have been the case. None of 865 * the extended inode fields is critical for correct filesystem operation. 866 * This macro checks if a certain field fits in the inode. Note that 867 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 868 */ 869 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 870 ((offsetof(typeof(*ext4_inode), field) + \ 871 sizeof((ext4_inode)->field)) \ 872 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 873 (einode)->i_extra_isize)) \ 874 875 /* 876 * We use an encoding that preserves the times for extra epoch "00": 877 * 878 * extra msb of adjust for signed 879 * epoch 32-bit 32-bit tv_sec to 880 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 881 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 882 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 883 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 884 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 885 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 886 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 887 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 888 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 889 * 890 * Note that previous versions of the kernel on 64-bit systems would 891 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 892 * 1970. e2fsck will correct this, assuming that it is run on the 893 * affected filesystem before 2242. 894 */ 895 896 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 897 { 898 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 899 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 900 } 901 902 static inline void ext4_decode_extra_time(struct timespec64 *time, 903 __le32 extra) 904 { 905 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 906 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 907 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 908 } 909 910 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 911 do { \ 912 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 913 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 914 (raw_inode)->xtime ## _extra = \ 915 ext4_encode_extra_time(&(inode)->xtime); \ 916 } \ 917 else \ 918 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 919 } while (0) 920 921 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 922 do { \ 923 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 924 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 925 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 926 (raw_inode)->xtime ## _extra = \ 927 ext4_encode_extra_time(&(einode)->xtime); \ 928 } while (0) 929 930 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 931 do { \ 932 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 933 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 934 ext4_decode_extra_time(&(inode)->xtime, \ 935 raw_inode->xtime ## _extra); \ 936 } \ 937 else \ 938 (inode)->xtime.tv_nsec = 0; \ 939 } while (0) 940 941 942 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 943 do { \ 944 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 945 (einode)->xtime.tv_sec = \ 946 (signed)le32_to_cpu((raw_inode)->xtime); \ 947 else \ 948 (einode)->xtime.tv_sec = 0; \ 949 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 950 ext4_decode_extra_time(&(einode)->xtime, \ 951 raw_inode->xtime ## _extra); \ 952 else \ 953 (einode)->xtime.tv_nsec = 0; \ 954 } while (0) 955 956 #define i_disk_version osd1.linux1.l_i_version 957 958 #if defined(__KERNEL__) || defined(__linux__) 959 #define i_reserved1 osd1.linux1.l_i_reserved1 960 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 961 #define i_blocks_high osd2.linux2.l_i_blocks_high 962 #define i_uid_low i_uid 963 #define i_gid_low i_gid 964 #define i_uid_high osd2.linux2.l_i_uid_high 965 #define i_gid_high osd2.linux2.l_i_gid_high 966 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 967 968 #elif defined(__GNU__) 969 970 #define i_translator osd1.hurd1.h_i_translator 971 #define i_uid_high osd2.hurd2.h_i_uid_high 972 #define i_gid_high osd2.hurd2.h_i_gid_high 973 #define i_author osd2.hurd2.h_i_author 974 975 #elif defined(__masix__) 976 977 #define i_reserved1 osd1.masix1.m_i_reserved1 978 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 979 #define i_reserved2 osd2.masix2.m_i_reserved2 980 981 #endif /* defined(__KERNEL__) || defined(__linux__) */ 982 983 #include "extents_status.h" 984 #include "fast_commit.h" 985 986 /* 987 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 988 * 989 * These are needed to avoid lockdep false positives when we need to 990 * allocate blocks to the quota inode during ext4_map_blocks(), while 991 * holding i_data_sem for a normal (non-quota) inode. Since we don't 992 * do quota tracking for the quota inode, this avoids deadlock (as 993 * well as infinite recursion, since it isn't turtles all the way 994 * down...) 995 * 996 * I_DATA_SEM_NORMAL - Used for most inodes 997 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 998 * where the second inode has larger inode number 999 * than the first 1000 * I_DATA_SEM_QUOTA - Used for quota inodes only 1001 */ 1002 enum { 1003 I_DATA_SEM_NORMAL = 0, 1004 I_DATA_SEM_OTHER, 1005 I_DATA_SEM_QUOTA, 1006 }; 1007 1008 1009 /* 1010 * fourth extended file system inode data in memory 1011 */ 1012 struct ext4_inode_info { 1013 __le32 i_data[15]; /* unconverted */ 1014 __u32 i_dtime; 1015 ext4_fsblk_t i_file_acl; 1016 1017 /* 1018 * i_block_group is the number of the block group which contains 1019 * this file's inode. Constant across the lifetime of the inode, 1020 * it is used for making block allocation decisions - we try to 1021 * place a file's data blocks near its inode block, and new inodes 1022 * near to their parent directory's inode. 1023 */ 1024 ext4_group_t i_block_group; 1025 ext4_lblk_t i_dir_start_lookup; 1026 #if (BITS_PER_LONG < 64) 1027 unsigned long i_state_flags; /* Dynamic state flags */ 1028 #endif 1029 unsigned long i_flags; 1030 1031 /* 1032 * Extended attributes can be read independently of the main file 1033 * data. Taking i_rwsem even when reading would cause contention 1034 * between readers of EAs and writers of regular file data, so 1035 * instead we synchronize on xattr_sem when reading or changing 1036 * EAs. 1037 */ 1038 struct rw_semaphore xattr_sem; 1039 1040 /* 1041 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1042 * i_orphan is used. 1043 */ 1044 union { 1045 struct list_head i_orphan; /* unlinked but open inodes */ 1046 unsigned int i_orphan_idx; /* Index in orphan file */ 1047 }; 1048 1049 /* Fast commit related info */ 1050 1051 /* For tracking dentry create updates */ 1052 struct list_head i_fc_dilist; 1053 struct list_head i_fc_list; /* 1054 * inodes that need fast commit 1055 * protected by sbi->s_fc_lock. 1056 */ 1057 1058 /* Start of lblk range that needs to be committed in this fast commit */ 1059 ext4_lblk_t i_fc_lblk_start; 1060 1061 /* End of lblk range that needs to be committed in this fast commit */ 1062 ext4_lblk_t i_fc_lblk_len; 1063 1064 /* Number of ongoing updates on this inode */ 1065 atomic_t i_fc_updates; 1066 1067 /* Fast commit wait queue for this inode */ 1068 wait_queue_head_t i_fc_wait; 1069 1070 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1071 struct mutex i_fc_lock; 1072 1073 /* 1074 * i_disksize keeps track of what the inode size is ON DISK, not 1075 * in memory. During truncate, i_size is set to the new size by 1076 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1077 * set i_disksize to 0 until the truncate is actually under way. 1078 * 1079 * The intent is that i_disksize always represents the blocks which 1080 * are used by this file. This allows recovery to restart truncate 1081 * on orphans if we crash during truncate. We actually write i_disksize 1082 * into the on-disk inode when writing inodes out, instead of i_size. 1083 * 1084 * The only time when i_disksize and i_size may be different is when 1085 * a truncate is in progress. The only things which change i_disksize 1086 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1087 */ 1088 loff_t i_disksize; 1089 1090 /* 1091 * i_data_sem is for serialising ext4_truncate() against 1092 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1093 * data tree are chopped off during truncate. We can't do that in 1094 * ext4 because whenever we perform intermediate commits during 1095 * truncate, the inode and all the metadata blocks *must* be in a 1096 * consistent state which allows truncation of the orphans to restart 1097 * during recovery. Hence we must fix the get_block-vs-truncate race 1098 * by other means, so we have i_data_sem. 1099 */ 1100 struct rw_semaphore i_data_sem; 1101 struct inode vfs_inode; 1102 struct jbd2_inode *jinode; 1103 1104 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1105 1106 /* 1107 * File creation time. Its function is same as that of 1108 * struct timespec64 i_{a,c,m}time in the generic inode. 1109 */ 1110 struct timespec64 i_crtime; 1111 1112 /* mballoc */ 1113 atomic_t i_prealloc_active; 1114 struct list_head i_prealloc_list; 1115 spinlock_t i_prealloc_lock; 1116 1117 /* extents status tree */ 1118 struct ext4_es_tree i_es_tree; 1119 rwlock_t i_es_lock; 1120 struct list_head i_es_list; 1121 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1122 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1123 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1124 extents to shrink. Protected by 1125 i_es_lock */ 1126 1127 /* ialloc */ 1128 ext4_group_t i_last_alloc_group; 1129 1130 /* allocation reservation info for delalloc */ 1131 /* In case of bigalloc, this refer to clusters rather than blocks */ 1132 unsigned int i_reserved_data_blocks; 1133 1134 /* pending cluster reservations for bigalloc file systems */ 1135 struct ext4_pending_tree i_pending_tree; 1136 1137 /* on-disk additional length */ 1138 __u16 i_extra_isize; 1139 1140 /* Indicate the inline data space. */ 1141 u16 i_inline_off; 1142 u16 i_inline_size; 1143 1144 #ifdef CONFIG_QUOTA 1145 /* quota space reservation, managed internally by quota code */ 1146 qsize_t i_reserved_quota; 1147 #endif 1148 1149 /* Lock protecting lists below */ 1150 spinlock_t i_completed_io_lock; 1151 /* 1152 * Completed IOs that need unwritten extents handling and have 1153 * transaction reserved 1154 */ 1155 struct list_head i_rsv_conversion_list; 1156 struct work_struct i_rsv_conversion_work; 1157 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1158 1159 spinlock_t i_block_reservation_lock; 1160 1161 /* 1162 * Transactions that contain inode's metadata needed to complete 1163 * fsync and fdatasync, respectively. 1164 */ 1165 tid_t i_sync_tid; 1166 tid_t i_datasync_tid; 1167 1168 #ifdef CONFIG_QUOTA 1169 struct dquot *i_dquot[MAXQUOTAS]; 1170 #endif 1171 1172 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1173 __u32 i_csum_seed; 1174 1175 kprojid_t i_projid; 1176 }; 1177 1178 /* 1179 * File system states 1180 */ 1181 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1182 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1183 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1184 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1185 1186 /* 1187 * Misc. filesystem flags 1188 */ 1189 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1190 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1191 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1192 1193 /* 1194 * Mount flags set via mount options or defaults 1195 */ 1196 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1197 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1198 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1199 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1200 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1201 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1202 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1203 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1204 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1205 #ifdef CONFIG_FS_DAX 1206 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1207 #else 1208 #define EXT4_MOUNT_DAX_ALWAYS 0 1209 #endif 1210 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1211 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1212 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1213 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1214 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1215 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1216 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1217 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1218 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1219 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1220 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1221 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1222 * enable enforcement for hidden 1223 * quota files */ 1224 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1225 * enforcement for hidden quota 1226 * files */ 1227 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1228 * enforcement */ 1229 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1230 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1231 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1232 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1233 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1234 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1235 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1236 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1237 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1238 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1239 1240 /* 1241 * Mount flags set either automatically (could not be set by mount option) 1242 * based on per file system feature or property or in special cases such as 1243 * distinguishing between explicit mount option definition and default. 1244 */ 1245 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1246 specified delalloc */ 1247 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1248 size of blocksize * 8 1249 blocks */ 1250 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1251 file systems */ 1252 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1253 specified journal checksum */ 1254 1255 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1256 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1257 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1258 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1259 * scanning in mballoc 1260 */ 1261 1262 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1263 ~EXT4_MOUNT_##opt 1264 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1265 EXT4_MOUNT_##opt 1266 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1267 EXT4_MOUNT_##opt) 1268 1269 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1270 ~EXT4_MOUNT2_##opt 1271 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1272 EXT4_MOUNT2_##opt 1273 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1274 EXT4_MOUNT2_##opt) 1275 1276 #define ext4_test_and_set_bit __test_and_set_bit_le 1277 #define ext4_set_bit __set_bit_le 1278 #define ext4_set_bit_atomic ext2_set_bit_atomic 1279 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1280 #define ext4_clear_bit __clear_bit_le 1281 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1282 #define ext4_test_bit test_bit_le 1283 #define ext4_find_next_zero_bit find_next_zero_bit_le 1284 #define ext4_find_next_bit find_next_bit_le 1285 1286 extern void mb_set_bits(void *bm, int cur, int len); 1287 1288 /* 1289 * Maximal mount counts between two filesystem checks 1290 */ 1291 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1292 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1293 1294 /* 1295 * Behaviour when detecting errors 1296 */ 1297 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1298 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1299 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1300 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1301 1302 /* Metadata checksum algorithm codes */ 1303 #define EXT4_CRC32C_CHKSUM 1 1304 1305 #define EXT4_LABEL_MAX 16 1306 1307 /* 1308 * Structure of the super block 1309 */ 1310 struct ext4_super_block { 1311 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1312 __le32 s_blocks_count_lo; /* Blocks count */ 1313 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1314 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1315 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1316 __le32 s_first_data_block; /* First Data Block */ 1317 __le32 s_log_block_size; /* Block size */ 1318 __le32 s_log_cluster_size; /* Allocation cluster size */ 1319 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1320 __le32 s_clusters_per_group; /* # Clusters per group */ 1321 __le32 s_inodes_per_group; /* # Inodes per group */ 1322 __le32 s_mtime; /* Mount time */ 1323 /*30*/ __le32 s_wtime; /* Write time */ 1324 __le16 s_mnt_count; /* Mount count */ 1325 __le16 s_max_mnt_count; /* Maximal mount count */ 1326 __le16 s_magic; /* Magic signature */ 1327 __le16 s_state; /* File system state */ 1328 __le16 s_errors; /* Behaviour when detecting errors */ 1329 __le16 s_minor_rev_level; /* minor revision level */ 1330 /*40*/ __le32 s_lastcheck; /* time of last check */ 1331 __le32 s_checkinterval; /* max. time between checks */ 1332 __le32 s_creator_os; /* OS */ 1333 __le32 s_rev_level; /* Revision level */ 1334 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1335 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1336 /* 1337 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1338 * 1339 * Note: the difference between the compatible feature set and 1340 * the incompatible feature set is that if there is a bit set 1341 * in the incompatible feature set that the kernel doesn't 1342 * know about, it should refuse to mount the filesystem. 1343 * 1344 * e2fsck's requirements are more strict; if it doesn't know 1345 * about a feature in either the compatible or incompatible 1346 * feature set, it must abort and not try to meddle with 1347 * things it doesn't understand... 1348 */ 1349 __le32 s_first_ino; /* First non-reserved inode */ 1350 __le16 s_inode_size; /* size of inode structure */ 1351 __le16 s_block_group_nr; /* block group # of this superblock */ 1352 __le32 s_feature_compat; /* compatible feature set */ 1353 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1354 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1355 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1356 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1357 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1358 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1359 /* 1360 * Performance hints. Directory preallocation should only 1361 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1362 */ 1363 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1364 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1365 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1366 /* 1367 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1368 */ 1369 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1370 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1371 __le32 s_journal_dev; /* device number of journal file */ 1372 __le32 s_last_orphan; /* start of list of inodes to delete */ 1373 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1374 __u8 s_def_hash_version; /* Default hash version to use */ 1375 __u8 s_jnl_backup_type; 1376 __le16 s_desc_size; /* size of group descriptor */ 1377 /*100*/ __le32 s_default_mount_opts; 1378 __le32 s_first_meta_bg; /* First metablock block group */ 1379 __le32 s_mkfs_time; /* When the filesystem was created */ 1380 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1381 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1382 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1383 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1384 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1385 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1386 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1387 __le32 s_flags; /* Miscellaneous flags */ 1388 __le16 s_raid_stride; /* RAID stride */ 1389 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1390 __le64 s_mmp_block; /* Block for multi-mount protection */ 1391 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1392 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1393 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1394 __u8 s_encryption_level; /* versioning level for encryption */ 1395 __u8 s_reserved_pad; /* Padding to next 32bits */ 1396 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1397 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1398 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1399 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1400 snapshot's future use */ 1401 __le32 s_snapshot_list; /* inode number of the head of the 1402 on-disk snapshot list */ 1403 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1404 __le32 s_error_count; /* number of fs errors */ 1405 __le32 s_first_error_time; /* first time an error happened */ 1406 __le32 s_first_error_ino; /* inode involved in first error */ 1407 __le64 s_first_error_block; /* block involved of first error */ 1408 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1409 __le32 s_first_error_line; /* line number where error happened */ 1410 __le32 s_last_error_time; /* most recent time of an error */ 1411 __le32 s_last_error_ino; /* inode involved in last error */ 1412 __le32 s_last_error_line; /* line number where error happened */ 1413 __le64 s_last_error_block; /* block involved of last error */ 1414 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1415 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1416 __u8 s_mount_opts[64]; 1417 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1418 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1419 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1420 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1421 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1422 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1423 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1424 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1425 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1426 __u8 s_wtime_hi; 1427 __u8 s_mtime_hi; 1428 __u8 s_mkfs_time_hi; 1429 __u8 s_lastcheck_hi; 1430 __u8 s_first_error_time_hi; 1431 __u8 s_last_error_time_hi; 1432 __u8 s_first_error_errcode; 1433 __u8 s_last_error_errcode; 1434 __le16 s_encoding; /* Filename charset encoding */ 1435 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1436 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1437 __le32 s_reserved[94]; /* Padding to the end of the block */ 1438 __le32 s_checksum; /* crc32c(superblock) */ 1439 }; 1440 1441 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1442 1443 #ifdef __KERNEL__ 1444 1445 /* Number of quota types we support */ 1446 #define EXT4_MAXQUOTAS 3 1447 1448 #define EXT4_ENC_UTF8_12_1 1 1449 1450 /* Types of ext4 journal triggers */ 1451 enum ext4_journal_trigger_type { 1452 EXT4_JTR_ORPHAN_FILE, 1453 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1454 }; 1455 1456 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1457 1458 struct ext4_journal_trigger { 1459 struct jbd2_buffer_trigger_type tr_triggers; 1460 struct super_block *sb; 1461 }; 1462 1463 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1464 struct jbd2_buffer_trigger_type *trigger) 1465 { 1466 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1467 } 1468 1469 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1470 1471 /* Structure at the tail of orphan block */ 1472 struct ext4_orphan_block_tail { 1473 __le32 ob_magic; 1474 __le32 ob_checksum; 1475 }; 1476 1477 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1478 { 1479 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1480 sizeof(u32); 1481 } 1482 1483 struct ext4_orphan_block { 1484 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1485 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1486 }; 1487 1488 /* 1489 * Info about orphan file. 1490 */ 1491 struct ext4_orphan_info { 1492 int of_blocks; /* Number of orphan blocks in a file */ 1493 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1494 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1495 * file blocks */ 1496 }; 1497 1498 /* 1499 * fourth extended-fs super-block data in memory 1500 */ 1501 struct ext4_sb_info { 1502 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1503 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1504 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1505 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1506 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1507 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1508 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1509 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1510 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1511 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1512 unsigned long s_overhead; /* # of fs overhead clusters */ 1513 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1514 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1515 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1516 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1517 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1518 struct buffer_head * __rcu *s_group_desc; 1519 unsigned int s_mount_opt; 1520 unsigned int s_mount_opt2; 1521 unsigned long s_mount_flags; 1522 unsigned int s_def_mount_opt; 1523 ext4_fsblk_t s_sb_block; 1524 atomic64_t s_resv_clusters; 1525 kuid_t s_resuid; 1526 kgid_t s_resgid; 1527 unsigned short s_mount_state; 1528 unsigned short s_pad; 1529 int s_addr_per_block_bits; 1530 int s_desc_per_block_bits; 1531 int s_inode_size; 1532 int s_first_ino; 1533 unsigned int s_inode_readahead_blks; 1534 unsigned int s_inode_goal; 1535 u32 s_hash_seed[4]; 1536 int s_def_hash_version; 1537 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1538 struct percpu_counter s_freeclusters_counter; 1539 struct percpu_counter s_freeinodes_counter; 1540 struct percpu_counter s_dirs_counter; 1541 struct percpu_counter s_dirtyclusters_counter; 1542 struct percpu_counter s_sra_exceeded_retry_limit; 1543 struct blockgroup_lock *s_blockgroup_lock; 1544 struct proc_dir_entry *s_proc; 1545 struct kobject s_kobj; 1546 struct completion s_kobj_unregister; 1547 struct super_block *s_sb; 1548 struct buffer_head *s_mmp_bh; 1549 1550 /* Journaling */ 1551 struct journal_s *s_journal; 1552 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1553 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1554 struct list_head s_orphan; /* List of orphaned inodes in on disk 1555 list */ 1556 struct ext4_orphan_info s_orphan_info; 1557 unsigned long s_commit_interval; 1558 u32 s_max_batch_time; 1559 u32 s_min_batch_time; 1560 struct block_device *s_journal_bdev; 1561 #ifdef CONFIG_QUOTA 1562 /* Names of quota files with journalled quota */ 1563 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1564 int s_jquota_fmt; /* Format of quota to use */ 1565 #endif 1566 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1567 struct ext4_system_blocks __rcu *s_system_blks; 1568 1569 #ifdef EXTENTS_STATS 1570 /* ext4 extents stats */ 1571 unsigned long s_ext_min; 1572 unsigned long s_ext_max; 1573 unsigned long s_depth_max; 1574 spinlock_t s_ext_stats_lock; 1575 unsigned long s_ext_blocks; 1576 unsigned long s_ext_extents; 1577 #endif 1578 1579 /* for buddy allocator */ 1580 struct ext4_group_info ** __rcu *s_group_info; 1581 struct inode *s_buddy_cache; 1582 spinlock_t s_md_lock; 1583 unsigned short *s_mb_offsets; 1584 unsigned int *s_mb_maxs; 1585 unsigned int s_group_info_size; 1586 unsigned int s_mb_free_pending; 1587 struct list_head s_freed_data_list; /* List of blocks to be freed 1588 after commit completed */ 1589 struct list_head s_discard_list; 1590 struct work_struct s_discard_work; 1591 atomic_t s_retry_alloc_pending; 1592 struct rb_root s_mb_avg_fragment_size_root; 1593 rwlock_t s_mb_rb_lock; 1594 struct list_head *s_mb_largest_free_orders; 1595 rwlock_t *s_mb_largest_free_orders_locks; 1596 1597 /* tunables */ 1598 unsigned long s_stripe; 1599 unsigned int s_mb_max_linear_groups; 1600 unsigned int s_mb_stream_request; 1601 unsigned int s_mb_max_to_scan; 1602 unsigned int s_mb_min_to_scan; 1603 unsigned int s_mb_stats; 1604 unsigned int s_mb_order2_reqs; 1605 unsigned int s_mb_group_prealloc; 1606 unsigned int s_mb_max_inode_prealloc; 1607 unsigned int s_max_dir_size_kb; 1608 /* where last allocation was done - for stream allocation */ 1609 unsigned long s_mb_last_group; 1610 unsigned long s_mb_last_start; 1611 unsigned int s_mb_prefetch; 1612 unsigned int s_mb_prefetch_limit; 1613 1614 /* stats for buddy allocator */ 1615 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1616 atomic_t s_bal_success; /* we found long enough chunks */ 1617 atomic_t s_bal_allocated; /* in blocks */ 1618 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1619 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1620 atomic_t s_bal_goals; /* goal hits */ 1621 atomic_t s_bal_breaks; /* too long searches */ 1622 atomic_t s_bal_2orders; /* 2^order hits */ 1623 atomic_t s_bal_cr0_bad_suggestions; 1624 atomic_t s_bal_cr1_bad_suggestions; 1625 atomic64_t s_bal_cX_groups_considered[4]; 1626 atomic64_t s_bal_cX_hits[4]; 1627 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1628 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1629 atomic64_t s_mb_generation_time; 1630 atomic_t s_mb_lost_chunks; 1631 atomic_t s_mb_preallocated; 1632 atomic_t s_mb_discarded; 1633 atomic_t s_lock_busy; 1634 1635 /* locality groups */ 1636 struct ext4_locality_group __percpu *s_locality_groups; 1637 1638 /* for write statistics */ 1639 unsigned long s_sectors_written_start; 1640 u64 s_kbytes_written; 1641 1642 /* the size of zero-out chunk */ 1643 unsigned int s_extent_max_zeroout_kb; 1644 1645 unsigned int s_log_groups_per_flex; 1646 struct flex_groups * __rcu *s_flex_groups; 1647 ext4_group_t s_flex_groups_allocated; 1648 1649 /* workqueue for reserved extent conversions (buffered io) */ 1650 struct workqueue_struct *rsv_conversion_wq; 1651 1652 /* timer for periodic error stats printing */ 1653 struct timer_list s_err_report; 1654 1655 /* Lazy inode table initialization info */ 1656 struct ext4_li_request *s_li_request; 1657 /* Wait multiplier for lazy initialization thread */ 1658 unsigned int s_li_wait_mult; 1659 1660 /* Kernel thread for multiple mount protection */ 1661 struct task_struct *s_mmp_tsk; 1662 1663 /* record the last minlen when FITRIM is called. */ 1664 unsigned long s_last_trim_minblks; 1665 1666 /* Reference to checksum algorithm driver via cryptoapi */ 1667 struct crypto_shash *s_chksum_driver; 1668 1669 /* Precomputed FS UUID checksum for seeding other checksums */ 1670 __u32 s_csum_seed; 1671 1672 /* Reclaim extents from extent status tree */ 1673 struct shrinker s_es_shrinker; 1674 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1675 long s_es_nr_inode; 1676 struct ext4_es_stats s_es_stats; 1677 struct mb_cache *s_ea_block_cache; 1678 struct mb_cache *s_ea_inode_cache; 1679 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1680 1681 /* Journal triggers for checksum computation */ 1682 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1683 1684 /* Ratelimit ext4 messages. */ 1685 struct ratelimit_state s_err_ratelimit_state; 1686 struct ratelimit_state s_warning_ratelimit_state; 1687 struct ratelimit_state s_msg_ratelimit_state; 1688 atomic_t s_warning_count; 1689 atomic_t s_msg_count; 1690 1691 /* Encryption policy for '-o test_dummy_encryption' */ 1692 struct fscrypt_dummy_policy s_dummy_enc_policy; 1693 1694 /* 1695 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1696 * or EXTENTS flag. 1697 */ 1698 struct percpu_rw_semaphore s_writepages_rwsem; 1699 struct dax_device *s_daxdev; 1700 u64 s_dax_part_off; 1701 #ifdef CONFIG_EXT4_DEBUG 1702 unsigned long s_simulate_fail; 1703 #endif 1704 /* Record the errseq of the backing block device */ 1705 errseq_t s_bdev_wb_err; 1706 spinlock_t s_bdev_wb_lock; 1707 1708 /* Information about errors that happened during this mount */ 1709 spinlock_t s_error_lock; 1710 int s_add_error_count; 1711 int s_first_error_code; 1712 __u32 s_first_error_line; 1713 __u32 s_first_error_ino; 1714 __u64 s_first_error_block; 1715 const char *s_first_error_func; 1716 time64_t s_first_error_time; 1717 int s_last_error_code; 1718 __u32 s_last_error_line; 1719 __u32 s_last_error_ino; 1720 __u64 s_last_error_block; 1721 const char *s_last_error_func; 1722 time64_t s_last_error_time; 1723 /* 1724 * If we are in a context where we cannot update error information in 1725 * the on-disk superblock, we queue this work to do it. 1726 */ 1727 struct work_struct s_error_work; 1728 1729 /* Ext4 fast commit sub transaction ID */ 1730 atomic_t s_fc_subtid; 1731 1732 /* 1733 * After commit starts, the main queue gets locked, and the further 1734 * updates get added in the staging queue. 1735 */ 1736 #define FC_Q_MAIN 0 1737 #define FC_Q_STAGING 1 1738 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1739 * that have data changes in them. 1740 */ 1741 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1742 unsigned int s_fc_bytes; 1743 /* 1744 * Main fast commit lock. This lock protects accesses to the 1745 * following fields: 1746 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1747 */ 1748 spinlock_t s_fc_lock; 1749 struct buffer_head *s_fc_bh; 1750 struct ext4_fc_stats s_fc_stats; 1751 tid_t s_fc_ineligible_tid; 1752 #ifdef CONFIG_EXT4_DEBUG 1753 int s_fc_debug_max_replay; 1754 #endif 1755 struct ext4_fc_replay_state s_fc_replay_state; 1756 }; 1757 1758 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1759 { 1760 return sb->s_fs_info; 1761 } 1762 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1763 { 1764 return container_of(inode, struct ext4_inode_info, vfs_inode); 1765 } 1766 1767 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1768 { 1769 return ino == EXT4_ROOT_INO || 1770 (ino >= EXT4_FIRST_INO(sb) && 1771 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1772 } 1773 1774 /* 1775 * Returns: sbi->field[index] 1776 * Used to access an array element from the following sbi fields which require 1777 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1778 * - s_group_desc 1779 * - s_group_info 1780 * - s_flex_group 1781 */ 1782 #define sbi_array_rcu_deref(sbi, field, index) \ 1783 ({ \ 1784 typeof(*((sbi)->field)) _v; \ 1785 rcu_read_lock(); \ 1786 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1787 rcu_read_unlock(); \ 1788 _v; \ 1789 }) 1790 1791 /* 1792 * run-time mount flags 1793 */ 1794 enum { 1795 EXT4_MF_MNTDIR_SAMPLED, 1796 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1797 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1798 }; 1799 1800 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1801 { 1802 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1803 } 1804 1805 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1806 { 1807 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1808 } 1809 1810 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1811 { 1812 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1813 } 1814 1815 1816 /* 1817 * Simulate_fail codes 1818 */ 1819 #define EXT4_SIM_BBITMAP_EIO 1 1820 #define EXT4_SIM_BBITMAP_CRC 2 1821 #define EXT4_SIM_IBITMAP_EIO 3 1822 #define EXT4_SIM_IBITMAP_CRC 4 1823 #define EXT4_SIM_INODE_EIO 5 1824 #define EXT4_SIM_INODE_CRC 6 1825 #define EXT4_SIM_DIRBLOCK_EIO 7 1826 #define EXT4_SIM_DIRBLOCK_CRC 8 1827 1828 static inline bool ext4_simulate_fail(struct super_block *sb, 1829 unsigned long code) 1830 { 1831 #ifdef CONFIG_EXT4_DEBUG 1832 struct ext4_sb_info *sbi = EXT4_SB(sb); 1833 1834 if (unlikely(sbi->s_simulate_fail == code)) { 1835 sbi->s_simulate_fail = 0; 1836 return true; 1837 } 1838 #endif 1839 return false; 1840 } 1841 1842 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1843 struct buffer_head *bh, 1844 unsigned long code) 1845 { 1846 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1847 clear_buffer_uptodate(bh); 1848 } 1849 1850 /* 1851 * Error number codes for s_{first,last}_error_errno 1852 * 1853 * Linux errno numbers are architecture specific, so we need to translate 1854 * them into something which is architecture independent. We don't define 1855 * codes for all errno's; just the ones which are most likely to be the cause 1856 * of an ext4_error() call. 1857 */ 1858 #define EXT4_ERR_UNKNOWN 1 1859 #define EXT4_ERR_EIO 2 1860 #define EXT4_ERR_ENOMEM 3 1861 #define EXT4_ERR_EFSBADCRC 4 1862 #define EXT4_ERR_EFSCORRUPTED 5 1863 #define EXT4_ERR_ENOSPC 6 1864 #define EXT4_ERR_ENOKEY 7 1865 #define EXT4_ERR_EROFS 8 1866 #define EXT4_ERR_EFBIG 9 1867 #define EXT4_ERR_EEXIST 10 1868 #define EXT4_ERR_ERANGE 11 1869 #define EXT4_ERR_EOVERFLOW 12 1870 #define EXT4_ERR_EBUSY 13 1871 #define EXT4_ERR_ENOTDIR 14 1872 #define EXT4_ERR_ENOTEMPTY 15 1873 #define EXT4_ERR_ESHUTDOWN 16 1874 #define EXT4_ERR_EFAULT 17 1875 1876 /* 1877 * Inode dynamic state flags 1878 */ 1879 enum { 1880 EXT4_STATE_JDATA, /* journaled data exists */ 1881 EXT4_STATE_NEW, /* inode is newly created */ 1882 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1883 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1884 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1885 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1886 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1887 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1888 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1889 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1890 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1891 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1892 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1893 }; 1894 1895 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1896 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1897 { \ 1898 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1899 } \ 1900 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1901 { \ 1902 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1903 } \ 1904 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1905 { \ 1906 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1907 } 1908 1909 /* Add these declarations here only so that these functions can be 1910 * found by name. Otherwise, they are very hard to locate. */ 1911 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1912 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1913 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1914 EXT4_INODE_BIT_FNS(flag, flags, 0) 1915 1916 /* Add these declarations here only so that these functions can be 1917 * found by name. Otherwise, they are very hard to locate. */ 1918 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1919 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1920 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1921 #if (BITS_PER_LONG < 64) 1922 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1923 1924 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1925 { 1926 (ei)->i_state_flags = 0; 1927 } 1928 #else 1929 EXT4_INODE_BIT_FNS(state, flags, 32) 1930 1931 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1932 { 1933 /* We depend on the fact that callers will set i_flags */ 1934 } 1935 #endif 1936 #else 1937 /* Assume that user mode programs are passing in an ext4fs superblock, not 1938 * a kernel struct super_block. This will allow us to call the feature-test 1939 * macros from user land. */ 1940 #define EXT4_SB(sb) (sb) 1941 #endif 1942 1943 static inline bool ext4_verity_in_progress(struct inode *inode) 1944 { 1945 return IS_ENABLED(CONFIG_FS_VERITY) && 1946 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1947 } 1948 1949 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1950 1951 /* 1952 * Codes for operating systems 1953 */ 1954 #define EXT4_OS_LINUX 0 1955 #define EXT4_OS_HURD 1 1956 #define EXT4_OS_MASIX 2 1957 #define EXT4_OS_FREEBSD 3 1958 #define EXT4_OS_LITES 4 1959 1960 /* 1961 * Revision levels 1962 */ 1963 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1964 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1965 1966 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1967 1968 #define EXT4_GOOD_OLD_INODE_SIZE 128 1969 1970 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1971 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1972 #define EXT4_TIMESTAMP_MIN S32_MIN 1973 1974 /* 1975 * Feature set definitions 1976 */ 1977 1978 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1979 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1980 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1981 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1982 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1983 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1984 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1985 /* 1986 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1987 * incompatible only if fast commit blocks are present in the FS. Since we 1988 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1989 * incompatible. We also have a JBD2 incompat feature, which gets set when 1990 * there are fast commit blocks present in the journal. 1991 */ 1992 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1993 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1994 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1995 1996 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1997 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1998 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1999 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2000 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2001 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2002 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2003 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2004 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2005 /* 2006 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2007 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2008 * all other data structures' checksums. However, the METADATA_CSUM and 2009 * GDT_CSUM bits are mutually exclusive. 2010 */ 2011 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2012 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2013 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2014 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2015 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2016 non-empty */ 2017 2018 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2019 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2020 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2021 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2022 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2023 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2024 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2025 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2026 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2027 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2028 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2029 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2030 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2031 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2032 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2033 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2034 2035 extern void ext4_update_dynamic_rev(struct super_block *sb); 2036 2037 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2038 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2039 { \ 2040 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2041 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2042 } \ 2043 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2044 { \ 2045 ext4_update_dynamic_rev(sb); \ 2046 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2047 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2048 } \ 2049 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2050 { \ 2051 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2052 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2053 } 2054 2055 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2056 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2057 { \ 2058 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2059 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2060 } \ 2061 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2062 { \ 2063 ext4_update_dynamic_rev(sb); \ 2064 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2065 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2066 } \ 2067 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2068 { \ 2069 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2070 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2071 } 2072 2073 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2074 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2075 { \ 2076 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2077 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2078 } \ 2079 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2080 { \ 2081 ext4_update_dynamic_rev(sb); \ 2082 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2083 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2084 } \ 2085 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2086 { \ 2087 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2088 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2089 } 2090 2091 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2092 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2093 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2094 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2095 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2096 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2097 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2098 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2099 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2100 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2101 2102 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2103 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2104 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2105 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2106 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2107 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2108 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2109 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2110 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2111 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2112 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2113 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2114 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2115 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2116 2117 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2118 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2119 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2120 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2121 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2122 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2123 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2124 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2125 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2126 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2127 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2128 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2129 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2130 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2131 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2132 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2133 2134 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2135 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2136 EXT4_FEATURE_INCOMPAT_META_BG) 2137 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2138 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2139 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2140 2141 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2142 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2143 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2144 EXT4_FEATURE_INCOMPAT_META_BG) 2145 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2146 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2147 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2148 2149 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2150 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2151 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2152 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2153 EXT4_FEATURE_INCOMPAT_META_BG| \ 2154 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2155 EXT4_FEATURE_INCOMPAT_64BIT| \ 2156 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2157 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2158 EXT4_FEATURE_INCOMPAT_MMP | \ 2159 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2160 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2161 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2162 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2163 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2164 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2165 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2166 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2167 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2168 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2169 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2170 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2171 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2172 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2173 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2174 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2175 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2176 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2177 2178 #define EXTN_FEATURE_FUNCS(ver) \ 2179 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2180 { \ 2181 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2182 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2183 } \ 2184 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2185 { \ 2186 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2187 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2188 } \ 2189 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2190 { \ 2191 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2192 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2193 } 2194 2195 EXTN_FEATURE_FUNCS(2) 2196 EXTN_FEATURE_FUNCS(3) 2197 EXTN_FEATURE_FUNCS(4) 2198 2199 static inline bool ext4_has_compat_features(struct super_block *sb) 2200 { 2201 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2202 } 2203 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2204 { 2205 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2206 } 2207 static inline bool ext4_has_incompat_features(struct super_block *sb) 2208 { 2209 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2210 } 2211 2212 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2213 2214 /* 2215 * Superblock flags 2216 */ 2217 #define EXT4_FLAGS_RESIZING 0 2218 #define EXT4_FLAGS_SHUTDOWN 1 2219 #define EXT4_FLAGS_BDEV_IS_DAX 2 2220 2221 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2222 { 2223 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2224 } 2225 2226 /* 2227 * Default values for user and/or group using reserved blocks 2228 */ 2229 #define EXT4_DEF_RESUID 0 2230 #define EXT4_DEF_RESGID 0 2231 2232 /* 2233 * Default project ID 2234 */ 2235 #define EXT4_DEF_PROJID 0 2236 2237 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2238 2239 /* 2240 * Default mount options 2241 */ 2242 #define EXT4_DEFM_DEBUG 0x0001 2243 #define EXT4_DEFM_BSDGROUPS 0x0002 2244 #define EXT4_DEFM_XATTR_USER 0x0004 2245 #define EXT4_DEFM_ACL 0x0008 2246 #define EXT4_DEFM_UID16 0x0010 2247 #define EXT4_DEFM_JMODE 0x0060 2248 #define EXT4_DEFM_JMODE_DATA 0x0020 2249 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2250 #define EXT4_DEFM_JMODE_WBACK 0x0060 2251 #define EXT4_DEFM_NOBARRIER 0x0100 2252 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2253 #define EXT4_DEFM_DISCARD 0x0400 2254 #define EXT4_DEFM_NODELALLOC 0x0800 2255 2256 /* 2257 * Default journal batch times 2258 */ 2259 #define EXT4_DEF_MIN_BATCH_TIME 0 2260 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2261 2262 /* 2263 * Minimum number of groups in a flexgroup before we separate out 2264 * directories into the first block group of a flexgroup 2265 */ 2266 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2267 2268 /* 2269 * Structure of a directory entry 2270 */ 2271 #define EXT4_NAME_LEN 255 2272 /* 2273 * Base length of the ext4 directory entry excluding the name length 2274 */ 2275 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2276 2277 struct ext4_dir_entry { 2278 __le32 inode; /* Inode number */ 2279 __le16 rec_len; /* Directory entry length */ 2280 __le16 name_len; /* Name length */ 2281 char name[EXT4_NAME_LEN]; /* File name */ 2282 }; 2283 2284 2285 /* 2286 * Encrypted Casefolded entries require saving the hash on disk. This structure 2287 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2288 * boundary. 2289 */ 2290 struct ext4_dir_entry_hash { 2291 __le32 hash; 2292 __le32 minor_hash; 2293 }; 2294 2295 /* 2296 * The new version of the directory entry. Since EXT4 structures are 2297 * stored in intel byte order, and the name_len field could never be 2298 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2299 * file_type field. 2300 */ 2301 struct ext4_dir_entry_2 { 2302 __le32 inode; /* Inode number */ 2303 __le16 rec_len; /* Directory entry length */ 2304 __u8 name_len; /* Name length */ 2305 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2306 char name[EXT4_NAME_LEN]; /* File name */ 2307 }; 2308 2309 /* 2310 * Access the hashes at the end of ext4_dir_entry_2 2311 */ 2312 #define EXT4_DIRENT_HASHES(entry) \ 2313 ((struct ext4_dir_entry_hash *) \ 2314 (((void *)(entry)) + \ 2315 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2316 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2317 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2318 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2319 2320 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2321 { 2322 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2323 } 2324 2325 /* 2326 * This is a bogus directory entry at the end of each leaf block that 2327 * records checksums. 2328 */ 2329 struct ext4_dir_entry_tail { 2330 __le32 det_reserved_zero1; /* Pretend to be unused */ 2331 __le16 det_rec_len; /* 12 */ 2332 __u8 det_reserved_zero2; /* Zero name length */ 2333 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2334 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2335 }; 2336 2337 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2338 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2339 ((blocksize) - \ 2340 sizeof(struct ext4_dir_entry_tail)))) 2341 2342 /* 2343 * Ext4 directory file types. Only the low 3 bits are used. The 2344 * other bits are reserved for now. 2345 */ 2346 #define EXT4_FT_UNKNOWN 0 2347 #define EXT4_FT_REG_FILE 1 2348 #define EXT4_FT_DIR 2 2349 #define EXT4_FT_CHRDEV 3 2350 #define EXT4_FT_BLKDEV 4 2351 #define EXT4_FT_FIFO 5 2352 #define EXT4_FT_SOCK 6 2353 #define EXT4_FT_SYMLINK 7 2354 2355 #define EXT4_FT_MAX 8 2356 2357 #define EXT4_FT_DIR_CSUM 0xDE 2358 2359 /* 2360 * EXT4_DIR_PAD defines the directory entries boundaries 2361 * 2362 * NOTE: It must be a multiple of 4 2363 */ 2364 #define EXT4_DIR_PAD 4 2365 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2366 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2367 2368 /* 2369 * The rec_len is dependent on the type of directory. Directories that are 2370 * casefolded and encrypted need to store the hash as well, so we add room for 2371 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2372 * pass NULL for dir, as those entries do not use the extra fields. 2373 */ 2374 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2375 const struct inode *dir) 2376 { 2377 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2378 2379 if (dir && ext4_hash_in_dirent(dir)) 2380 rec_len += sizeof(struct ext4_dir_entry_hash); 2381 return (rec_len & ~EXT4_DIR_ROUND); 2382 } 2383 2384 /* 2385 * If we ever get support for fs block sizes > page_size, we'll need 2386 * to remove the #if statements in the next two functions... 2387 */ 2388 static inline unsigned int 2389 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2390 { 2391 unsigned len = le16_to_cpu(dlen); 2392 2393 #if (PAGE_SIZE >= 65536) 2394 if (len == EXT4_MAX_REC_LEN || len == 0) 2395 return blocksize; 2396 return (len & 65532) | ((len & 3) << 16); 2397 #else 2398 return len; 2399 #endif 2400 } 2401 2402 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2403 { 2404 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2405 #if (PAGE_SIZE >= 65536) 2406 if (len < 65536) 2407 return cpu_to_le16(len); 2408 if (len == blocksize) { 2409 if (blocksize == 65536) 2410 return cpu_to_le16(EXT4_MAX_REC_LEN); 2411 else 2412 return cpu_to_le16(0); 2413 } 2414 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2415 #else 2416 return cpu_to_le16(len); 2417 #endif 2418 } 2419 2420 /* 2421 * Hash Tree Directory indexing 2422 * (c) Daniel Phillips, 2001 2423 */ 2424 2425 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2426 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2427 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2428 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2429 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2430 2431 /* Legal values for the dx_root hash_version field: */ 2432 2433 #define DX_HASH_LEGACY 0 2434 #define DX_HASH_HALF_MD4 1 2435 #define DX_HASH_TEA 2 2436 #define DX_HASH_LEGACY_UNSIGNED 3 2437 #define DX_HASH_HALF_MD4_UNSIGNED 4 2438 #define DX_HASH_TEA_UNSIGNED 5 2439 #define DX_HASH_SIPHASH 6 2440 2441 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2442 const void *address, unsigned int length) 2443 { 2444 struct { 2445 struct shash_desc shash; 2446 char ctx[4]; 2447 } desc; 2448 2449 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2450 2451 desc.shash.tfm = sbi->s_chksum_driver; 2452 *(u32 *)desc.ctx = crc; 2453 2454 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2455 2456 return *(u32 *)desc.ctx; 2457 } 2458 2459 #ifdef __KERNEL__ 2460 2461 /* hash info structure used by the directory hash */ 2462 struct dx_hash_info 2463 { 2464 u32 hash; 2465 u32 minor_hash; 2466 int hash_version; 2467 u32 *seed; 2468 }; 2469 2470 2471 /* 32 and 64 bit signed EOF for dx directories */ 2472 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2473 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2474 2475 2476 /* 2477 * Control parameters used by ext4_htree_next_block 2478 */ 2479 #define HASH_NB_ALWAYS 1 2480 2481 struct ext4_filename { 2482 const struct qstr *usr_fname; 2483 struct fscrypt_str disk_name; 2484 struct dx_hash_info hinfo; 2485 #ifdef CONFIG_FS_ENCRYPTION 2486 struct fscrypt_str crypto_buf; 2487 #endif 2488 #if IS_ENABLED(CONFIG_UNICODE) 2489 struct fscrypt_str cf_name; 2490 #endif 2491 }; 2492 2493 #define fname_name(p) ((p)->disk_name.name) 2494 #define fname_usr_name(p) ((p)->usr_fname->name) 2495 #define fname_len(p) ((p)->disk_name.len) 2496 2497 /* 2498 * Describe an inode's exact location on disk and in memory 2499 */ 2500 struct ext4_iloc 2501 { 2502 struct buffer_head *bh; 2503 unsigned long offset; 2504 ext4_group_t block_group; 2505 }; 2506 2507 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2508 { 2509 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2510 } 2511 2512 static inline bool ext4_is_quota_file(struct inode *inode) 2513 { 2514 return IS_NOQUOTA(inode) && 2515 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2516 } 2517 2518 /* 2519 * This structure is stuffed into the struct file's private_data field 2520 * for directories. It is where we put information so that we can do 2521 * readdir operations in hash tree order. 2522 */ 2523 struct dir_private_info { 2524 struct rb_root root; 2525 struct rb_node *curr_node; 2526 struct fname *extra_fname; 2527 loff_t last_pos; 2528 __u32 curr_hash; 2529 __u32 curr_minor_hash; 2530 __u32 next_hash; 2531 }; 2532 2533 /* calculate the first block number of the group */ 2534 static inline ext4_fsblk_t 2535 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2536 { 2537 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2538 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2539 } 2540 2541 /* 2542 * Special error return code only used by dx_probe() and its callers. 2543 */ 2544 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2545 2546 /* htree levels for ext4 */ 2547 #define EXT4_HTREE_LEVEL_COMPAT 2 2548 #define EXT4_HTREE_LEVEL 3 2549 2550 static inline int ext4_dir_htree_level(struct super_block *sb) 2551 { 2552 return ext4_has_feature_largedir(sb) ? 2553 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2554 } 2555 2556 /* 2557 * Timeout and state flag for lazy initialization inode thread. 2558 */ 2559 #define EXT4_DEF_LI_WAIT_MULT 10 2560 #define EXT4_DEF_LI_MAX_START_DELAY 5 2561 #define EXT4_LAZYINIT_QUIT 0x0001 2562 #define EXT4_LAZYINIT_RUNNING 0x0002 2563 2564 /* 2565 * Lazy inode table initialization info 2566 */ 2567 struct ext4_lazy_init { 2568 unsigned long li_state; 2569 struct list_head li_request_list; 2570 struct mutex li_list_mtx; 2571 }; 2572 2573 enum ext4_li_mode { 2574 EXT4_LI_MODE_PREFETCH_BBITMAP, 2575 EXT4_LI_MODE_ITABLE, 2576 }; 2577 2578 struct ext4_li_request { 2579 struct super_block *lr_super; 2580 enum ext4_li_mode lr_mode; 2581 ext4_group_t lr_first_not_zeroed; 2582 ext4_group_t lr_next_group; 2583 struct list_head lr_request; 2584 unsigned long lr_next_sched; 2585 unsigned long lr_timeout; 2586 }; 2587 2588 struct ext4_features { 2589 struct kobject f_kobj; 2590 struct completion f_kobj_unregister; 2591 }; 2592 2593 /* 2594 * This structure will be used for multiple mount protection. It will be 2595 * written into the block number saved in the s_mmp_block field in the 2596 * superblock. Programs that check MMP should assume that if 2597 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2598 * to use the filesystem, regardless of how old the timestamp is. 2599 */ 2600 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2601 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2602 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2603 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2604 2605 struct mmp_struct { 2606 __le32 mmp_magic; /* Magic number for MMP */ 2607 __le32 mmp_seq; /* Sequence no. updated periodically */ 2608 2609 /* 2610 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2611 * purposes and do not affect the correctness of the algorithm 2612 */ 2613 __le64 mmp_time; /* Time last updated */ 2614 char mmp_nodename[64]; /* Node which last updated MMP block */ 2615 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2616 2617 /* 2618 * mmp_check_interval is used to verify if the MMP block has been 2619 * updated on the block device. The value is updated based on the 2620 * maximum time to write the MMP block during an update cycle. 2621 */ 2622 __le16 mmp_check_interval; 2623 2624 __le16 mmp_pad1; 2625 __le32 mmp_pad2[226]; 2626 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2627 }; 2628 2629 /* arguments passed to the mmp thread */ 2630 struct mmpd_data { 2631 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2632 struct super_block *sb; /* super block of the fs */ 2633 }; 2634 2635 /* 2636 * Check interval multiplier 2637 * The MMP block is written every update interval and initially checked every 2638 * update interval x the multiplier (the value is then adapted based on the 2639 * write latency). The reason is that writes can be delayed under load and we 2640 * don't want readers to incorrectly assume that the filesystem is no longer 2641 * in use. 2642 */ 2643 #define EXT4_MMP_CHECK_MULT 2UL 2644 2645 /* 2646 * Minimum interval for MMP checking in seconds. 2647 */ 2648 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2649 2650 /* 2651 * Maximum interval for MMP checking in seconds. 2652 */ 2653 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2654 2655 /* 2656 * Function prototypes 2657 */ 2658 2659 /* 2660 * Ok, these declarations are also in <linux/kernel.h> but none of the 2661 * ext4 source programs needs to include it so they are duplicated here. 2662 */ 2663 # define NORET_TYPE /**/ 2664 # define ATTRIB_NORET __attribute__((noreturn)) 2665 # define NORET_AND noreturn, 2666 2667 /* bitmap.c */ 2668 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2669 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2670 struct ext4_group_desc *gdp, 2671 struct buffer_head *bh, int sz); 2672 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2673 struct ext4_group_desc *gdp, 2674 struct buffer_head *bh, int sz); 2675 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2676 struct ext4_group_desc *gdp, 2677 struct buffer_head *bh); 2678 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2679 struct ext4_group_desc *gdp, 2680 struct buffer_head *bh); 2681 2682 /* balloc.c */ 2683 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2684 ext4_fsblk_t blocknr, 2685 ext4_group_t *blockgrpp, 2686 ext4_grpblk_t *offsetp); 2687 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2688 ext4_fsblk_t block); 2689 2690 extern unsigned int ext4_block_group(struct super_block *sb, 2691 ext4_fsblk_t blocknr); 2692 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2693 ext4_fsblk_t blocknr); 2694 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2695 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2696 ext4_group_t group); 2697 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2698 ext4_fsblk_t goal, 2699 unsigned int flags, 2700 unsigned long *count, 2701 int *errp); 2702 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2703 s64 nclusters, unsigned int flags); 2704 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2705 extern void ext4_check_blocks_bitmap(struct super_block *); 2706 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2707 ext4_group_t block_group, 2708 struct buffer_head ** bh); 2709 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2710 2711 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2712 ext4_group_t block_group, 2713 bool ignore_locked); 2714 extern int ext4_wait_block_bitmap(struct super_block *sb, 2715 ext4_group_t block_group, 2716 struct buffer_head *bh); 2717 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2718 ext4_group_t block_group); 2719 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2720 ext4_group_t block_group, 2721 struct ext4_group_desc *gdp); 2722 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2723 2724 #if IS_ENABLED(CONFIG_UNICODE) 2725 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2726 const struct qstr *iname, 2727 struct ext4_filename *fname); 2728 #endif 2729 2730 /* ext4 encryption related stuff goes here crypto.c */ 2731 #ifdef CONFIG_FS_ENCRYPTION 2732 extern const struct fscrypt_operations ext4_cryptops; 2733 2734 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2735 int lookup, struct ext4_filename *fname); 2736 2737 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2738 struct ext4_filename *fname); 2739 2740 void ext4_fname_free_filename(struct ext4_filename *fname); 2741 2742 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2743 2744 #else /* !CONFIG_FS_ENCRYPTION */ 2745 static inline int ext4_fname_setup_filename(struct inode *dir, 2746 const struct qstr *iname, 2747 int lookup, 2748 struct ext4_filename *fname) 2749 { 2750 int err = 0; 2751 fname->usr_fname = iname; 2752 fname->disk_name.name = (unsigned char *) iname->name; 2753 fname->disk_name.len = iname->len; 2754 2755 #if IS_ENABLED(CONFIG_UNICODE) 2756 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2757 #endif 2758 2759 return err; 2760 } 2761 2762 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2763 struct dentry *dentry, 2764 struct ext4_filename *fname) 2765 { 2766 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2767 } 2768 2769 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2770 { 2771 #if IS_ENABLED(CONFIG_UNICODE) 2772 kfree(fname->cf_name.name); 2773 fname->cf_name.name = NULL; 2774 #endif 2775 } 2776 2777 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2778 void __user *arg) 2779 { 2780 return -EOPNOTSUPP; 2781 } 2782 #endif /* !CONFIG_FS_ENCRYPTION */ 2783 2784 /* dir.c */ 2785 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2786 struct file *, 2787 struct ext4_dir_entry_2 *, 2788 struct buffer_head *, char *, int, 2789 unsigned int); 2790 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2791 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2792 (de), (bh), (buf), (size), (offset))) 2793 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2794 __u32 minor_hash, 2795 struct ext4_dir_entry_2 *dirent, 2796 struct fscrypt_str *ent_name); 2797 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2798 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2799 struct buffer_head *bh, 2800 void *buf, int buf_size, 2801 struct ext4_filename *fname, 2802 struct ext4_dir_entry_2 **dest_de); 2803 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2804 struct ext4_dir_entry_2 *de, 2805 int buf_size, 2806 struct ext4_filename *fname); 2807 static inline void ext4_update_dx_flag(struct inode *inode) 2808 { 2809 if (!ext4_has_feature_dir_index(inode->i_sb) && 2810 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2811 /* ext4_iget() should have caught this... */ 2812 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2813 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2814 } 2815 } 2816 static const unsigned char ext4_filetype_table[] = { 2817 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2818 }; 2819 2820 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2821 { 2822 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2823 return DT_UNKNOWN; 2824 2825 return ext4_filetype_table[filetype]; 2826 } 2827 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2828 void *buf, int buf_size); 2829 2830 /* fsync.c */ 2831 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2832 2833 /* hash.c */ 2834 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2835 struct dx_hash_info *hinfo); 2836 2837 /* ialloc.c */ 2838 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2839 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *, 2840 struct inode *, umode_t, 2841 const struct qstr *qstr, __u32 goal, 2842 uid_t *owner, __u32 i_flags, 2843 int handle_type, unsigned int line_no, 2844 int nblocks); 2845 2846 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2847 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \ 2848 (goal), (owner), i_flags, 0, 0, 0) 2849 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \ 2850 type, nblocks) \ 2851 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2852 0, (type), __LINE__, (nblocks)) 2853 2854 2855 extern void ext4_free_inode(handle_t *, struct inode *); 2856 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2857 extern unsigned long ext4_count_free_inodes(struct super_block *); 2858 extern unsigned long ext4_count_dirs(struct super_block *); 2859 extern void ext4_check_inodes_bitmap(struct super_block *); 2860 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2861 extern int ext4_init_inode_table(struct super_block *sb, 2862 ext4_group_t group, int barrier); 2863 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2864 2865 /* fast_commit.c */ 2866 int ext4_fc_info_show(struct seq_file *seq, void *v); 2867 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2868 void ext4_fc_init_inode(struct inode *inode); 2869 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2870 ext4_lblk_t end); 2871 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2872 struct dentry *dentry); 2873 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2874 struct dentry *dentry); 2875 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2876 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2877 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2878 struct dentry *dentry); 2879 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2880 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2881 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2882 void ext4_fc_start_update(struct inode *inode); 2883 void ext4_fc_stop_update(struct inode *inode); 2884 void ext4_fc_del(struct inode *inode); 2885 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2886 void ext4_fc_replay_cleanup(struct super_block *sb); 2887 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2888 int __init ext4_fc_init_dentry_cache(void); 2889 void ext4_fc_destroy_dentry_cache(void); 2890 int ext4_fc_record_regions(struct super_block *sb, int ino, 2891 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2892 int len, int replay); 2893 2894 /* mballoc.c */ 2895 extern const struct seq_operations ext4_mb_seq_groups_ops; 2896 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2897 extern long ext4_mb_stats; 2898 extern long ext4_mb_max_to_scan; 2899 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2900 extern int ext4_mb_init(struct super_block *); 2901 extern int ext4_mb_release(struct super_block *); 2902 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2903 struct ext4_allocation_request *, int *); 2904 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2905 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2906 extern int __init ext4_init_mballoc(void); 2907 extern void ext4_exit_mballoc(void); 2908 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2909 ext4_group_t group, 2910 unsigned int nr, int *cnt); 2911 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2912 unsigned int nr); 2913 2914 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2915 struct buffer_head *bh, ext4_fsblk_t block, 2916 unsigned long count, int flags); 2917 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2918 ext4_group_t ngroups); 2919 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2920 ext4_group_t i, struct ext4_group_desc *desc); 2921 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2922 ext4_fsblk_t block, unsigned long count); 2923 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2924 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2925 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2926 int len, int state); 2927 2928 /* inode.c */ 2929 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2930 struct ext4_inode_info *ei); 2931 int ext4_inode_is_fast_symlink(struct inode *inode); 2932 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2933 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2934 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2935 bool wait, struct buffer_head **bhs); 2936 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2937 struct buffer_head *bh_result, int create); 2938 int ext4_get_block(struct inode *inode, sector_t iblock, 2939 struct buffer_head *bh_result, int create); 2940 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2941 struct buffer_head *bh, int create); 2942 int ext4_walk_page_buffers(handle_t *handle, 2943 struct inode *inode, 2944 struct buffer_head *head, 2945 unsigned from, 2946 unsigned to, 2947 int *partial, 2948 int (*fn)(handle_t *handle, struct inode *inode, 2949 struct buffer_head *bh)); 2950 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2951 struct buffer_head *bh); 2952 #define FALL_BACK_TO_NONDELALLOC 1 2953 #define CONVERT_INLINE_DATA 2 2954 2955 typedef enum { 2956 EXT4_IGET_NORMAL = 0, 2957 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2958 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2959 } ext4_iget_flags; 2960 2961 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2962 ext4_iget_flags flags, const char *function, 2963 unsigned int line); 2964 2965 #define ext4_iget(sb, ino, flags) \ 2966 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2967 2968 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2969 extern int ext4_setattr(struct user_namespace *, struct dentry *, 2970 struct iattr *); 2971 extern int ext4_getattr(struct user_namespace *, const struct path *, 2972 struct kstat *, u32, unsigned int); 2973 extern void ext4_evict_inode(struct inode *); 2974 extern void ext4_clear_inode(struct inode *); 2975 extern int ext4_file_getattr(struct user_namespace *, const struct path *, 2976 struct kstat *, u32, unsigned int); 2977 extern int ext4_sync_inode(handle_t *, struct inode *); 2978 extern void ext4_dirty_inode(struct inode *, int); 2979 extern int ext4_change_inode_journal_flag(struct inode *, int); 2980 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2981 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2982 struct ext4_iloc *iloc); 2983 extern int ext4_inode_attach_jinode(struct inode *inode); 2984 extern int ext4_can_truncate(struct inode *inode); 2985 extern int ext4_truncate(struct inode *); 2986 extern int ext4_break_layouts(struct inode *); 2987 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2988 extern void ext4_set_inode_flags(struct inode *, bool init); 2989 extern int ext4_alloc_da_blocks(struct inode *inode); 2990 extern void ext4_set_aops(struct inode *inode); 2991 extern int ext4_writepage_trans_blocks(struct inode *); 2992 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2993 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2994 loff_t lstart, loff_t lend); 2995 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2996 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2997 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2998 extern void ext4_da_release_space(struct inode *inode, int to_free); 2999 extern void ext4_da_update_reserve_space(struct inode *inode, 3000 int used, int quota_claim); 3001 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3002 ext4_fsblk_t pblk, ext4_lblk_t len); 3003 3004 /* indirect.c */ 3005 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3006 struct ext4_map_blocks *map, int flags); 3007 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3008 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3009 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3010 ext4_lblk_t start, ext4_lblk_t end); 3011 3012 /* ioctl.c */ 3013 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3014 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3015 int ext4_fileattr_set(struct user_namespace *mnt_userns, 3016 struct dentry *dentry, struct fileattr *fa); 3017 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3018 extern void ext4_reset_inode_seed(struct inode *inode); 3019 int ext4_update_overhead(struct super_block *sb); 3020 3021 /* migrate.c */ 3022 extern int ext4_ext_migrate(struct inode *); 3023 extern int ext4_ind_migrate(struct inode *inode); 3024 3025 /* namei.c */ 3026 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3027 struct inode *inode); 3028 extern int ext4_dirblock_csum_verify(struct inode *inode, 3029 struct buffer_head *bh); 3030 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3031 __u32 start_minor_hash, __u32 *next_hash); 3032 extern int ext4_search_dir(struct buffer_head *bh, 3033 char *search_buf, 3034 int buf_size, 3035 struct inode *dir, 3036 struct ext4_filename *fname, 3037 unsigned int offset, 3038 struct ext4_dir_entry_2 **res_dir); 3039 extern int ext4_generic_delete_entry(struct inode *dir, 3040 struct ext4_dir_entry_2 *de_del, 3041 struct buffer_head *bh, 3042 void *entry_buf, 3043 int buf_size, 3044 int csum_size); 3045 extern bool ext4_empty_dir(struct inode *inode); 3046 3047 /* resize.c */ 3048 extern void ext4_kvfree_array_rcu(void *to_free); 3049 extern int ext4_group_add(struct super_block *sb, 3050 struct ext4_new_group_data *input); 3051 extern int ext4_group_extend(struct super_block *sb, 3052 struct ext4_super_block *es, 3053 ext4_fsblk_t n_blocks_count); 3054 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3055 extern unsigned int ext4_list_backups(struct super_block *sb, 3056 unsigned int *three, unsigned int *five, 3057 unsigned int *seven); 3058 3059 /* super.c */ 3060 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3061 sector_t block, int op_flags); 3062 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3063 sector_t block); 3064 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags, 3065 bh_end_io_t *end_io); 3066 extern int ext4_read_bh(struct buffer_head *bh, int op_flags, 3067 bh_end_io_t *end_io); 3068 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait); 3069 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3070 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3071 extern int ext4_calculate_overhead(struct super_block *sb); 3072 extern __le32 ext4_superblock_csum(struct super_block *sb, 3073 struct ext4_super_block *es); 3074 extern void ext4_superblock_csum_set(struct super_block *sb); 3075 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3076 ext4_group_t ngroup); 3077 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3078 char nbuf[16]); 3079 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3080 ext4_group_t block_group, 3081 unsigned int flags); 3082 3083 extern __printf(7, 8) 3084 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3085 int, __u64, const char *, ...); 3086 extern __printf(6, 7) 3087 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3088 ext4_fsblk_t, int, const char *, ...); 3089 extern __printf(5, 6) 3090 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3091 const char *, ...); 3092 extern void __ext4_std_error(struct super_block *, const char *, 3093 unsigned int, int); 3094 extern __printf(4, 5) 3095 void __ext4_warning(struct super_block *, const char *, unsigned int, 3096 const char *, ...); 3097 extern __printf(4, 5) 3098 void __ext4_warning_inode(const struct inode *inode, const char *function, 3099 unsigned int line, const char *fmt, ...); 3100 extern __printf(3, 4) 3101 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3102 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3103 const char *, unsigned int, const char *); 3104 extern __printf(7, 8) 3105 void __ext4_grp_locked_error(const char *, unsigned int, 3106 struct super_block *, ext4_group_t, 3107 unsigned long, ext4_fsblk_t, 3108 const char *, ...); 3109 3110 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3111 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3112 3113 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3114 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3115 3116 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3117 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3118 (fmt), ## a) 3119 3120 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3121 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3122 3123 #define ext4_abort(sb, err, fmt, a...) \ 3124 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3125 3126 #ifdef CONFIG_PRINTK 3127 3128 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3129 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3130 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3131 __ext4_error_inode((inode), (func), (line), (block), \ 3132 (err), (fmt), ##__VA_ARGS__) 3133 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3134 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3135 #define ext4_error(sb, fmt, ...) \ 3136 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3137 ##__VA_ARGS__) 3138 #define ext4_error_err(sb, err, fmt, ...) \ 3139 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3140 ##__VA_ARGS__) 3141 #define ext4_warning(sb, fmt, ...) \ 3142 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3143 #define ext4_warning_inode(inode, fmt, ...) \ 3144 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3145 #define ext4_msg(sb, level, fmt, ...) \ 3146 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3147 #define dump_mmp_msg(sb, mmp, msg) \ 3148 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3149 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3150 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3151 fmt, ##__VA_ARGS__) 3152 3153 #else 3154 3155 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3156 do { \ 3157 no_printk(fmt, ##__VA_ARGS__); \ 3158 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3159 } while (0) 3160 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3161 do { \ 3162 no_printk(fmt, ##__VA_ARGS__); \ 3163 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3164 } while (0) 3165 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3166 do { \ 3167 no_printk(fmt, ##__VA_ARGS__); \ 3168 __ext4_error_file(file, "", 0, block, " "); \ 3169 } while (0) 3170 #define ext4_error(sb, fmt, ...) \ 3171 do { \ 3172 no_printk(fmt, ##__VA_ARGS__); \ 3173 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3174 } while (0) 3175 #define ext4_error_err(sb, err, fmt, ...) \ 3176 do { \ 3177 no_printk(fmt, ##__VA_ARGS__); \ 3178 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3179 } while (0) 3180 #define ext4_warning(sb, fmt, ...) \ 3181 do { \ 3182 no_printk(fmt, ##__VA_ARGS__); \ 3183 __ext4_warning(sb, "", 0, " "); \ 3184 } while (0) 3185 #define ext4_warning_inode(inode, fmt, ...) \ 3186 do { \ 3187 no_printk(fmt, ##__VA_ARGS__); \ 3188 __ext4_warning_inode(inode, "", 0, " "); \ 3189 } while (0) 3190 #define ext4_msg(sb, level, fmt, ...) \ 3191 do { \ 3192 no_printk(fmt, ##__VA_ARGS__); \ 3193 __ext4_msg(sb, "", " "); \ 3194 } while (0) 3195 #define dump_mmp_msg(sb, mmp, msg) \ 3196 __dump_mmp_msg(sb, mmp, "", 0, "") 3197 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3198 do { \ 3199 no_printk(fmt, ##__VA_ARGS__); \ 3200 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3201 } while (0) 3202 3203 #endif 3204 3205 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3206 struct ext4_group_desc *bg); 3207 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3208 struct ext4_group_desc *bg); 3209 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3210 struct ext4_group_desc *bg); 3211 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3212 struct ext4_group_desc *bg); 3213 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3214 struct ext4_group_desc *bg); 3215 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3216 struct ext4_group_desc *bg); 3217 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3218 struct ext4_group_desc *bg); 3219 extern void ext4_block_bitmap_set(struct super_block *sb, 3220 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3221 extern void ext4_inode_bitmap_set(struct super_block *sb, 3222 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3223 extern void ext4_inode_table_set(struct super_block *sb, 3224 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3225 extern void ext4_free_group_clusters_set(struct super_block *sb, 3226 struct ext4_group_desc *bg, 3227 __u32 count); 3228 extern void ext4_free_inodes_set(struct super_block *sb, 3229 struct ext4_group_desc *bg, __u32 count); 3230 extern void ext4_used_dirs_set(struct super_block *sb, 3231 struct ext4_group_desc *bg, __u32 count); 3232 extern void ext4_itable_unused_set(struct super_block *sb, 3233 struct ext4_group_desc *bg, __u32 count); 3234 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3235 struct ext4_group_desc *gdp); 3236 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3237 struct ext4_group_desc *gdp); 3238 extern int ext4_register_li_request(struct super_block *sb, 3239 ext4_group_t first_not_zeroed); 3240 3241 static inline int ext4_has_metadata_csum(struct super_block *sb) 3242 { 3243 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3244 !EXT4_SB(sb)->s_chksum_driver); 3245 3246 return ext4_has_feature_metadata_csum(sb) && 3247 (EXT4_SB(sb)->s_chksum_driver != NULL); 3248 } 3249 3250 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3251 { 3252 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3253 } 3254 3255 #define ext4_read_incompat_64bit_val(es, name) \ 3256 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3257 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3258 le32_to_cpu(es->name##_lo)) 3259 3260 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3261 { 3262 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3263 } 3264 3265 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3266 { 3267 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3268 } 3269 3270 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3271 { 3272 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3273 } 3274 3275 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3276 ext4_fsblk_t blk) 3277 { 3278 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3279 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3280 } 3281 3282 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3283 ext4_fsblk_t blk) 3284 { 3285 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3286 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3287 } 3288 3289 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3290 ext4_fsblk_t blk) 3291 { 3292 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3293 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3294 } 3295 3296 static inline loff_t ext4_isize(struct super_block *sb, 3297 struct ext4_inode *raw_inode) 3298 { 3299 if (ext4_has_feature_largedir(sb) || 3300 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3301 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3302 le32_to_cpu(raw_inode->i_size_lo); 3303 3304 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3305 } 3306 3307 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3308 { 3309 raw_inode->i_size_lo = cpu_to_le32(i_size); 3310 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3311 } 3312 3313 static inline 3314 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3315 ext4_group_t group) 3316 { 3317 struct ext4_group_info **grp_info; 3318 long indexv, indexh; 3319 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3320 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3321 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3322 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3323 return grp_info[indexh]; 3324 } 3325 3326 /* 3327 * Reading s_groups_count requires using smp_rmb() afterwards. See 3328 * the locking protocol documented in the comments of ext4_group_add() 3329 * in resize.c 3330 */ 3331 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3332 { 3333 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3334 3335 smp_rmb(); 3336 return ngroups; 3337 } 3338 3339 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3340 ext4_group_t block_group) 3341 { 3342 return block_group >> sbi->s_log_groups_per_flex; 3343 } 3344 3345 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3346 { 3347 return 1 << sbi->s_log_groups_per_flex; 3348 } 3349 3350 #define ext4_std_error(sb, errno) \ 3351 do { \ 3352 if ((errno)) \ 3353 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3354 } while (0) 3355 3356 #ifdef CONFIG_SMP 3357 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3358 * counters. So we need to make sure we have free clusters more 3359 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3360 */ 3361 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3362 #else 3363 #define EXT4_FREECLUSTERS_WATERMARK 0 3364 #endif 3365 3366 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3367 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3368 { 3369 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3370 !inode_is_locked(inode)); 3371 down_write(&EXT4_I(inode)->i_data_sem); 3372 if (newsize > EXT4_I(inode)->i_disksize) 3373 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3374 up_write(&EXT4_I(inode)->i_data_sem); 3375 } 3376 3377 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3378 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3379 { 3380 int changed = 0; 3381 3382 if (newsize > inode->i_size) { 3383 i_size_write(inode, newsize); 3384 changed = 1; 3385 } 3386 if (newsize > EXT4_I(inode)->i_disksize) { 3387 ext4_update_i_disksize(inode, newsize); 3388 changed |= 2; 3389 } 3390 return changed; 3391 } 3392 3393 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3394 loff_t len); 3395 3396 struct ext4_group_info { 3397 unsigned long bb_state; 3398 #ifdef AGGRESSIVE_CHECK 3399 unsigned long bb_check_counter; 3400 #endif 3401 struct rb_root bb_free_root; 3402 ext4_grpblk_t bb_first_free; /* first free block */ 3403 ext4_grpblk_t bb_free; /* total free blocks */ 3404 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3405 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3406 ext4_group_t bb_group; /* Group number */ 3407 struct list_head bb_prealloc_list; 3408 #ifdef DOUBLE_CHECK 3409 void *bb_bitmap; 3410 #endif 3411 struct rw_semaphore alloc_sem; 3412 struct rb_node bb_avg_fragment_size_rb; 3413 struct list_head bb_largest_free_order_node; 3414 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3415 * regions, index is order. 3416 * bb_counters[3] = 5 means 3417 * 5 free 8-block regions. */ 3418 }; 3419 3420 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3421 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3422 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3423 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3424 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3425 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3426 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3427 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3428 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3429 3430 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3431 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3432 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3433 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3434 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3435 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3436 3437 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3438 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3439 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3440 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3441 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3442 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3443 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3444 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3445 3446 #define EXT4_MAX_CONTENTION 8 3447 #define EXT4_CONTENTION_THRESHOLD 2 3448 3449 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3450 ext4_group_t group) 3451 { 3452 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3453 } 3454 3455 /* 3456 * Returns true if the filesystem is busy enough that attempts to 3457 * access the block group locks has run into contention. 3458 */ 3459 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3460 { 3461 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3462 } 3463 3464 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3465 { 3466 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3467 if (spin_trylock(lock)) 3468 /* 3469 * We're able to grab the lock right away, so drop the 3470 * lock contention counter. 3471 */ 3472 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3473 else { 3474 /* 3475 * The lock is busy, so bump the contention counter, 3476 * and then wait on the spin lock. 3477 */ 3478 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3479 EXT4_MAX_CONTENTION); 3480 spin_lock(lock); 3481 } 3482 } 3483 3484 static inline void ext4_unlock_group(struct super_block *sb, 3485 ext4_group_t group) 3486 { 3487 spin_unlock(ext4_group_lock_ptr(sb, group)); 3488 } 3489 3490 #ifdef CONFIG_QUOTA 3491 static inline bool ext4_quota_capable(struct super_block *sb) 3492 { 3493 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3494 } 3495 3496 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3497 { 3498 struct ext4_sb_info *sbi = EXT4_SB(sb); 3499 3500 return (ext4_has_feature_quota(sb) || 3501 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3502 } 3503 int ext4_enable_quotas(struct super_block *sb); 3504 #endif 3505 3506 /* 3507 * Block validity checking 3508 */ 3509 #define ext4_check_indirect_blockref(inode, bh) \ 3510 ext4_check_blockref(__func__, __LINE__, inode, \ 3511 (__le32 *)(bh)->b_data, \ 3512 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3513 3514 #define ext4_ind_check_inode(inode) \ 3515 ext4_check_blockref(__func__, __LINE__, inode, \ 3516 EXT4_I(inode)->i_data, \ 3517 EXT4_NDIR_BLOCKS) 3518 3519 /* 3520 * Inodes and files operations 3521 */ 3522 3523 /* dir.c */ 3524 extern const struct file_operations ext4_dir_operations; 3525 3526 /* file.c */ 3527 extern const struct inode_operations ext4_file_inode_operations; 3528 extern const struct file_operations ext4_file_operations; 3529 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3530 3531 /* inline.c */ 3532 extern int ext4_get_max_inline_size(struct inode *inode); 3533 extern int ext4_find_inline_data_nolock(struct inode *inode); 3534 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3535 unsigned int len); 3536 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3537 3538 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3539 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3540 struct inode *inode, 3541 loff_t pos, unsigned len, 3542 struct page **pagep); 3543 extern int ext4_write_inline_data_end(struct inode *inode, 3544 loff_t pos, unsigned len, 3545 unsigned copied, 3546 struct page *page); 3547 extern struct buffer_head * 3548 ext4_journalled_write_inline_data(struct inode *inode, 3549 unsigned len, 3550 struct page *page); 3551 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3552 struct inode *inode, 3553 loff_t pos, unsigned len, 3554 struct page **pagep, 3555 void **fsdata); 3556 extern int ext4_try_add_inline_entry(handle_t *handle, 3557 struct ext4_filename *fname, 3558 struct inode *dir, struct inode *inode); 3559 extern int ext4_try_create_inline_dir(handle_t *handle, 3560 struct inode *parent, 3561 struct inode *inode); 3562 extern int ext4_read_inline_dir(struct file *filp, 3563 struct dir_context *ctx, 3564 int *has_inline_data); 3565 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3566 struct inode *dir, ext4_lblk_t block, 3567 struct dx_hash_info *hinfo, 3568 __u32 start_hash, __u32 start_minor_hash, 3569 int *has_inline_data); 3570 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3571 struct ext4_filename *fname, 3572 struct ext4_dir_entry_2 **res_dir, 3573 int *has_inline_data); 3574 extern int ext4_delete_inline_entry(handle_t *handle, 3575 struct inode *dir, 3576 struct ext4_dir_entry_2 *de_del, 3577 struct buffer_head *bh, 3578 int *has_inline_data); 3579 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3580 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3581 struct ext4_dir_entry_2 **parent_de, 3582 int *retval); 3583 extern int ext4_inline_data_fiemap(struct inode *inode, 3584 struct fiemap_extent_info *fieinfo, 3585 int *has_inline, __u64 start, __u64 len); 3586 3587 struct iomap; 3588 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3589 3590 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3591 3592 extern int ext4_convert_inline_data(struct inode *inode); 3593 3594 static inline int ext4_has_inline_data(struct inode *inode) 3595 { 3596 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3597 EXT4_I(inode)->i_inline_off; 3598 } 3599 3600 /* namei.c */ 3601 extern const struct inode_operations ext4_dir_inode_operations; 3602 extern const struct inode_operations ext4_special_inode_operations; 3603 extern struct dentry *ext4_get_parent(struct dentry *child); 3604 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3605 struct ext4_dir_entry_2 *de, 3606 int blocksize, int csum_size, 3607 unsigned int parent_ino, int dotdot_real_len); 3608 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3609 unsigned int blocksize); 3610 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3611 struct buffer_head *bh); 3612 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name, 3613 struct inode *inode); 3614 extern int __ext4_link(struct inode *dir, struct inode *inode, 3615 struct dentry *dentry); 3616 3617 #define S_SHIFT 12 3618 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3619 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3620 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3621 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3622 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3623 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3624 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3625 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3626 }; 3627 3628 static inline void ext4_set_de_type(struct super_block *sb, 3629 struct ext4_dir_entry_2 *de, 3630 umode_t mode) { 3631 if (ext4_has_feature_filetype(sb)) 3632 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3633 } 3634 3635 /* readpages.c */ 3636 extern int ext4_mpage_readpages(struct inode *inode, 3637 struct readahead_control *rac, struct page *page); 3638 extern int __init ext4_init_post_read_processing(void); 3639 extern void ext4_exit_post_read_processing(void); 3640 3641 /* symlink.c */ 3642 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3643 extern const struct inode_operations ext4_symlink_inode_operations; 3644 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3645 3646 /* sysfs.c */ 3647 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3648 extern int ext4_register_sysfs(struct super_block *sb); 3649 extern void ext4_unregister_sysfs(struct super_block *sb); 3650 extern int __init ext4_init_sysfs(void); 3651 extern void ext4_exit_sysfs(void); 3652 3653 /* block_validity */ 3654 extern void ext4_release_system_zone(struct super_block *sb); 3655 extern int ext4_setup_system_zone(struct super_block *sb); 3656 extern int __init ext4_init_system_zone(void); 3657 extern void ext4_exit_system_zone(void); 3658 extern int ext4_inode_block_valid(struct inode *inode, 3659 ext4_fsblk_t start_blk, 3660 unsigned int count); 3661 extern int ext4_check_blockref(const char *, unsigned int, 3662 struct inode *, __le32 *, unsigned int); 3663 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3664 ext4_fsblk_t start_blk, unsigned int count); 3665 3666 3667 /* extents.c */ 3668 struct ext4_ext_path; 3669 struct ext4_extent; 3670 3671 /* 3672 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3673 * __le32. 3674 */ 3675 #define EXT_MAX_BLOCKS 0xffffffff 3676 3677 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3678 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3679 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3680 struct ext4_map_blocks *map, int flags); 3681 extern int ext4_ext_truncate(handle_t *, struct inode *); 3682 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3683 ext4_lblk_t end); 3684 extern void ext4_ext_init(struct super_block *); 3685 extern void ext4_ext_release(struct super_block *); 3686 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3687 loff_t len); 3688 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3689 loff_t offset, ssize_t len); 3690 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3691 ext4_io_end_t *io_end); 3692 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3693 struct ext4_map_blocks *map, int flags); 3694 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3695 int num, 3696 struct ext4_ext_path *path); 3697 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3698 struct ext4_ext_path **, 3699 struct ext4_extent *, int); 3700 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3701 struct ext4_ext_path **, 3702 int flags); 3703 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3704 extern int ext4_ext_check_inode(struct inode *inode); 3705 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3706 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3707 __u64 start, __u64 len); 3708 extern int ext4_get_es_cache(struct inode *inode, 3709 struct fiemap_extent_info *fieinfo, 3710 __u64 start, __u64 len); 3711 extern int ext4_ext_precache(struct inode *inode); 3712 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3713 struct inode *inode2, ext4_lblk_t lblk1, 3714 ext4_lblk_t lblk2, ext4_lblk_t count, 3715 int mark_unwritten,int *err); 3716 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3717 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3718 int check_cred, int restart_cred, 3719 int revoke_cred); 3720 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3721 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3722 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3723 int len, int unwritten, ext4_fsblk_t pblk); 3724 extern int ext4_ext_clear_bb(struct inode *inode); 3725 3726 3727 /* move_extent.c */ 3728 extern void ext4_double_down_write_data_sem(struct inode *first, 3729 struct inode *second); 3730 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3731 struct inode *donor_inode); 3732 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3733 __u64 start_orig, __u64 start_donor, 3734 __u64 len, __u64 *moved_len); 3735 3736 /* page-io.c */ 3737 extern int __init ext4_init_pageio(void); 3738 extern void ext4_exit_pageio(void); 3739 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3740 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3741 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3742 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3743 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3744 struct writeback_control *wbc); 3745 extern void ext4_end_io_rsv_work(struct work_struct *work); 3746 extern void ext4_io_submit(struct ext4_io_submit *io); 3747 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3748 struct page *page, 3749 int len, 3750 bool keep_towrite); 3751 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3752 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3753 3754 /* mmp.c */ 3755 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3756 3757 /* mmp.c */ 3758 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3759 3760 /* verity.c */ 3761 extern const struct fsverity_operations ext4_verityops; 3762 3763 /* orphan.c */ 3764 extern int ext4_orphan_add(handle_t *, struct inode *); 3765 extern int ext4_orphan_del(handle_t *, struct inode *); 3766 extern void ext4_orphan_cleanup(struct super_block *sb, 3767 struct ext4_super_block *es); 3768 extern void ext4_release_orphan_info(struct super_block *sb); 3769 extern int ext4_init_orphan_info(struct super_block *sb); 3770 extern int ext4_orphan_file_empty(struct super_block *sb); 3771 extern void ext4_orphan_file_block_trigger( 3772 struct jbd2_buffer_trigger_type *triggers, 3773 struct buffer_head *bh, 3774 void *data, size_t size); 3775 3776 /* 3777 * Add new method to test whether block and inode bitmaps are properly 3778 * initialized. With uninit_bg reading the block from disk is not enough 3779 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3780 */ 3781 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3782 3783 static inline int bitmap_uptodate(struct buffer_head *bh) 3784 { 3785 return (buffer_uptodate(bh) && 3786 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3787 } 3788 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3789 { 3790 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3791 } 3792 3793 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3794 3795 /* For ioend & aio unwritten conversion wait queues */ 3796 #define EXT4_WQ_HASH_SZ 37 3797 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3798 EXT4_WQ_HASH_SZ]) 3799 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3800 3801 extern int ext4_resize_begin(struct super_block *sb); 3802 extern void ext4_resize_end(struct super_block *sb); 3803 3804 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3805 struct ext4_io_end *io_end) 3806 { 3807 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3808 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3809 atomic_inc(&EXT4_I(inode)->i_unwritten); 3810 } 3811 } 3812 3813 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3814 { 3815 struct inode *inode = io_end->inode; 3816 3817 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3818 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3819 /* Wake up anyone waiting on unwritten extent conversion */ 3820 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3821 wake_up_all(ext4_ioend_wq(inode)); 3822 } 3823 } 3824 3825 extern const struct iomap_ops ext4_iomap_ops; 3826 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3827 extern const struct iomap_ops ext4_iomap_report_ops; 3828 3829 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3830 { 3831 /* 3832 * If the buffer has the write error flag, we have failed 3833 * to write out data in the block. In this case, we don't 3834 * have to read the block because we may read the old data 3835 * successfully. 3836 */ 3837 if (buffer_write_io_error(bh)) 3838 set_buffer_uptodate(bh); 3839 return buffer_uptodate(bh); 3840 } 3841 3842 #endif /* __KERNEL__ */ 3843 3844 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3845 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3846 3847 #endif /* _EXT4_H */ 3848