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 #include <uapi/linux/ext4.h> 44 45 #include <linux/fscrypt.h> 46 #include <linux/fsverity.h> 47 48 #include <linux/compiler.h> 49 50 /* 51 * The fourth extended filesystem constants/structures 52 */ 53 54 /* 55 * with AGGRESSIVE_CHECK allocator runs consistency checks over 56 * structures. these checks slow things down a lot 57 */ 58 #define AGGRESSIVE_CHECK__ 59 60 /* 61 * with DOUBLE_CHECK defined mballoc creates persistent in-core 62 * bitmaps, maintains and uses them to check for double allocations 63 */ 64 #define DOUBLE_CHECK__ 65 66 /* 67 * Define EXT4FS_DEBUG to produce debug messages 68 */ 69 #undef EXT4FS_DEBUG 70 71 /* 72 * Debug code 73 */ 74 #ifdef EXT4FS_DEBUG 75 #define ext4_debug(f, a...) \ 76 do { \ 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 78 __FILE__, __LINE__, __func__); \ 79 printk(KERN_DEBUG f, ## a); \ 80 } while (0) 81 #else 82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 83 #endif 84 85 /* 86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 87 */ 88 #define EXT_DEBUG__ 89 90 /* 91 * Dynamic printk for controlled extents debugging. 92 */ 93 #ifdef CONFIG_EXT4_DEBUG 94 #define ext_debug(ino, fmt, ...) \ 95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 96 current->comm, task_pid_nr(current), \ 97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 98 __func__, ##__VA_ARGS__) 99 #else 100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 101 #endif 102 103 #define ASSERT(assert) \ 104 do { \ 105 if (unlikely(!(assert))) { \ 106 printk(KERN_EMERG \ 107 "Assertion failure in %s() at %s:%d: '%s'\n", \ 108 __func__, __FILE__, __LINE__, #assert); \ 109 BUG(); \ 110 } \ 111 } while (0) 112 113 /* data type for block offset of block group */ 114 typedef int ext4_grpblk_t; 115 116 /* data type for filesystem-wide blocks number */ 117 typedef unsigned long long ext4_fsblk_t; 118 119 /* data type for file logical block number */ 120 typedef __u32 ext4_lblk_t; 121 122 /* data type for block group number */ 123 typedef unsigned int ext4_group_t; 124 125 enum SHIFT_DIRECTION { 126 SHIFT_LEFT = 0, 127 SHIFT_RIGHT, 128 }; 129 130 /* 131 * Number of criterias defined. For each criteria, mballoc has slightly 132 * different way of finding the required blocks nad usually, higher the 133 * criteria the slower the allocation. We start at lower criterias and keep 134 * falling back to higher ones if we are not able to find any blocks. 135 */ 136 #define EXT4_MB_NUM_CRS 4 137 /* 138 * All possible allocation criterias for mballoc 139 */ 140 enum criteria { 141 CR0, 142 CR1, 143 CR2, 144 CR3, 145 }; 146 147 /* 148 * Flags used in mballoc's allocation_context flags field. 149 * 150 * Also used to show what's going on for debugging purposes when the 151 * flag field is exported via the traceport interface 152 */ 153 154 /* prefer goal again. length */ 155 #define EXT4_MB_HINT_MERGE 0x0001 156 /* blocks already reserved */ 157 #define EXT4_MB_HINT_RESERVED 0x0002 158 /* metadata is being allocated */ 159 #define EXT4_MB_HINT_METADATA 0x0004 160 /* first blocks in the file */ 161 #define EXT4_MB_HINT_FIRST 0x0008 162 /* search for the best chunk */ 163 #define EXT4_MB_HINT_BEST 0x0010 164 /* data is being allocated */ 165 #define EXT4_MB_HINT_DATA 0x0020 166 /* don't preallocate (for tails) */ 167 #define EXT4_MB_HINT_NOPREALLOC 0x0040 168 /* allocate for locality group */ 169 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 170 /* allocate goal blocks or none */ 171 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 172 /* goal is meaningful */ 173 #define EXT4_MB_HINT_TRY_GOAL 0x0200 174 /* blocks already pre-reserved by delayed allocation */ 175 #define EXT4_MB_DELALLOC_RESERVED 0x0400 176 /* We are doing stream allocation */ 177 #define EXT4_MB_STREAM_ALLOC 0x0800 178 /* Use reserved root blocks if needed */ 179 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 180 /* Use blocks from reserved pool */ 181 #define EXT4_MB_USE_RESERVED 0x2000 182 /* Do strict check for free blocks while retrying block allocation */ 183 #define EXT4_MB_STRICT_CHECK 0x4000 184 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 185 #define EXT4_MB_CR0_OPTIMIZED 0x8000 186 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 187 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 188 struct ext4_allocation_request { 189 /* target inode for block we're allocating */ 190 struct inode *inode; 191 /* how many blocks we want to allocate */ 192 unsigned int len; 193 /* logical block in target inode */ 194 ext4_lblk_t logical; 195 /* the closest logical allocated block to the left */ 196 ext4_lblk_t lleft; 197 /* the closest logical allocated block to the right */ 198 ext4_lblk_t lright; 199 /* phys. target (a hint) */ 200 ext4_fsblk_t goal; 201 /* phys. block for the closest logical allocated block to the left */ 202 ext4_fsblk_t pleft; 203 /* phys. block for the closest logical allocated block to the right */ 204 ext4_fsblk_t pright; 205 /* flags. see above EXT4_MB_HINT_* */ 206 unsigned int flags; 207 }; 208 209 /* 210 * Logical to physical block mapping, used by ext4_map_blocks() 211 * 212 * This structure is used to pass requests into ext4_map_blocks() as 213 * well as to store the information returned by ext4_map_blocks(). It 214 * takes less room on the stack than a struct buffer_head. 215 */ 216 #define EXT4_MAP_NEW BIT(BH_New) 217 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 218 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 219 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 220 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 221 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 222 223 struct ext4_map_blocks { 224 ext4_fsblk_t m_pblk; 225 ext4_lblk_t m_lblk; 226 unsigned int m_len; 227 unsigned int m_flags; 228 }; 229 230 /* 231 * Block validity checking, system zone rbtree. 232 */ 233 struct ext4_system_blocks { 234 struct rb_root root; 235 struct rcu_head rcu; 236 }; 237 238 /* 239 * Flags for ext4_io_end->flags 240 */ 241 #define EXT4_IO_END_UNWRITTEN 0x0001 242 243 struct ext4_io_end_vec { 244 struct list_head list; /* list of io_end_vec */ 245 loff_t offset; /* offset in the file */ 246 ssize_t size; /* size of the extent */ 247 }; 248 249 /* 250 * For converting unwritten extents on a work queue. 'handle' is used for 251 * buffered writeback. 252 */ 253 typedef struct ext4_io_end { 254 struct list_head list; /* per-file finished IO list */ 255 handle_t *handle; /* handle reserved for extent 256 * conversion */ 257 struct inode *inode; /* file being written to */ 258 struct bio *bio; /* Linked list of completed 259 * bios covering the extent */ 260 unsigned int flag; /* unwritten or not */ 261 refcount_t count; /* reference counter */ 262 struct list_head list_vec; /* list of ext4_io_end_vec */ 263 } ext4_io_end_t; 264 265 struct ext4_io_submit { 266 struct writeback_control *io_wbc; 267 struct bio *io_bio; 268 ext4_io_end_t *io_end; 269 sector_t io_next_block; 270 }; 271 272 /* 273 * Special inodes numbers 274 */ 275 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 276 #define EXT4_ROOT_INO 2 /* Root inode */ 277 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 278 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 279 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 280 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 281 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 282 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 283 284 /* First non-reserved inode for old ext4 filesystems */ 285 #define EXT4_GOOD_OLD_FIRST_INO 11 286 287 /* 288 * Maximal count of links to a file 289 */ 290 #define EXT4_LINK_MAX 65000 291 292 /* 293 * Macro-instructions used to manage several block sizes 294 */ 295 #define EXT4_MIN_BLOCK_SIZE 1024 296 #define EXT4_MAX_BLOCK_SIZE 65536 297 #define EXT4_MIN_BLOCK_LOG_SIZE 10 298 #define EXT4_MAX_BLOCK_LOG_SIZE 16 299 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 300 #ifdef __KERNEL__ 301 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 302 #else 303 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 304 #endif 305 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 306 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 307 EXT4_SB(s)->s_cluster_bits) 308 #ifdef __KERNEL__ 309 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 310 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 311 #else 312 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 313 #endif 314 #ifdef __KERNEL__ 315 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 316 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 317 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 318 #else 319 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 320 EXT4_GOOD_OLD_INODE_SIZE : \ 321 (s)->s_inode_size) 322 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 323 EXT4_GOOD_OLD_FIRST_INO : \ 324 (s)->s_first_ino) 325 #endif 326 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 327 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 328 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 329 blkbits)) 330 331 /* Translate a block number to a cluster number */ 332 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 333 /* Translate a cluster number to a block number */ 334 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 335 /* Translate # of blks to # of clusters */ 336 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 337 (sbi)->s_cluster_bits) 338 /* Mask out the low bits to get the starting block of the cluster */ 339 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 340 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 341 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 342 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 343 /* Fill in the low bits to get the last block of the cluster */ 344 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 345 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 346 /* Get the cluster offset */ 347 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 348 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 349 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 350 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 351 352 /* 353 * Structure of a blocks group descriptor 354 */ 355 struct ext4_group_desc 356 { 357 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 358 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 359 __le32 bg_inode_table_lo; /* Inodes table block */ 360 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 361 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 362 __le16 bg_used_dirs_count_lo; /* Directories count */ 363 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 364 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 365 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 366 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 367 __le16 bg_itable_unused_lo; /* Unused inodes count */ 368 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 369 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 370 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 371 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 372 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 373 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 374 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 375 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 376 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 377 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 378 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 379 __u32 bg_reserved; 380 }; 381 382 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 383 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 384 sizeof(__le16)) 385 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 386 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 387 sizeof(__le16)) 388 389 /* 390 * Structure of a flex block group info 391 */ 392 393 struct flex_groups { 394 atomic64_t free_clusters; 395 atomic_t free_inodes; 396 atomic_t used_dirs; 397 }; 398 399 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 400 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 401 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 402 403 /* 404 * Macro-instructions used to manage group descriptors 405 */ 406 #define EXT4_MIN_DESC_SIZE 32 407 #define EXT4_MIN_DESC_SIZE_64BIT 64 408 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 409 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 410 #ifdef __KERNEL__ 411 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 412 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 413 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 414 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 415 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 416 #else 417 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 418 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 419 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 420 #endif 421 422 /* 423 * Constants relative to the data blocks 424 */ 425 #define EXT4_NDIR_BLOCKS 12 426 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 427 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 428 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 429 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 430 431 /* 432 * Inode flags 433 */ 434 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 435 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 436 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 437 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 438 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 439 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 440 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 441 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 442 /* Reserved for compression usage... */ 443 #define EXT4_DIRTY_FL 0x00000100 444 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 445 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 446 /* nb: was previously EXT2_ECOMPR_FL */ 447 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 448 /* End compression flags --- maybe not all used */ 449 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 450 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 451 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 452 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 453 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 454 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 455 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 456 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 457 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 458 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 459 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 460 461 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 462 463 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 464 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 465 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 466 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 467 468 /* User modifiable flags */ 469 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 470 EXT4_UNRM_FL | \ 471 EXT4_COMPR_FL | \ 472 EXT4_SYNC_FL | \ 473 EXT4_IMMUTABLE_FL | \ 474 EXT4_APPEND_FL | \ 475 EXT4_NODUMP_FL | \ 476 EXT4_NOATIME_FL | \ 477 EXT4_JOURNAL_DATA_FL | \ 478 EXT4_NOTAIL_FL | \ 479 EXT4_DIRSYNC_FL | \ 480 EXT4_TOPDIR_FL | \ 481 EXT4_EXTENTS_FL | \ 482 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 483 EXT4_DAX_FL | \ 484 EXT4_PROJINHERIT_FL | \ 485 EXT4_CASEFOLD_FL) 486 487 /* User visible flags */ 488 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 489 EXT4_DIRTY_FL | \ 490 EXT4_COMPRBLK_FL | \ 491 EXT4_NOCOMPR_FL | \ 492 EXT4_ENCRYPT_FL | \ 493 EXT4_INDEX_FL | \ 494 EXT4_VERITY_FL | \ 495 EXT4_INLINE_DATA_FL) 496 497 /* Flags that should be inherited by new inodes from their parent. */ 498 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 499 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 500 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 501 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 502 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 503 EXT4_DAX_FL) 504 505 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 506 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 507 EXT4_PROJINHERIT_FL)) 508 509 /* Flags that are appropriate for non-directories/regular files. */ 510 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 511 512 /* The only flags that should be swapped */ 513 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 514 515 /* Flags which are mutually exclusive to DAX */ 516 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 517 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 518 519 /* Mask out flags that are inappropriate for the given type of inode. */ 520 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 521 { 522 if (S_ISDIR(mode)) 523 return flags; 524 else if (S_ISREG(mode)) 525 return flags & EXT4_REG_FLMASK; 526 else 527 return flags & EXT4_OTHER_FLMASK; 528 } 529 530 /* 531 * Inode flags used for atomic set/get 532 */ 533 enum { 534 EXT4_INODE_SECRM = 0, /* Secure deletion */ 535 EXT4_INODE_UNRM = 1, /* Undelete */ 536 EXT4_INODE_COMPR = 2, /* Compress file */ 537 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 538 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 539 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 540 EXT4_INODE_NODUMP = 6, /* do not dump file */ 541 EXT4_INODE_NOATIME = 7, /* do not update atime */ 542 /* Reserved for compression usage... */ 543 EXT4_INODE_DIRTY = 8, 544 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 545 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 546 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 547 /* End compression flags --- maybe not all used */ 548 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 549 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 550 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 551 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 552 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 553 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 554 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 555 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 556 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 557 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 558 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 559 EXT4_INODE_DAX = 25, /* Inode is DAX */ 560 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 561 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 562 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 563 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 564 }; 565 566 /* 567 * Since it's pretty easy to mix up bit numbers and hex values, we use a 568 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 569 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 570 * any extra space in the compiled kernel image, otherwise, the build will fail. 571 * It's important that these values are the same, since we are using 572 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 573 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 574 * values found in ext2, ext3 and ext4 filesystems, and of course the values 575 * defined in e2fsprogs. 576 * 577 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 578 */ 579 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 580 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 581 582 static inline void ext4_check_flag_values(void) 583 { 584 CHECK_FLAG_VALUE(SECRM); 585 CHECK_FLAG_VALUE(UNRM); 586 CHECK_FLAG_VALUE(COMPR); 587 CHECK_FLAG_VALUE(SYNC); 588 CHECK_FLAG_VALUE(IMMUTABLE); 589 CHECK_FLAG_VALUE(APPEND); 590 CHECK_FLAG_VALUE(NODUMP); 591 CHECK_FLAG_VALUE(NOATIME); 592 CHECK_FLAG_VALUE(DIRTY); 593 CHECK_FLAG_VALUE(COMPRBLK); 594 CHECK_FLAG_VALUE(NOCOMPR); 595 CHECK_FLAG_VALUE(ENCRYPT); 596 CHECK_FLAG_VALUE(INDEX); 597 CHECK_FLAG_VALUE(IMAGIC); 598 CHECK_FLAG_VALUE(JOURNAL_DATA); 599 CHECK_FLAG_VALUE(NOTAIL); 600 CHECK_FLAG_VALUE(DIRSYNC); 601 CHECK_FLAG_VALUE(TOPDIR); 602 CHECK_FLAG_VALUE(HUGE_FILE); 603 CHECK_FLAG_VALUE(EXTENTS); 604 CHECK_FLAG_VALUE(VERITY); 605 CHECK_FLAG_VALUE(EA_INODE); 606 CHECK_FLAG_VALUE(INLINE_DATA); 607 CHECK_FLAG_VALUE(PROJINHERIT); 608 CHECK_FLAG_VALUE(CASEFOLD); 609 CHECK_FLAG_VALUE(RESERVED); 610 } 611 612 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 613 struct compat_ext4_new_group_input { 614 u32 group; 615 compat_u64 block_bitmap; 616 compat_u64 inode_bitmap; 617 compat_u64 inode_table; 618 u32 blocks_count; 619 u16 reserved_blocks; 620 u16 unused; 621 }; 622 #endif 623 624 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 625 struct ext4_new_group_data { 626 __u32 group; 627 __u64 block_bitmap; 628 __u64 inode_bitmap; 629 __u64 inode_table; 630 __u32 blocks_count; 631 __u16 reserved_blocks; 632 __u16 mdata_blocks; 633 __u32 free_clusters_count; 634 }; 635 636 /* Indexes used to index group tables in ext4_new_group_data */ 637 enum { 638 BLOCK_BITMAP = 0, /* block bitmap */ 639 INODE_BITMAP, /* inode bitmap */ 640 INODE_TABLE, /* inode tables */ 641 GROUP_TABLE_COUNT, 642 }; 643 644 /* 645 * Flags used by ext4_map_blocks() 646 */ 647 /* Allocate any needed blocks and/or convert an unwritten 648 extent to be an initialized ext4 */ 649 #define EXT4_GET_BLOCKS_CREATE 0x0001 650 /* Request the creation of an unwritten extent */ 651 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 652 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 653 EXT4_GET_BLOCKS_CREATE) 654 /* Caller is from the delayed allocation writeout path 655 * finally doing the actual allocation of delayed blocks */ 656 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 657 /* caller is from the direct IO path, request to creation of an 658 unwritten extents if not allocated, split the unwritten 659 extent if blocks has been preallocated already*/ 660 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 661 #define EXT4_GET_BLOCKS_CONVERT 0x0010 662 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 663 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 664 /* Convert extent to initialized after IO complete */ 665 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 666 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 667 /* Eventual metadata allocation (due to growing extent tree) 668 * should not fail, so try to use reserved blocks for that.*/ 669 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 670 /* Don't normalize allocation size (used for fallocate) */ 671 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 672 /* Convert written extents to unwritten */ 673 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 674 /* Write zeros to newly created written extents */ 675 #define EXT4_GET_BLOCKS_ZERO 0x0200 676 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 677 EXT4_GET_BLOCKS_ZERO) 678 /* Caller will submit data before dropping transaction handle. This 679 * allows jbd2 to avoid submitting data before commit. */ 680 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 681 /* Caller is in the atomic contex, find extent if it has been cached */ 682 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 683 684 /* 685 * The bit position of these flags must not overlap with any of the 686 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 687 * read_extent_tree_block(), ext4_split_extent_at(), 688 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 689 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 690 * caching the extents when reading from the extent tree while a 691 * truncate or punch hole operation is in progress. 692 */ 693 #define EXT4_EX_NOCACHE 0x40000000 694 #define EXT4_EX_FORCE_CACHE 0x20000000 695 #define EXT4_EX_NOFAIL 0x10000000 696 697 /* 698 * Flags used by ext4_free_blocks 699 */ 700 #define EXT4_FREE_BLOCKS_METADATA 0x0001 701 #define EXT4_FREE_BLOCKS_FORGET 0x0002 702 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 703 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 704 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 705 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 706 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 707 708 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 709 /* 710 * ioctl commands in 32 bit emulation 711 */ 712 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 713 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 714 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 715 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 716 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 717 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 718 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 719 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 720 #endif 721 722 /* Max physical block we can address w/o extents */ 723 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 724 725 /* Max logical block we can support */ 726 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 727 728 /* 729 * Structure of an inode on the disk 730 */ 731 struct ext4_inode { 732 __le16 i_mode; /* File mode */ 733 __le16 i_uid; /* Low 16 bits of Owner Uid */ 734 __le32 i_size_lo; /* Size in bytes */ 735 __le32 i_atime; /* Access time */ 736 __le32 i_ctime; /* Inode Change time */ 737 __le32 i_mtime; /* Modification time */ 738 __le32 i_dtime; /* Deletion Time */ 739 __le16 i_gid; /* Low 16 bits of Group Id */ 740 __le16 i_links_count; /* Links count */ 741 __le32 i_blocks_lo; /* Blocks count */ 742 __le32 i_flags; /* File flags */ 743 union { 744 struct { 745 __le32 l_i_version; 746 } linux1; 747 struct { 748 __u32 h_i_translator; 749 } hurd1; 750 struct { 751 __u32 m_i_reserved1; 752 } masix1; 753 } osd1; /* OS dependent 1 */ 754 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 755 __le32 i_generation; /* File version (for NFS) */ 756 __le32 i_file_acl_lo; /* File ACL */ 757 __le32 i_size_high; 758 __le32 i_obso_faddr; /* Obsoleted fragment address */ 759 union { 760 struct { 761 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 762 __le16 l_i_file_acl_high; 763 __le16 l_i_uid_high; /* these 2 fields */ 764 __le16 l_i_gid_high; /* were reserved2[0] */ 765 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 766 __le16 l_i_reserved; 767 } linux2; 768 struct { 769 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 770 __u16 h_i_mode_high; 771 __u16 h_i_uid_high; 772 __u16 h_i_gid_high; 773 __u32 h_i_author; 774 } hurd2; 775 struct { 776 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 777 __le16 m_i_file_acl_high; 778 __u32 m_i_reserved2[2]; 779 } masix2; 780 } osd2; /* OS dependent 2 */ 781 __le16 i_extra_isize; 782 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 783 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 784 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 785 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 786 __le32 i_crtime; /* File Creation time */ 787 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 788 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 789 __le32 i_projid; /* Project ID */ 790 }; 791 792 #define EXT4_EPOCH_BITS 2 793 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 794 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 795 796 /* 797 * Extended fields will fit into an inode if the filesystem was formatted 798 * with large inodes (-I 256 or larger) and there are not currently any EAs 799 * consuming all of the available space. For new inodes we always reserve 800 * enough space for the kernel's known extended fields, but for inodes 801 * created with an old kernel this might not have been the case. None of 802 * the extended inode fields is critical for correct filesystem operation. 803 * This macro checks if a certain field fits in the inode. Note that 804 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 805 */ 806 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 807 ((offsetof(typeof(*ext4_inode), field) + \ 808 sizeof((ext4_inode)->field)) \ 809 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 810 (einode)->i_extra_isize)) \ 811 812 /* 813 * We use an encoding that preserves the times for extra epoch "00": 814 * 815 * extra msb of adjust for signed 816 * epoch 32-bit 32-bit tv_sec to 817 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 818 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 819 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 820 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 821 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 822 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 823 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 824 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 825 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 826 * 827 * Note that previous versions of the kernel on 64-bit systems would 828 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 829 * 1970. e2fsck will correct this, assuming that it is run on the 830 * affected filesystem before 2242. 831 */ 832 833 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 834 { 835 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 836 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 837 } 838 839 static inline void ext4_decode_extra_time(struct timespec64 *time, 840 __le32 extra) 841 { 842 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 843 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 844 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 845 } 846 847 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 848 do { \ 849 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 850 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 851 (raw_inode)->xtime ## _extra = \ 852 ext4_encode_extra_time(&(inode)->xtime); \ 853 } \ 854 else \ 855 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 856 } while (0) 857 858 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 859 do { \ 860 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 861 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 862 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 863 (raw_inode)->xtime ## _extra = \ 864 ext4_encode_extra_time(&(einode)->xtime); \ 865 } while (0) 866 867 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 868 do { \ 869 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 870 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 871 ext4_decode_extra_time(&(inode)->xtime, \ 872 raw_inode->xtime ## _extra); \ 873 } \ 874 else \ 875 (inode)->xtime.tv_nsec = 0; \ 876 } while (0) 877 878 879 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 880 do { \ 881 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 882 (einode)->xtime.tv_sec = \ 883 (signed)le32_to_cpu((raw_inode)->xtime); \ 884 else \ 885 (einode)->xtime.tv_sec = 0; \ 886 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 887 ext4_decode_extra_time(&(einode)->xtime, \ 888 raw_inode->xtime ## _extra); \ 889 else \ 890 (einode)->xtime.tv_nsec = 0; \ 891 } while (0) 892 893 #define i_disk_version osd1.linux1.l_i_version 894 895 #if defined(__KERNEL__) || defined(__linux__) 896 #define i_reserved1 osd1.linux1.l_i_reserved1 897 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 898 #define i_blocks_high osd2.linux2.l_i_blocks_high 899 #define i_uid_low i_uid 900 #define i_gid_low i_gid 901 #define i_uid_high osd2.linux2.l_i_uid_high 902 #define i_gid_high osd2.linux2.l_i_gid_high 903 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 904 905 #elif defined(__GNU__) 906 907 #define i_translator osd1.hurd1.h_i_translator 908 #define i_uid_high osd2.hurd2.h_i_uid_high 909 #define i_gid_high osd2.hurd2.h_i_gid_high 910 #define i_author osd2.hurd2.h_i_author 911 912 #elif defined(__masix__) 913 914 #define i_reserved1 osd1.masix1.m_i_reserved1 915 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 916 #define i_reserved2 osd2.masix2.m_i_reserved2 917 918 #endif /* defined(__KERNEL__) || defined(__linux__) */ 919 920 #include "extents_status.h" 921 #include "fast_commit.h" 922 923 /* 924 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 925 * 926 * These are needed to avoid lockdep false positives when we need to 927 * allocate blocks to the quota inode during ext4_map_blocks(), while 928 * holding i_data_sem for a normal (non-quota) inode. Since we don't 929 * do quota tracking for the quota inode, this avoids deadlock (as 930 * well as infinite recursion, since it isn't turtles all the way 931 * down...) 932 * 933 * I_DATA_SEM_NORMAL - Used for most inodes 934 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 935 * where the second inode has larger inode number 936 * than the first 937 * I_DATA_SEM_QUOTA - Used for quota inodes only 938 * I_DATA_SEM_EA - Used for ea_inodes only 939 */ 940 enum { 941 I_DATA_SEM_NORMAL = 0, 942 I_DATA_SEM_OTHER, 943 I_DATA_SEM_QUOTA, 944 I_DATA_SEM_EA 945 }; 946 947 948 /* 949 * fourth extended file system inode data in memory 950 */ 951 struct ext4_inode_info { 952 __le32 i_data[15]; /* unconverted */ 953 __u32 i_dtime; 954 ext4_fsblk_t i_file_acl; 955 956 /* 957 * i_block_group is the number of the block group which contains 958 * this file's inode. Constant across the lifetime of the inode, 959 * it is used for making block allocation decisions - we try to 960 * place a file's data blocks near its inode block, and new inodes 961 * near to their parent directory's inode. 962 */ 963 ext4_group_t i_block_group; 964 ext4_lblk_t i_dir_start_lookup; 965 #if (BITS_PER_LONG < 64) 966 unsigned long i_state_flags; /* Dynamic state flags */ 967 #endif 968 unsigned long i_flags; 969 970 /* 971 * Extended attributes can be read independently of the main file 972 * data. Taking i_rwsem even when reading would cause contention 973 * between readers of EAs and writers of regular file data, so 974 * instead we synchronize on xattr_sem when reading or changing 975 * EAs. 976 */ 977 struct rw_semaphore xattr_sem; 978 979 /* 980 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 981 * i_orphan is used. 982 */ 983 union { 984 struct list_head i_orphan; /* unlinked but open inodes */ 985 unsigned int i_orphan_idx; /* Index in orphan file */ 986 }; 987 988 /* Fast commit related info */ 989 990 /* For tracking dentry create updates */ 991 struct list_head i_fc_dilist; 992 struct list_head i_fc_list; /* 993 * inodes that need fast commit 994 * protected by sbi->s_fc_lock. 995 */ 996 997 /* Start of lblk range that needs to be committed in this fast commit */ 998 ext4_lblk_t i_fc_lblk_start; 999 1000 /* End of lblk range that needs to be committed in this fast commit */ 1001 ext4_lblk_t i_fc_lblk_len; 1002 1003 /* Number of ongoing updates on this inode */ 1004 atomic_t i_fc_updates; 1005 1006 /* Fast commit wait queue for this inode */ 1007 wait_queue_head_t i_fc_wait; 1008 1009 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1010 struct mutex i_fc_lock; 1011 1012 /* 1013 * i_disksize keeps track of what the inode size is ON DISK, not 1014 * in memory. During truncate, i_size is set to the new size by 1015 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1016 * set i_disksize to 0 until the truncate is actually under way. 1017 * 1018 * The intent is that i_disksize always represents the blocks which 1019 * are used by this file. This allows recovery to restart truncate 1020 * on orphans if we crash during truncate. We actually write i_disksize 1021 * into the on-disk inode when writing inodes out, instead of i_size. 1022 * 1023 * The only time when i_disksize and i_size may be different is when 1024 * a truncate is in progress. The only things which change i_disksize 1025 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1026 */ 1027 loff_t i_disksize; 1028 1029 /* 1030 * i_data_sem is for serialising ext4_truncate() against 1031 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1032 * data tree are chopped off during truncate. We can't do that in 1033 * ext4 because whenever we perform intermediate commits during 1034 * truncate, the inode and all the metadata blocks *must* be in a 1035 * consistent state which allows truncation of the orphans to restart 1036 * during recovery. Hence we must fix the get_block-vs-truncate race 1037 * by other means, so we have i_data_sem. 1038 */ 1039 struct rw_semaphore i_data_sem; 1040 struct inode vfs_inode; 1041 struct jbd2_inode *jinode; 1042 1043 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1044 1045 /* 1046 * File creation time. Its function is same as that of 1047 * struct timespec64 i_{a,c,m}time in the generic inode. 1048 */ 1049 struct timespec64 i_crtime; 1050 1051 /* mballoc */ 1052 atomic_t i_prealloc_active; 1053 struct rb_root i_prealloc_node; 1054 rwlock_t i_prealloc_lock; 1055 1056 /* extents status tree */ 1057 struct ext4_es_tree i_es_tree; 1058 rwlock_t i_es_lock; 1059 struct list_head i_es_list; 1060 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1061 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1062 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1063 extents to shrink. Protected by 1064 i_es_lock */ 1065 1066 /* ialloc */ 1067 ext4_group_t i_last_alloc_group; 1068 1069 /* allocation reservation info for delalloc */ 1070 /* In case of bigalloc, this refer to clusters rather than blocks */ 1071 unsigned int i_reserved_data_blocks; 1072 1073 /* pending cluster reservations for bigalloc file systems */ 1074 struct ext4_pending_tree i_pending_tree; 1075 1076 /* on-disk additional length */ 1077 __u16 i_extra_isize; 1078 1079 /* Indicate the inline data space. */ 1080 u16 i_inline_off; 1081 u16 i_inline_size; 1082 1083 #ifdef CONFIG_QUOTA 1084 /* quota space reservation, managed internally by quota code */ 1085 qsize_t i_reserved_quota; 1086 #endif 1087 1088 /* Lock protecting lists below */ 1089 spinlock_t i_completed_io_lock; 1090 /* 1091 * Completed IOs that need unwritten extents handling and have 1092 * transaction reserved 1093 */ 1094 struct list_head i_rsv_conversion_list; 1095 struct work_struct i_rsv_conversion_work; 1096 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1097 1098 spinlock_t i_block_reservation_lock; 1099 1100 /* 1101 * Transactions that contain inode's metadata needed to complete 1102 * fsync and fdatasync, respectively. 1103 */ 1104 tid_t i_sync_tid; 1105 tid_t i_datasync_tid; 1106 1107 #ifdef CONFIG_QUOTA 1108 struct dquot *i_dquot[MAXQUOTAS]; 1109 #endif 1110 1111 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1112 __u32 i_csum_seed; 1113 1114 kprojid_t i_projid; 1115 }; 1116 1117 /* 1118 * File system states 1119 */ 1120 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1121 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1122 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1123 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1124 1125 /* 1126 * Misc. filesystem flags 1127 */ 1128 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1129 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1130 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1131 1132 /* 1133 * Mount flags set via mount options or defaults 1134 */ 1135 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1136 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1137 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1138 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1139 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1140 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1141 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1142 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1143 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1144 #ifdef CONFIG_FS_DAX 1145 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1146 #else 1147 #define EXT4_MOUNT_DAX_ALWAYS 0 1148 #endif 1149 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1150 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1151 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1152 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1153 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1154 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1155 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1156 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1157 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1158 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1159 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1160 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1161 * enable enforcement for hidden 1162 * quota files */ 1163 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1164 * enforcement for hidden quota 1165 * files */ 1166 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1167 * enforcement */ 1168 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1169 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1170 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1171 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1172 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1173 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1174 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1175 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1176 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1177 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1178 1179 /* 1180 * Mount flags set either automatically (could not be set by mount option) 1181 * based on per file system feature or property or in special cases such as 1182 * distinguishing between explicit mount option definition and default. 1183 */ 1184 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1185 specified delalloc */ 1186 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1187 size of blocksize * 8 1188 blocks */ 1189 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1190 file systems */ 1191 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1192 specified journal checksum */ 1193 1194 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1195 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1196 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1197 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1198 * scanning in mballoc 1199 */ 1200 1201 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1202 ~EXT4_MOUNT_##opt 1203 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1204 EXT4_MOUNT_##opt 1205 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1206 EXT4_MOUNT_##opt) 1207 1208 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1209 ~EXT4_MOUNT2_##opt 1210 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1211 EXT4_MOUNT2_##opt 1212 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1213 EXT4_MOUNT2_##opt) 1214 1215 #define ext4_test_and_set_bit __test_and_set_bit_le 1216 #define ext4_set_bit __set_bit_le 1217 #define ext4_set_bit_atomic ext2_set_bit_atomic 1218 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1219 #define ext4_clear_bit __clear_bit_le 1220 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1221 #define ext4_test_bit test_bit_le 1222 #define ext4_find_next_zero_bit find_next_zero_bit_le 1223 #define ext4_find_next_bit find_next_bit_le 1224 1225 extern void mb_set_bits(void *bm, int cur, int len); 1226 1227 /* 1228 * Maximal mount counts between two filesystem checks 1229 */ 1230 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1231 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1232 1233 /* 1234 * Behaviour when detecting errors 1235 */ 1236 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1237 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1238 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1239 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1240 1241 /* Metadata checksum algorithm codes */ 1242 #define EXT4_CRC32C_CHKSUM 1 1243 1244 #define EXT4_LABEL_MAX 16 1245 1246 /* 1247 * Structure of the super block 1248 */ 1249 struct ext4_super_block { 1250 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1251 __le32 s_blocks_count_lo; /* Blocks count */ 1252 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1253 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1254 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1255 __le32 s_first_data_block; /* First Data Block */ 1256 __le32 s_log_block_size; /* Block size */ 1257 __le32 s_log_cluster_size; /* Allocation cluster size */ 1258 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1259 __le32 s_clusters_per_group; /* # Clusters per group */ 1260 __le32 s_inodes_per_group; /* # Inodes per group */ 1261 __le32 s_mtime; /* Mount time */ 1262 /*30*/ __le32 s_wtime; /* Write time */ 1263 __le16 s_mnt_count; /* Mount count */ 1264 __le16 s_max_mnt_count; /* Maximal mount count */ 1265 __le16 s_magic; /* Magic signature */ 1266 __le16 s_state; /* File system state */ 1267 __le16 s_errors; /* Behaviour when detecting errors */ 1268 __le16 s_minor_rev_level; /* minor revision level */ 1269 /*40*/ __le32 s_lastcheck; /* time of last check */ 1270 __le32 s_checkinterval; /* max. time between checks */ 1271 __le32 s_creator_os; /* OS */ 1272 __le32 s_rev_level; /* Revision level */ 1273 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1274 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1275 /* 1276 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1277 * 1278 * Note: the difference between the compatible feature set and 1279 * the incompatible feature set is that if there is a bit set 1280 * in the incompatible feature set that the kernel doesn't 1281 * know about, it should refuse to mount the filesystem. 1282 * 1283 * e2fsck's requirements are more strict; if it doesn't know 1284 * about a feature in either the compatible or incompatible 1285 * feature set, it must abort and not try to meddle with 1286 * things it doesn't understand... 1287 */ 1288 __le32 s_first_ino; /* First non-reserved inode */ 1289 __le16 s_inode_size; /* size of inode structure */ 1290 __le16 s_block_group_nr; /* block group # of this superblock */ 1291 __le32 s_feature_compat; /* compatible feature set */ 1292 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1293 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1294 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1295 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1296 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1297 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1298 /* 1299 * Performance hints. Directory preallocation should only 1300 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1301 */ 1302 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1303 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1304 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1305 /* 1306 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1307 */ 1308 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1309 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1310 __le32 s_journal_dev; /* device number of journal file */ 1311 __le32 s_last_orphan; /* start of list of inodes to delete */ 1312 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1313 __u8 s_def_hash_version; /* Default hash version to use */ 1314 __u8 s_jnl_backup_type; 1315 __le16 s_desc_size; /* size of group descriptor */ 1316 /*100*/ __le32 s_default_mount_opts; 1317 __le32 s_first_meta_bg; /* First metablock block group */ 1318 __le32 s_mkfs_time; /* When the filesystem was created */ 1319 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1320 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1321 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1322 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1323 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1324 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1325 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1326 __le32 s_flags; /* Miscellaneous flags */ 1327 __le16 s_raid_stride; /* RAID stride */ 1328 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1329 __le64 s_mmp_block; /* Block for multi-mount protection */ 1330 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1331 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1332 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1333 __u8 s_encryption_level; /* versioning level for encryption */ 1334 __u8 s_reserved_pad; /* Padding to next 32bits */ 1335 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1336 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1337 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1338 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1339 snapshot's future use */ 1340 __le32 s_snapshot_list; /* inode number of the head of the 1341 on-disk snapshot list */ 1342 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1343 __le32 s_error_count; /* number of fs errors */ 1344 __le32 s_first_error_time; /* first time an error happened */ 1345 __le32 s_first_error_ino; /* inode involved in first error */ 1346 __le64 s_first_error_block; /* block involved of first error */ 1347 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1348 __le32 s_first_error_line; /* line number where error happened */ 1349 __le32 s_last_error_time; /* most recent time of an error */ 1350 __le32 s_last_error_ino; /* inode involved in last error */ 1351 __le32 s_last_error_line; /* line number where error happened */ 1352 __le64 s_last_error_block; /* block involved of last error */ 1353 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1354 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1355 __u8 s_mount_opts[64]; 1356 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1357 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1358 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1359 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1360 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1361 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1362 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1363 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1364 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1365 __u8 s_wtime_hi; 1366 __u8 s_mtime_hi; 1367 __u8 s_mkfs_time_hi; 1368 __u8 s_lastcheck_hi; 1369 __u8 s_first_error_time_hi; 1370 __u8 s_last_error_time_hi; 1371 __u8 s_first_error_errcode; 1372 __u8 s_last_error_errcode; 1373 __le16 s_encoding; /* Filename charset encoding */ 1374 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1375 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1376 __le32 s_reserved[94]; /* Padding to the end of the block */ 1377 __le32 s_checksum; /* crc32c(superblock) */ 1378 }; 1379 1380 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1381 1382 #ifdef __KERNEL__ 1383 1384 /* Number of quota types we support */ 1385 #define EXT4_MAXQUOTAS 3 1386 1387 #define EXT4_ENC_UTF8_12_1 1 1388 1389 /* Types of ext4 journal triggers */ 1390 enum ext4_journal_trigger_type { 1391 EXT4_JTR_ORPHAN_FILE, 1392 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1393 }; 1394 1395 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1396 1397 struct ext4_journal_trigger { 1398 struct jbd2_buffer_trigger_type tr_triggers; 1399 struct super_block *sb; 1400 }; 1401 1402 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1403 struct jbd2_buffer_trigger_type *trigger) 1404 { 1405 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1406 } 1407 1408 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1409 1410 /* Structure at the tail of orphan block */ 1411 struct ext4_orphan_block_tail { 1412 __le32 ob_magic; 1413 __le32 ob_checksum; 1414 }; 1415 1416 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1417 { 1418 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1419 sizeof(u32); 1420 } 1421 1422 struct ext4_orphan_block { 1423 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1424 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1425 }; 1426 1427 /* 1428 * Info about orphan file. 1429 */ 1430 struct ext4_orphan_info { 1431 int of_blocks; /* Number of orphan blocks in a file */ 1432 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1433 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1434 * file blocks */ 1435 }; 1436 1437 /* 1438 * fourth extended-fs super-block data in memory 1439 */ 1440 struct ext4_sb_info { 1441 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1442 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1443 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1444 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1445 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1446 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1447 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1448 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1449 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1450 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1451 unsigned long s_overhead; /* # of fs overhead clusters */ 1452 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1453 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1454 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1455 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1456 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1457 struct buffer_head * __rcu *s_group_desc; 1458 unsigned int s_mount_opt; 1459 unsigned int s_mount_opt2; 1460 unsigned long s_mount_flags; 1461 unsigned int s_def_mount_opt; 1462 unsigned int s_def_mount_opt2; 1463 ext4_fsblk_t s_sb_block; 1464 atomic64_t s_resv_clusters; 1465 kuid_t s_resuid; 1466 kgid_t s_resgid; 1467 unsigned short s_mount_state; 1468 unsigned short s_pad; 1469 int s_addr_per_block_bits; 1470 int s_desc_per_block_bits; 1471 int s_inode_size; 1472 int s_first_ino; 1473 unsigned int s_inode_readahead_blks; 1474 unsigned int s_inode_goal; 1475 u32 s_hash_seed[4]; 1476 int s_def_hash_version; 1477 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1478 struct percpu_counter s_freeclusters_counter; 1479 struct percpu_counter s_freeinodes_counter; 1480 struct percpu_counter s_dirs_counter; 1481 struct percpu_counter s_dirtyclusters_counter; 1482 struct percpu_counter s_sra_exceeded_retry_limit; 1483 struct blockgroup_lock *s_blockgroup_lock; 1484 struct proc_dir_entry *s_proc; 1485 struct kobject s_kobj; 1486 struct completion s_kobj_unregister; 1487 struct super_block *s_sb; 1488 struct buffer_head *s_mmp_bh; 1489 1490 /* Journaling */ 1491 struct journal_s *s_journal; 1492 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1493 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1494 struct list_head s_orphan; /* List of orphaned inodes in on disk 1495 list */ 1496 struct ext4_orphan_info s_orphan_info; 1497 unsigned long s_commit_interval; 1498 u32 s_max_batch_time; 1499 u32 s_min_batch_time; 1500 struct block_device *s_journal_bdev; 1501 #ifdef CONFIG_QUOTA 1502 /* Names of quota files with journalled quota */ 1503 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1504 int s_jquota_fmt; /* Format of quota to use */ 1505 #endif 1506 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1507 struct ext4_system_blocks __rcu *s_system_blks; 1508 1509 #ifdef EXTENTS_STATS 1510 /* ext4 extents stats */ 1511 unsigned long s_ext_min; 1512 unsigned long s_ext_max; 1513 unsigned long s_depth_max; 1514 spinlock_t s_ext_stats_lock; 1515 unsigned long s_ext_blocks; 1516 unsigned long s_ext_extents; 1517 #endif 1518 1519 /* for buddy allocator */ 1520 struct ext4_group_info ** __rcu *s_group_info; 1521 struct inode *s_buddy_cache; 1522 spinlock_t s_md_lock; 1523 unsigned short *s_mb_offsets; 1524 unsigned int *s_mb_maxs; 1525 unsigned int s_group_info_size; 1526 unsigned int s_mb_free_pending; 1527 struct list_head s_freed_data_list; /* List of blocks to be freed 1528 after commit completed */ 1529 struct list_head s_discard_list; 1530 struct work_struct s_discard_work; 1531 atomic_t s_retry_alloc_pending; 1532 struct list_head *s_mb_avg_fragment_size; 1533 rwlock_t *s_mb_avg_fragment_size_locks; 1534 struct list_head *s_mb_largest_free_orders; 1535 rwlock_t *s_mb_largest_free_orders_locks; 1536 1537 /* tunables */ 1538 unsigned long s_stripe; 1539 unsigned int s_mb_max_linear_groups; 1540 unsigned int s_mb_stream_request; 1541 unsigned int s_mb_max_to_scan; 1542 unsigned int s_mb_min_to_scan; 1543 unsigned int s_mb_stats; 1544 unsigned int s_mb_order2_reqs; 1545 unsigned int s_mb_group_prealloc; 1546 unsigned int s_max_dir_size_kb; 1547 /* where last allocation was done - for stream allocation */ 1548 unsigned long s_mb_last_group; 1549 unsigned long s_mb_last_start; 1550 unsigned int s_mb_prefetch; 1551 unsigned int s_mb_prefetch_limit; 1552 1553 /* stats for buddy allocator */ 1554 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1555 atomic_t s_bal_success; /* we found long enough chunks */ 1556 atomic_t s_bal_allocated; /* in blocks */ 1557 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1558 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1559 atomic_t s_bal_goals; /* goal hits */ 1560 atomic_t s_bal_breaks; /* too long searches */ 1561 atomic_t s_bal_2orders; /* 2^order hits */ 1562 atomic_t s_bal_cr0_bad_suggestions; 1563 atomic_t s_bal_cr1_bad_suggestions; 1564 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1565 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1566 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1567 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1568 atomic64_t s_mb_generation_time; 1569 atomic_t s_mb_lost_chunks; 1570 atomic_t s_mb_preallocated; 1571 atomic_t s_mb_discarded; 1572 atomic_t s_lock_busy; 1573 1574 /* locality groups */ 1575 struct ext4_locality_group __percpu *s_locality_groups; 1576 1577 /* for write statistics */ 1578 unsigned long s_sectors_written_start; 1579 u64 s_kbytes_written; 1580 1581 /* the size of zero-out chunk */ 1582 unsigned int s_extent_max_zeroout_kb; 1583 1584 unsigned int s_log_groups_per_flex; 1585 struct flex_groups * __rcu *s_flex_groups; 1586 ext4_group_t s_flex_groups_allocated; 1587 1588 /* workqueue for reserved extent conversions (buffered io) */ 1589 struct workqueue_struct *rsv_conversion_wq; 1590 1591 /* timer for periodic error stats printing */ 1592 struct timer_list s_err_report; 1593 1594 /* Lazy inode table initialization info */ 1595 struct ext4_li_request *s_li_request; 1596 /* Wait multiplier for lazy initialization thread */ 1597 unsigned int s_li_wait_mult; 1598 1599 /* Kernel thread for multiple mount protection */ 1600 struct task_struct *s_mmp_tsk; 1601 1602 /* record the last minlen when FITRIM is called. */ 1603 unsigned long s_last_trim_minblks; 1604 1605 /* Reference to checksum algorithm driver via cryptoapi */ 1606 struct crypto_shash *s_chksum_driver; 1607 1608 /* Precomputed FS UUID checksum for seeding other checksums */ 1609 __u32 s_csum_seed; 1610 1611 /* Reclaim extents from extent status tree */ 1612 struct shrinker s_es_shrinker; 1613 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1614 long s_es_nr_inode; 1615 struct ext4_es_stats s_es_stats; 1616 struct mb_cache *s_ea_block_cache; 1617 struct mb_cache *s_ea_inode_cache; 1618 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1619 1620 /* Journal triggers for checksum computation */ 1621 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1622 1623 /* Ratelimit ext4 messages. */ 1624 struct ratelimit_state s_err_ratelimit_state; 1625 struct ratelimit_state s_warning_ratelimit_state; 1626 struct ratelimit_state s_msg_ratelimit_state; 1627 atomic_t s_warning_count; 1628 atomic_t s_msg_count; 1629 1630 /* Encryption policy for '-o test_dummy_encryption' */ 1631 struct fscrypt_dummy_policy s_dummy_enc_policy; 1632 1633 /* 1634 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1635 * or EXTENTS flag. 1636 */ 1637 struct percpu_rw_semaphore s_writepages_rwsem; 1638 struct dax_device *s_daxdev; 1639 u64 s_dax_part_off; 1640 #ifdef CONFIG_EXT4_DEBUG 1641 unsigned long s_simulate_fail; 1642 #endif 1643 /* Record the errseq of the backing block device */ 1644 errseq_t s_bdev_wb_err; 1645 spinlock_t s_bdev_wb_lock; 1646 1647 /* Information about errors that happened during this mount */ 1648 spinlock_t s_error_lock; 1649 int s_add_error_count; 1650 int s_first_error_code; 1651 __u32 s_first_error_line; 1652 __u32 s_first_error_ino; 1653 __u64 s_first_error_block; 1654 const char *s_first_error_func; 1655 time64_t s_first_error_time; 1656 int s_last_error_code; 1657 __u32 s_last_error_line; 1658 __u32 s_last_error_ino; 1659 __u64 s_last_error_block; 1660 const char *s_last_error_func; 1661 time64_t s_last_error_time; 1662 /* 1663 * If we are in a context where we cannot update error information in 1664 * the on-disk superblock, we queue this work to do it. 1665 */ 1666 struct work_struct s_error_work; 1667 1668 /* Ext4 fast commit sub transaction ID */ 1669 atomic_t s_fc_subtid; 1670 1671 /* 1672 * After commit starts, the main queue gets locked, and the further 1673 * updates get added in the staging queue. 1674 */ 1675 #define FC_Q_MAIN 0 1676 #define FC_Q_STAGING 1 1677 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1678 * that have data changes in them. 1679 */ 1680 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1681 unsigned int s_fc_bytes; 1682 /* 1683 * Main fast commit lock. This lock protects accesses to the 1684 * following fields: 1685 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1686 */ 1687 spinlock_t s_fc_lock; 1688 struct buffer_head *s_fc_bh; 1689 struct ext4_fc_stats s_fc_stats; 1690 tid_t s_fc_ineligible_tid; 1691 #ifdef CONFIG_EXT4_DEBUG 1692 int s_fc_debug_max_replay; 1693 #endif 1694 struct ext4_fc_replay_state s_fc_replay_state; 1695 }; 1696 1697 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1698 { 1699 return sb->s_fs_info; 1700 } 1701 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1702 { 1703 return container_of(inode, struct ext4_inode_info, vfs_inode); 1704 } 1705 1706 static inline int ext4_writepages_down_read(struct super_block *sb) 1707 { 1708 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1709 return memalloc_nofs_save(); 1710 } 1711 1712 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1713 { 1714 memalloc_nofs_restore(ctx); 1715 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1716 } 1717 1718 static inline int ext4_writepages_down_write(struct super_block *sb) 1719 { 1720 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1721 return memalloc_nofs_save(); 1722 } 1723 1724 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1725 { 1726 memalloc_nofs_restore(ctx); 1727 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1728 } 1729 1730 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1731 { 1732 return ino == EXT4_ROOT_INO || 1733 (ino >= EXT4_FIRST_INO(sb) && 1734 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1735 } 1736 1737 /* 1738 * Returns: sbi->field[index] 1739 * Used to access an array element from the following sbi fields which require 1740 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1741 * - s_group_desc 1742 * - s_group_info 1743 * - s_flex_group 1744 */ 1745 #define sbi_array_rcu_deref(sbi, field, index) \ 1746 ({ \ 1747 typeof(*((sbi)->field)) _v; \ 1748 rcu_read_lock(); \ 1749 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1750 rcu_read_unlock(); \ 1751 _v; \ 1752 }) 1753 1754 /* 1755 * run-time mount flags 1756 */ 1757 enum { 1758 EXT4_MF_MNTDIR_SAMPLED, 1759 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1760 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1761 }; 1762 1763 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1764 { 1765 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1766 } 1767 1768 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1769 { 1770 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1771 } 1772 1773 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1774 { 1775 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1776 } 1777 1778 1779 /* 1780 * Simulate_fail codes 1781 */ 1782 #define EXT4_SIM_BBITMAP_EIO 1 1783 #define EXT4_SIM_BBITMAP_CRC 2 1784 #define EXT4_SIM_IBITMAP_EIO 3 1785 #define EXT4_SIM_IBITMAP_CRC 4 1786 #define EXT4_SIM_INODE_EIO 5 1787 #define EXT4_SIM_INODE_CRC 6 1788 #define EXT4_SIM_DIRBLOCK_EIO 7 1789 #define EXT4_SIM_DIRBLOCK_CRC 8 1790 1791 static inline bool ext4_simulate_fail(struct super_block *sb, 1792 unsigned long code) 1793 { 1794 #ifdef CONFIG_EXT4_DEBUG 1795 struct ext4_sb_info *sbi = EXT4_SB(sb); 1796 1797 if (unlikely(sbi->s_simulate_fail == code)) { 1798 sbi->s_simulate_fail = 0; 1799 return true; 1800 } 1801 #endif 1802 return false; 1803 } 1804 1805 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1806 struct buffer_head *bh, 1807 unsigned long code) 1808 { 1809 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1810 clear_buffer_uptodate(bh); 1811 } 1812 1813 /* 1814 * Error number codes for s_{first,last}_error_errno 1815 * 1816 * Linux errno numbers are architecture specific, so we need to translate 1817 * them into something which is architecture independent. We don't define 1818 * codes for all errno's; just the ones which are most likely to be the cause 1819 * of an ext4_error() call. 1820 */ 1821 #define EXT4_ERR_UNKNOWN 1 1822 #define EXT4_ERR_EIO 2 1823 #define EXT4_ERR_ENOMEM 3 1824 #define EXT4_ERR_EFSBADCRC 4 1825 #define EXT4_ERR_EFSCORRUPTED 5 1826 #define EXT4_ERR_ENOSPC 6 1827 #define EXT4_ERR_ENOKEY 7 1828 #define EXT4_ERR_EROFS 8 1829 #define EXT4_ERR_EFBIG 9 1830 #define EXT4_ERR_EEXIST 10 1831 #define EXT4_ERR_ERANGE 11 1832 #define EXT4_ERR_EOVERFLOW 12 1833 #define EXT4_ERR_EBUSY 13 1834 #define EXT4_ERR_ENOTDIR 14 1835 #define EXT4_ERR_ENOTEMPTY 15 1836 #define EXT4_ERR_ESHUTDOWN 16 1837 #define EXT4_ERR_EFAULT 17 1838 1839 /* 1840 * Inode dynamic state flags 1841 */ 1842 enum { 1843 EXT4_STATE_NEW, /* inode is newly created */ 1844 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1845 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1846 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1847 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1848 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1849 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1850 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1851 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1852 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1853 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1854 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1855 }; 1856 1857 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1858 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1859 { \ 1860 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1861 } \ 1862 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1863 { \ 1864 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1865 } \ 1866 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1867 { \ 1868 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1869 } 1870 1871 /* Add these declarations here only so that these functions can be 1872 * found by name. Otherwise, they are very hard to locate. */ 1873 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1874 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1875 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1876 EXT4_INODE_BIT_FNS(flag, flags, 0) 1877 1878 /* Add these declarations here only so that these functions can be 1879 * found by name. Otherwise, they are very hard to locate. */ 1880 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1881 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1882 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1883 #if (BITS_PER_LONG < 64) 1884 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1885 1886 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1887 { 1888 (ei)->i_state_flags = 0; 1889 } 1890 #else 1891 EXT4_INODE_BIT_FNS(state, flags, 32) 1892 1893 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1894 { 1895 /* We depend on the fact that callers will set i_flags */ 1896 } 1897 #endif 1898 #else 1899 /* Assume that user mode programs are passing in an ext4fs superblock, not 1900 * a kernel struct super_block. This will allow us to call the feature-test 1901 * macros from user land. */ 1902 #define EXT4_SB(sb) (sb) 1903 #endif 1904 1905 static inline bool ext4_verity_in_progress(struct inode *inode) 1906 { 1907 return IS_ENABLED(CONFIG_FS_VERITY) && 1908 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1909 } 1910 1911 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1912 1913 /* 1914 * Codes for operating systems 1915 */ 1916 #define EXT4_OS_LINUX 0 1917 #define EXT4_OS_HURD 1 1918 #define EXT4_OS_MASIX 2 1919 #define EXT4_OS_FREEBSD 3 1920 #define EXT4_OS_LITES 4 1921 1922 /* 1923 * Revision levels 1924 */ 1925 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1926 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1927 1928 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1929 1930 #define EXT4_GOOD_OLD_INODE_SIZE 128 1931 1932 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1933 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1934 #define EXT4_TIMESTAMP_MIN S32_MIN 1935 1936 /* 1937 * Feature set definitions 1938 */ 1939 1940 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1941 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1942 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1943 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1944 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1945 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1946 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1947 /* 1948 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1949 * incompatible only if fast commit blocks are present in the FS. Since we 1950 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1951 * incompatible. We also have a JBD2 incompat feature, which gets set when 1952 * there are fast commit blocks present in the journal. 1953 */ 1954 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1955 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1956 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1957 1958 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1959 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1960 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1961 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1962 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1963 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1964 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1965 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1966 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1967 /* 1968 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1969 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1970 * all other data structures' checksums. However, the METADATA_CSUM and 1971 * GDT_CSUM bits are mutually exclusive. 1972 */ 1973 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1974 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1975 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1976 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 1977 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 1978 non-empty */ 1979 1980 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1981 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1982 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1983 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1984 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1985 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1986 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1987 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1988 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1989 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1990 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1991 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1992 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1993 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1994 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1995 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1996 1997 extern void ext4_update_dynamic_rev(struct super_block *sb); 1998 1999 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2000 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2001 { \ 2002 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2003 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2004 } \ 2005 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2006 { \ 2007 ext4_update_dynamic_rev(sb); \ 2008 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2009 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2010 } \ 2011 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2012 { \ 2013 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2014 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2015 } 2016 2017 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2018 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2019 { \ 2020 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2021 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2022 } \ 2023 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2024 { \ 2025 ext4_update_dynamic_rev(sb); \ 2026 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2027 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2028 } \ 2029 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2030 { \ 2031 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2032 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2033 } 2034 2035 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2036 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2037 { \ 2038 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2039 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2040 } \ 2041 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2042 { \ 2043 ext4_update_dynamic_rev(sb); \ 2044 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2045 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2046 } \ 2047 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2048 { \ 2049 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2050 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2051 } 2052 2053 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2054 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2055 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2056 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2057 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2058 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2059 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2060 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2061 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2062 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2063 2064 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2065 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2066 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2067 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2068 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2069 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2070 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2071 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2072 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2073 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2074 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2075 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2076 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2077 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2078 2079 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2080 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2081 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2082 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2083 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2084 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2085 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2086 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2087 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2088 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2089 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2090 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2091 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2092 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2093 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2094 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2095 2096 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2097 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2098 EXT4_FEATURE_INCOMPAT_META_BG) 2099 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2100 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2101 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2102 2103 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2104 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2105 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2106 EXT4_FEATURE_INCOMPAT_META_BG) 2107 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2108 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2109 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2110 2111 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2112 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2113 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2114 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2115 EXT4_FEATURE_INCOMPAT_META_BG| \ 2116 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2117 EXT4_FEATURE_INCOMPAT_64BIT| \ 2118 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2119 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2120 EXT4_FEATURE_INCOMPAT_MMP | \ 2121 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2122 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2123 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2124 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2125 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2126 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2127 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2128 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2129 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2130 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2131 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2132 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2133 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2134 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2135 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2136 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2137 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2138 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2139 2140 #define EXTN_FEATURE_FUNCS(ver) \ 2141 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2142 { \ 2143 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2144 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2145 } \ 2146 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2147 { \ 2148 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2149 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2150 } \ 2151 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2152 { \ 2153 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2154 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2155 } 2156 2157 EXTN_FEATURE_FUNCS(2) 2158 EXTN_FEATURE_FUNCS(3) 2159 EXTN_FEATURE_FUNCS(4) 2160 2161 static inline bool ext4_has_compat_features(struct super_block *sb) 2162 { 2163 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2164 } 2165 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2166 { 2167 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2168 } 2169 static inline bool ext4_has_incompat_features(struct super_block *sb) 2170 { 2171 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2172 } 2173 2174 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2175 2176 /* 2177 * Superblock flags 2178 */ 2179 #define EXT4_FLAGS_RESIZING 0 2180 #define EXT4_FLAGS_SHUTDOWN 1 2181 #define EXT4_FLAGS_BDEV_IS_DAX 2 2182 2183 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2184 { 2185 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2186 } 2187 2188 /* 2189 * Default values for user and/or group using reserved blocks 2190 */ 2191 #define EXT4_DEF_RESUID 0 2192 #define EXT4_DEF_RESGID 0 2193 2194 /* 2195 * Default project ID 2196 */ 2197 #define EXT4_DEF_PROJID 0 2198 2199 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2200 2201 /* 2202 * Default mount options 2203 */ 2204 #define EXT4_DEFM_DEBUG 0x0001 2205 #define EXT4_DEFM_BSDGROUPS 0x0002 2206 #define EXT4_DEFM_XATTR_USER 0x0004 2207 #define EXT4_DEFM_ACL 0x0008 2208 #define EXT4_DEFM_UID16 0x0010 2209 #define EXT4_DEFM_JMODE 0x0060 2210 #define EXT4_DEFM_JMODE_DATA 0x0020 2211 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2212 #define EXT4_DEFM_JMODE_WBACK 0x0060 2213 #define EXT4_DEFM_NOBARRIER 0x0100 2214 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2215 #define EXT4_DEFM_DISCARD 0x0400 2216 #define EXT4_DEFM_NODELALLOC 0x0800 2217 2218 /* 2219 * Default journal batch times 2220 */ 2221 #define EXT4_DEF_MIN_BATCH_TIME 0 2222 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2223 2224 /* 2225 * Minimum number of groups in a flexgroup before we separate out 2226 * directories into the first block group of a flexgroup 2227 */ 2228 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2229 2230 /* 2231 * Structure of a directory entry 2232 */ 2233 #define EXT4_NAME_LEN 255 2234 /* 2235 * Base length of the ext4 directory entry excluding the name length 2236 */ 2237 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2238 2239 struct ext4_dir_entry { 2240 __le32 inode; /* Inode number */ 2241 __le16 rec_len; /* Directory entry length */ 2242 __le16 name_len; /* Name length */ 2243 char name[EXT4_NAME_LEN]; /* File name */ 2244 }; 2245 2246 2247 /* 2248 * Encrypted Casefolded entries require saving the hash on disk. This structure 2249 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2250 * boundary. 2251 */ 2252 struct ext4_dir_entry_hash { 2253 __le32 hash; 2254 __le32 minor_hash; 2255 }; 2256 2257 /* 2258 * The new version of the directory entry. Since EXT4 structures are 2259 * stored in intel byte order, and the name_len field could never be 2260 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2261 * file_type field. 2262 */ 2263 struct ext4_dir_entry_2 { 2264 __le32 inode; /* Inode number */ 2265 __le16 rec_len; /* Directory entry length */ 2266 __u8 name_len; /* Name length */ 2267 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2268 char name[EXT4_NAME_LEN]; /* File name */ 2269 }; 2270 2271 /* 2272 * Access the hashes at the end of ext4_dir_entry_2 2273 */ 2274 #define EXT4_DIRENT_HASHES(entry) \ 2275 ((struct ext4_dir_entry_hash *) \ 2276 (((void *)(entry)) + \ 2277 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2278 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2279 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2280 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2281 2282 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2283 { 2284 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2285 } 2286 2287 /* 2288 * This is a bogus directory entry at the end of each leaf block that 2289 * records checksums. 2290 */ 2291 struct ext4_dir_entry_tail { 2292 __le32 det_reserved_zero1; /* Pretend to be unused */ 2293 __le16 det_rec_len; /* 12 */ 2294 __u8 det_reserved_zero2; /* Zero name length */ 2295 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2296 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2297 }; 2298 2299 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2300 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2301 ((blocksize) - \ 2302 sizeof(struct ext4_dir_entry_tail)))) 2303 2304 /* 2305 * Ext4 directory file types. Only the low 3 bits are used. The 2306 * other bits are reserved for now. 2307 */ 2308 #define EXT4_FT_UNKNOWN 0 2309 #define EXT4_FT_REG_FILE 1 2310 #define EXT4_FT_DIR 2 2311 #define EXT4_FT_CHRDEV 3 2312 #define EXT4_FT_BLKDEV 4 2313 #define EXT4_FT_FIFO 5 2314 #define EXT4_FT_SOCK 6 2315 #define EXT4_FT_SYMLINK 7 2316 2317 #define EXT4_FT_MAX 8 2318 2319 #define EXT4_FT_DIR_CSUM 0xDE 2320 2321 /* 2322 * EXT4_DIR_PAD defines the directory entries boundaries 2323 * 2324 * NOTE: It must be a multiple of 4 2325 */ 2326 #define EXT4_DIR_PAD 4 2327 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2328 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2329 2330 /* 2331 * The rec_len is dependent on the type of directory. Directories that are 2332 * casefolded and encrypted need to store the hash as well, so we add room for 2333 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2334 * pass NULL for dir, as those entries do not use the extra fields. 2335 */ 2336 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2337 const struct inode *dir) 2338 { 2339 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2340 2341 if (dir && ext4_hash_in_dirent(dir)) 2342 rec_len += sizeof(struct ext4_dir_entry_hash); 2343 return (rec_len & ~EXT4_DIR_ROUND); 2344 } 2345 2346 /* 2347 * If we ever get support for fs block sizes > page_size, we'll need 2348 * to remove the #if statements in the next two functions... 2349 */ 2350 static inline unsigned int 2351 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2352 { 2353 unsigned len = le16_to_cpu(dlen); 2354 2355 #if (PAGE_SIZE >= 65536) 2356 if (len == EXT4_MAX_REC_LEN || len == 0) 2357 return blocksize; 2358 return (len & 65532) | ((len & 3) << 16); 2359 #else 2360 return len; 2361 #endif 2362 } 2363 2364 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2365 { 2366 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2367 #if (PAGE_SIZE >= 65536) 2368 if (len < 65536) 2369 return cpu_to_le16(len); 2370 if (len == blocksize) { 2371 if (blocksize == 65536) 2372 return cpu_to_le16(EXT4_MAX_REC_LEN); 2373 else 2374 return cpu_to_le16(0); 2375 } 2376 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2377 #else 2378 return cpu_to_le16(len); 2379 #endif 2380 } 2381 2382 /* 2383 * Hash Tree Directory indexing 2384 * (c) Daniel Phillips, 2001 2385 */ 2386 2387 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2388 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2389 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2390 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2391 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2392 2393 /* Legal values for the dx_root hash_version field: */ 2394 2395 #define DX_HASH_LEGACY 0 2396 #define DX_HASH_HALF_MD4 1 2397 #define DX_HASH_TEA 2 2398 #define DX_HASH_LEGACY_UNSIGNED 3 2399 #define DX_HASH_HALF_MD4_UNSIGNED 4 2400 #define DX_HASH_TEA_UNSIGNED 5 2401 #define DX_HASH_SIPHASH 6 2402 2403 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2404 const void *address, unsigned int length) 2405 { 2406 struct { 2407 struct shash_desc shash; 2408 char ctx[4]; 2409 } desc; 2410 2411 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2412 2413 desc.shash.tfm = sbi->s_chksum_driver; 2414 *(u32 *)desc.ctx = crc; 2415 2416 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2417 2418 return *(u32 *)desc.ctx; 2419 } 2420 2421 #ifdef __KERNEL__ 2422 2423 /* hash info structure used by the directory hash */ 2424 struct dx_hash_info 2425 { 2426 u32 hash; 2427 u32 minor_hash; 2428 int hash_version; 2429 u32 *seed; 2430 }; 2431 2432 2433 /* 32 and 64 bit signed EOF for dx directories */ 2434 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2435 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2436 2437 2438 /* 2439 * Control parameters used by ext4_htree_next_block 2440 */ 2441 #define HASH_NB_ALWAYS 1 2442 2443 struct ext4_filename { 2444 const struct qstr *usr_fname; 2445 struct fscrypt_str disk_name; 2446 struct dx_hash_info hinfo; 2447 #ifdef CONFIG_FS_ENCRYPTION 2448 struct fscrypt_str crypto_buf; 2449 #endif 2450 #if IS_ENABLED(CONFIG_UNICODE) 2451 struct fscrypt_str cf_name; 2452 #endif 2453 }; 2454 2455 #define fname_name(p) ((p)->disk_name.name) 2456 #define fname_usr_name(p) ((p)->usr_fname->name) 2457 #define fname_len(p) ((p)->disk_name.len) 2458 2459 /* 2460 * Describe an inode's exact location on disk and in memory 2461 */ 2462 struct ext4_iloc 2463 { 2464 struct buffer_head *bh; 2465 unsigned long offset; 2466 ext4_group_t block_group; 2467 }; 2468 2469 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2470 { 2471 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2472 } 2473 2474 static inline bool ext4_is_quota_file(struct inode *inode) 2475 { 2476 return IS_NOQUOTA(inode) && 2477 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2478 } 2479 2480 /* 2481 * This structure is stuffed into the struct file's private_data field 2482 * for directories. It is where we put information so that we can do 2483 * readdir operations in hash tree order. 2484 */ 2485 struct dir_private_info { 2486 struct rb_root root; 2487 struct rb_node *curr_node; 2488 struct fname *extra_fname; 2489 loff_t last_pos; 2490 __u32 curr_hash; 2491 __u32 curr_minor_hash; 2492 __u32 next_hash; 2493 }; 2494 2495 /* calculate the first block number of the group */ 2496 static inline ext4_fsblk_t 2497 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2498 { 2499 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2500 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2501 } 2502 2503 /* 2504 * Special error return code only used by dx_probe() and its callers. 2505 */ 2506 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2507 2508 /* htree levels for ext4 */ 2509 #define EXT4_HTREE_LEVEL_COMPAT 2 2510 #define EXT4_HTREE_LEVEL 3 2511 2512 static inline int ext4_dir_htree_level(struct super_block *sb) 2513 { 2514 return ext4_has_feature_largedir(sb) ? 2515 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2516 } 2517 2518 /* 2519 * Timeout and state flag for lazy initialization inode thread. 2520 */ 2521 #define EXT4_DEF_LI_WAIT_MULT 10 2522 #define EXT4_DEF_LI_MAX_START_DELAY 5 2523 #define EXT4_LAZYINIT_QUIT 0x0001 2524 #define EXT4_LAZYINIT_RUNNING 0x0002 2525 2526 /* 2527 * Lazy inode table initialization info 2528 */ 2529 struct ext4_lazy_init { 2530 unsigned long li_state; 2531 struct list_head li_request_list; 2532 struct mutex li_list_mtx; 2533 }; 2534 2535 enum ext4_li_mode { 2536 EXT4_LI_MODE_PREFETCH_BBITMAP, 2537 EXT4_LI_MODE_ITABLE, 2538 }; 2539 2540 struct ext4_li_request { 2541 struct super_block *lr_super; 2542 enum ext4_li_mode lr_mode; 2543 ext4_group_t lr_first_not_zeroed; 2544 ext4_group_t lr_next_group; 2545 struct list_head lr_request; 2546 unsigned long lr_next_sched; 2547 unsigned long lr_timeout; 2548 }; 2549 2550 struct ext4_features { 2551 struct kobject f_kobj; 2552 struct completion f_kobj_unregister; 2553 }; 2554 2555 /* 2556 * This structure will be used for multiple mount protection. It will be 2557 * written into the block number saved in the s_mmp_block field in the 2558 * superblock. Programs that check MMP should assume that if 2559 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2560 * to use the filesystem, regardless of how old the timestamp is. 2561 */ 2562 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2563 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2564 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2565 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2566 2567 struct mmp_struct { 2568 __le32 mmp_magic; /* Magic number for MMP */ 2569 __le32 mmp_seq; /* Sequence no. updated periodically */ 2570 2571 /* 2572 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2573 * purposes and do not affect the correctness of the algorithm 2574 */ 2575 __le64 mmp_time; /* Time last updated */ 2576 char mmp_nodename[64]; /* Node which last updated MMP block */ 2577 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2578 2579 /* 2580 * mmp_check_interval is used to verify if the MMP block has been 2581 * updated on the block device. The value is updated based on the 2582 * maximum time to write the MMP block during an update cycle. 2583 */ 2584 __le16 mmp_check_interval; 2585 2586 __le16 mmp_pad1; 2587 __le32 mmp_pad2[226]; 2588 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2589 }; 2590 2591 /* arguments passed to the mmp thread */ 2592 struct mmpd_data { 2593 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2594 struct super_block *sb; /* super block of the fs */ 2595 }; 2596 2597 /* 2598 * Check interval multiplier 2599 * The MMP block is written every update interval and initially checked every 2600 * update interval x the multiplier (the value is then adapted based on the 2601 * write latency). The reason is that writes can be delayed under load and we 2602 * don't want readers to incorrectly assume that the filesystem is no longer 2603 * in use. 2604 */ 2605 #define EXT4_MMP_CHECK_MULT 2UL 2606 2607 /* 2608 * Minimum interval for MMP checking in seconds. 2609 */ 2610 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2611 2612 /* 2613 * Maximum interval for MMP checking in seconds. 2614 */ 2615 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2616 2617 /* 2618 * Function prototypes 2619 */ 2620 2621 /* 2622 * Ok, these declarations are also in <linux/kernel.h> but none of the 2623 * ext4 source programs needs to include it so they are duplicated here. 2624 */ 2625 # define NORET_TYPE /**/ 2626 # define ATTRIB_NORET __attribute__((noreturn)) 2627 # define NORET_AND noreturn, 2628 2629 /* bitmap.c */ 2630 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2631 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2632 struct ext4_group_desc *gdp, 2633 struct buffer_head *bh, int sz); 2634 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2635 struct ext4_group_desc *gdp, 2636 struct buffer_head *bh, int sz); 2637 void ext4_block_bitmap_csum_set(struct super_block *sb, 2638 struct ext4_group_desc *gdp, 2639 struct buffer_head *bh); 2640 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2641 struct ext4_group_desc *gdp, 2642 struct buffer_head *bh); 2643 2644 /* balloc.c */ 2645 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2646 ext4_fsblk_t blocknr, 2647 ext4_group_t *blockgrpp, 2648 ext4_grpblk_t *offsetp); 2649 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2650 ext4_fsblk_t block); 2651 2652 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2653 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2654 ext4_group_t group); 2655 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2656 ext4_fsblk_t goal, 2657 unsigned int flags, 2658 unsigned long *count, 2659 int *errp); 2660 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2661 s64 nclusters, unsigned int flags); 2662 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2663 extern void ext4_check_blocks_bitmap(struct super_block *); 2664 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2665 ext4_group_t block_group, 2666 struct buffer_head ** bh); 2667 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2668 ext4_group_t group); 2669 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2670 2671 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2672 ext4_group_t block_group, 2673 bool ignore_locked); 2674 extern int ext4_wait_block_bitmap(struct super_block *sb, 2675 ext4_group_t block_group, 2676 struct buffer_head *bh); 2677 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2678 ext4_group_t block_group); 2679 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2680 ext4_group_t block_group, 2681 struct ext4_group_desc *gdp); 2682 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2683 2684 #if IS_ENABLED(CONFIG_UNICODE) 2685 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2686 const struct qstr *iname, 2687 struct ext4_filename *fname); 2688 #endif 2689 2690 /* ext4 encryption related stuff goes here crypto.c */ 2691 #ifdef CONFIG_FS_ENCRYPTION 2692 extern const struct fscrypt_operations ext4_cryptops; 2693 2694 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2695 int lookup, struct ext4_filename *fname); 2696 2697 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2698 struct ext4_filename *fname); 2699 2700 void ext4_fname_free_filename(struct ext4_filename *fname); 2701 2702 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2703 2704 #else /* !CONFIG_FS_ENCRYPTION */ 2705 static inline int ext4_fname_setup_filename(struct inode *dir, 2706 const struct qstr *iname, 2707 int lookup, 2708 struct ext4_filename *fname) 2709 { 2710 int err = 0; 2711 fname->usr_fname = iname; 2712 fname->disk_name.name = (unsigned char *) iname->name; 2713 fname->disk_name.len = iname->len; 2714 2715 #if IS_ENABLED(CONFIG_UNICODE) 2716 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2717 #endif 2718 2719 return err; 2720 } 2721 2722 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2723 struct dentry *dentry, 2724 struct ext4_filename *fname) 2725 { 2726 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2727 } 2728 2729 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2730 { 2731 #if IS_ENABLED(CONFIG_UNICODE) 2732 kfree(fname->cf_name.name); 2733 fname->cf_name.name = NULL; 2734 #endif 2735 } 2736 2737 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2738 void __user *arg) 2739 { 2740 return -EOPNOTSUPP; 2741 } 2742 #endif /* !CONFIG_FS_ENCRYPTION */ 2743 2744 /* dir.c */ 2745 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2746 struct file *, 2747 struct ext4_dir_entry_2 *, 2748 struct buffer_head *, char *, int, 2749 unsigned int); 2750 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2751 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2752 (de), (bh), (buf), (size), (offset))) 2753 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2754 __u32 minor_hash, 2755 struct ext4_dir_entry_2 *dirent, 2756 struct fscrypt_str *ent_name); 2757 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2758 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2759 struct buffer_head *bh, 2760 void *buf, int buf_size, 2761 struct ext4_filename *fname, 2762 struct ext4_dir_entry_2 **dest_de); 2763 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2764 struct ext4_dir_entry_2 *de, 2765 int buf_size, 2766 struct ext4_filename *fname); 2767 static inline void ext4_update_dx_flag(struct inode *inode) 2768 { 2769 if (!ext4_has_feature_dir_index(inode->i_sb) && 2770 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2771 /* ext4_iget() should have caught this... */ 2772 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2773 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2774 } 2775 } 2776 static const unsigned char ext4_filetype_table[] = { 2777 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2778 }; 2779 2780 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2781 { 2782 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2783 return DT_UNKNOWN; 2784 2785 return ext4_filetype_table[filetype]; 2786 } 2787 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2788 void *buf, int buf_size); 2789 2790 /* fsync.c */ 2791 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2792 2793 /* hash.c */ 2794 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2795 struct dx_hash_info *hinfo); 2796 2797 /* ialloc.c */ 2798 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2799 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2800 struct inode *, umode_t, 2801 const struct qstr *qstr, __u32 goal, 2802 uid_t *owner, __u32 i_flags, 2803 int handle_type, unsigned int line_no, 2804 int nblocks); 2805 2806 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2807 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2808 (goal), (owner), i_flags, 0, 0, 0) 2809 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2810 type, nblocks) \ 2811 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2812 0, (type), __LINE__, (nblocks)) 2813 2814 2815 extern void ext4_free_inode(handle_t *, struct inode *); 2816 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2817 extern unsigned long ext4_count_free_inodes(struct super_block *); 2818 extern unsigned long ext4_count_dirs(struct super_block *); 2819 extern void ext4_check_inodes_bitmap(struct super_block *); 2820 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2821 extern int ext4_init_inode_table(struct super_block *sb, 2822 ext4_group_t group, int barrier); 2823 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2824 2825 /* fast_commit.c */ 2826 int ext4_fc_info_show(struct seq_file *seq, void *v); 2827 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2828 void ext4_fc_init_inode(struct inode *inode); 2829 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2830 ext4_lblk_t end); 2831 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2832 struct dentry *dentry); 2833 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2834 struct dentry *dentry); 2835 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2836 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2837 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2838 struct dentry *dentry); 2839 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2840 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2841 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2842 void ext4_fc_start_update(struct inode *inode); 2843 void ext4_fc_stop_update(struct inode *inode); 2844 void ext4_fc_del(struct inode *inode); 2845 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2846 void ext4_fc_replay_cleanup(struct super_block *sb); 2847 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2848 int __init ext4_fc_init_dentry_cache(void); 2849 void ext4_fc_destroy_dentry_cache(void); 2850 int ext4_fc_record_regions(struct super_block *sb, int ino, 2851 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2852 int len, int replay); 2853 2854 /* mballoc.c */ 2855 extern const struct seq_operations ext4_mb_seq_groups_ops; 2856 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2857 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2858 extern int ext4_mb_init(struct super_block *); 2859 extern int ext4_mb_release(struct super_block *); 2860 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2861 struct ext4_allocation_request *, int *); 2862 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2863 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2864 extern int __init ext4_init_mballoc(void); 2865 extern void ext4_exit_mballoc(void); 2866 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2867 ext4_group_t group, 2868 unsigned int nr, int *cnt); 2869 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2870 unsigned int nr); 2871 2872 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2873 struct buffer_head *bh, ext4_fsblk_t block, 2874 unsigned long count, int flags); 2875 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2876 ext4_group_t ngroups); 2877 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2878 ext4_group_t i, struct ext4_group_desc *desc); 2879 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2880 ext4_fsblk_t block, unsigned long count); 2881 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2882 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2883 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2884 int len, int state); 2885 2886 /* inode.c */ 2887 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2888 struct ext4_inode_info *ei); 2889 int ext4_inode_is_fast_symlink(struct inode *inode); 2890 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2891 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2892 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2893 bool wait, struct buffer_head **bhs); 2894 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2895 struct buffer_head *bh_result, int create); 2896 int ext4_get_block(struct inode *inode, sector_t iblock, 2897 struct buffer_head *bh_result, int create); 2898 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2899 struct buffer_head *bh, int create); 2900 int ext4_walk_page_buffers(handle_t *handle, 2901 struct inode *inode, 2902 struct buffer_head *head, 2903 unsigned from, 2904 unsigned to, 2905 int *partial, 2906 int (*fn)(handle_t *handle, struct inode *inode, 2907 struct buffer_head *bh)); 2908 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2909 struct buffer_head *bh); 2910 #define FALL_BACK_TO_NONDELALLOC 1 2911 #define CONVERT_INLINE_DATA 2 2912 2913 typedef enum { 2914 EXT4_IGET_NORMAL = 0, 2915 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2916 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2917 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 2918 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 2919 } ext4_iget_flags; 2920 2921 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2922 ext4_iget_flags flags, const char *function, 2923 unsigned int line); 2924 2925 #define ext4_iget(sb, ino, flags) \ 2926 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2927 2928 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2929 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2930 struct iattr *); 2931 extern u32 ext4_dio_alignment(struct inode *inode); 2932 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2933 struct kstat *, u32, unsigned int); 2934 extern void ext4_evict_inode(struct inode *); 2935 extern void ext4_clear_inode(struct inode *); 2936 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2937 struct kstat *, u32, unsigned int); 2938 extern int ext4_sync_inode(handle_t *, struct inode *); 2939 extern void ext4_dirty_inode(struct inode *, int); 2940 extern int ext4_change_inode_journal_flag(struct inode *, int); 2941 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2942 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2943 struct ext4_iloc *iloc); 2944 extern int ext4_inode_attach_jinode(struct inode *inode); 2945 extern int ext4_can_truncate(struct inode *inode); 2946 extern int ext4_truncate(struct inode *); 2947 extern int ext4_break_layouts(struct inode *); 2948 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2949 extern void ext4_set_inode_flags(struct inode *, bool init); 2950 extern int ext4_alloc_da_blocks(struct inode *inode); 2951 extern void ext4_set_aops(struct inode *inode); 2952 extern int ext4_writepage_trans_blocks(struct inode *); 2953 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 2954 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2955 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2956 loff_t lstart, loff_t lend); 2957 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2958 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2959 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2960 extern void ext4_da_release_space(struct inode *inode, int to_free); 2961 extern void ext4_da_update_reserve_space(struct inode *inode, 2962 int used, int quota_claim); 2963 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2964 ext4_fsblk_t pblk, ext4_lblk_t len); 2965 2966 /* indirect.c */ 2967 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2968 struct ext4_map_blocks *map, int flags); 2969 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2970 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2971 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2972 ext4_lblk_t start, ext4_lblk_t end); 2973 2974 /* ioctl.c */ 2975 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2976 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2977 int ext4_fileattr_set(struct mnt_idmap *idmap, 2978 struct dentry *dentry, struct fileattr *fa); 2979 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 2980 extern void ext4_reset_inode_seed(struct inode *inode); 2981 int ext4_update_overhead(struct super_block *sb, bool force); 2982 2983 /* migrate.c */ 2984 extern int ext4_ext_migrate(struct inode *); 2985 extern int ext4_ind_migrate(struct inode *inode); 2986 2987 /* namei.c */ 2988 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 2989 struct inode *inode); 2990 extern int ext4_dirblock_csum_verify(struct inode *inode, 2991 struct buffer_head *bh); 2992 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2993 __u32 start_minor_hash, __u32 *next_hash); 2994 extern int ext4_search_dir(struct buffer_head *bh, 2995 char *search_buf, 2996 int buf_size, 2997 struct inode *dir, 2998 struct ext4_filename *fname, 2999 unsigned int offset, 3000 struct ext4_dir_entry_2 **res_dir); 3001 extern int ext4_generic_delete_entry(struct inode *dir, 3002 struct ext4_dir_entry_2 *de_del, 3003 struct buffer_head *bh, 3004 void *entry_buf, 3005 int buf_size, 3006 int csum_size); 3007 extern bool ext4_empty_dir(struct inode *inode); 3008 3009 /* resize.c */ 3010 extern void ext4_kvfree_array_rcu(void *to_free); 3011 extern int ext4_group_add(struct super_block *sb, 3012 struct ext4_new_group_data *input); 3013 extern int ext4_group_extend(struct super_block *sb, 3014 struct ext4_super_block *es, 3015 ext4_fsblk_t n_blocks_count); 3016 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3017 extern unsigned int ext4_list_backups(struct super_block *sb, 3018 unsigned int *three, unsigned int *five, 3019 unsigned int *seven); 3020 3021 /* super.c */ 3022 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3023 sector_t block, blk_opf_t op_flags); 3024 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3025 sector_t block); 3026 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3027 bh_end_io_t *end_io); 3028 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3029 bh_end_io_t *end_io); 3030 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3031 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3032 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3033 extern int ext4_calculate_overhead(struct super_block *sb); 3034 extern __le32 ext4_superblock_csum(struct super_block *sb, 3035 struct ext4_super_block *es); 3036 extern void ext4_superblock_csum_set(struct super_block *sb); 3037 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3038 ext4_group_t ngroup); 3039 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3040 char nbuf[16]); 3041 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3042 ext4_group_t block_group, 3043 unsigned int flags); 3044 3045 extern __printf(7, 8) 3046 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3047 int, __u64, const char *, ...); 3048 extern __printf(6, 7) 3049 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3050 ext4_fsblk_t, int, const char *, ...); 3051 extern __printf(5, 6) 3052 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3053 const char *, ...); 3054 extern void __ext4_std_error(struct super_block *, const char *, 3055 unsigned int, int); 3056 extern __printf(4, 5) 3057 void __ext4_warning(struct super_block *, const char *, unsigned int, 3058 const char *, ...); 3059 extern __printf(4, 5) 3060 void __ext4_warning_inode(const struct inode *inode, const char *function, 3061 unsigned int line, const char *fmt, ...); 3062 extern __printf(3, 4) 3063 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3064 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3065 const char *, unsigned int, const char *); 3066 extern __printf(7, 8) 3067 void __ext4_grp_locked_error(const char *, unsigned int, 3068 struct super_block *, ext4_group_t, 3069 unsigned long, ext4_fsblk_t, 3070 const char *, ...); 3071 3072 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3073 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3074 3075 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3076 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3077 3078 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3079 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3080 (fmt), ## a) 3081 3082 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3083 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3084 3085 #define ext4_abort(sb, err, fmt, a...) \ 3086 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3087 3088 #ifdef CONFIG_PRINTK 3089 3090 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3091 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3092 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3093 __ext4_error_inode((inode), (func), (line), (block), \ 3094 (err), (fmt), ##__VA_ARGS__) 3095 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3096 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3097 #define ext4_error(sb, fmt, ...) \ 3098 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3099 ##__VA_ARGS__) 3100 #define ext4_error_err(sb, err, fmt, ...) \ 3101 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3102 ##__VA_ARGS__) 3103 #define ext4_warning(sb, fmt, ...) \ 3104 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3105 #define ext4_warning_inode(inode, fmt, ...) \ 3106 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3107 #define ext4_msg(sb, level, fmt, ...) \ 3108 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3109 #define dump_mmp_msg(sb, mmp, msg) \ 3110 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3111 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3112 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3113 fmt, ##__VA_ARGS__) 3114 3115 #else 3116 3117 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3118 do { \ 3119 no_printk(fmt, ##__VA_ARGS__); \ 3120 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3121 } while (0) 3122 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3123 do { \ 3124 no_printk(fmt, ##__VA_ARGS__); \ 3125 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3126 } while (0) 3127 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3128 do { \ 3129 no_printk(fmt, ##__VA_ARGS__); \ 3130 __ext4_error_file(file, "", 0, block, " "); \ 3131 } while (0) 3132 #define ext4_error(sb, fmt, ...) \ 3133 do { \ 3134 no_printk(fmt, ##__VA_ARGS__); \ 3135 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3136 } while (0) 3137 #define ext4_error_err(sb, err, fmt, ...) \ 3138 do { \ 3139 no_printk(fmt, ##__VA_ARGS__); \ 3140 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3141 } while (0) 3142 #define ext4_warning(sb, fmt, ...) \ 3143 do { \ 3144 no_printk(fmt, ##__VA_ARGS__); \ 3145 __ext4_warning(sb, "", 0, " "); \ 3146 } while (0) 3147 #define ext4_warning_inode(inode, fmt, ...) \ 3148 do { \ 3149 no_printk(fmt, ##__VA_ARGS__); \ 3150 __ext4_warning_inode(inode, "", 0, " "); \ 3151 } while (0) 3152 #define ext4_msg(sb, level, fmt, ...) \ 3153 do { \ 3154 no_printk(fmt, ##__VA_ARGS__); \ 3155 __ext4_msg(sb, "", " "); \ 3156 } while (0) 3157 #define dump_mmp_msg(sb, mmp, msg) \ 3158 __dump_mmp_msg(sb, mmp, "", 0, "") 3159 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3160 do { \ 3161 no_printk(fmt, ##__VA_ARGS__); \ 3162 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3163 } while (0) 3164 3165 #endif 3166 3167 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3168 struct ext4_group_desc *bg); 3169 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3170 struct ext4_group_desc *bg); 3171 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3172 struct ext4_group_desc *bg); 3173 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3174 struct ext4_group_desc *bg); 3175 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3176 struct ext4_group_desc *bg); 3177 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3178 struct ext4_group_desc *bg); 3179 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3180 struct ext4_group_desc *bg); 3181 extern void ext4_block_bitmap_set(struct super_block *sb, 3182 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3183 extern void ext4_inode_bitmap_set(struct super_block *sb, 3184 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3185 extern void ext4_inode_table_set(struct super_block *sb, 3186 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3187 extern void ext4_free_group_clusters_set(struct super_block *sb, 3188 struct ext4_group_desc *bg, 3189 __u32 count); 3190 extern void ext4_free_inodes_set(struct super_block *sb, 3191 struct ext4_group_desc *bg, __u32 count); 3192 extern void ext4_used_dirs_set(struct super_block *sb, 3193 struct ext4_group_desc *bg, __u32 count); 3194 extern void ext4_itable_unused_set(struct super_block *sb, 3195 struct ext4_group_desc *bg, __u32 count); 3196 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3197 struct ext4_group_desc *gdp); 3198 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3199 struct ext4_group_desc *gdp); 3200 extern int ext4_register_li_request(struct super_block *sb, 3201 ext4_group_t first_not_zeroed); 3202 3203 static inline int ext4_has_metadata_csum(struct super_block *sb) 3204 { 3205 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3206 !EXT4_SB(sb)->s_chksum_driver); 3207 3208 return ext4_has_feature_metadata_csum(sb) && 3209 (EXT4_SB(sb)->s_chksum_driver != NULL); 3210 } 3211 3212 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3213 { 3214 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3215 } 3216 3217 #define ext4_read_incompat_64bit_val(es, name) \ 3218 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3219 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3220 le32_to_cpu(es->name##_lo)) 3221 3222 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3223 { 3224 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3225 } 3226 3227 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3228 { 3229 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3230 } 3231 3232 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3233 { 3234 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3235 } 3236 3237 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3238 ext4_fsblk_t blk) 3239 { 3240 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3241 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3242 } 3243 3244 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3245 ext4_fsblk_t blk) 3246 { 3247 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3248 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3249 } 3250 3251 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3252 ext4_fsblk_t blk) 3253 { 3254 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3255 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3256 } 3257 3258 static inline loff_t ext4_isize(struct super_block *sb, 3259 struct ext4_inode *raw_inode) 3260 { 3261 if (ext4_has_feature_largedir(sb) || 3262 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3263 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3264 le32_to_cpu(raw_inode->i_size_lo); 3265 3266 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3267 } 3268 3269 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3270 { 3271 raw_inode->i_size_lo = cpu_to_le32(i_size); 3272 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3273 } 3274 3275 /* 3276 * Reading s_groups_count requires using smp_rmb() afterwards. See 3277 * the locking protocol documented in the comments of ext4_group_add() 3278 * in resize.c 3279 */ 3280 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3281 { 3282 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3283 3284 smp_rmb(); 3285 return ngroups; 3286 } 3287 3288 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3289 ext4_group_t block_group) 3290 { 3291 return block_group >> sbi->s_log_groups_per_flex; 3292 } 3293 3294 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3295 { 3296 return 1 << sbi->s_log_groups_per_flex; 3297 } 3298 3299 #define ext4_std_error(sb, errno) \ 3300 do { \ 3301 if ((errno)) \ 3302 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3303 } while (0) 3304 3305 #ifdef CONFIG_SMP 3306 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3307 * counters. So we need to make sure we have free clusters more 3308 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3309 */ 3310 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3311 #else 3312 #define EXT4_FREECLUSTERS_WATERMARK 0 3313 #endif 3314 3315 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3316 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3317 { 3318 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3319 !inode_is_locked(inode)); 3320 down_write(&EXT4_I(inode)->i_data_sem); 3321 if (newsize > EXT4_I(inode)->i_disksize) 3322 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3323 up_write(&EXT4_I(inode)->i_data_sem); 3324 } 3325 3326 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3327 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3328 { 3329 int changed = 0; 3330 3331 if (newsize > inode->i_size) { 3332 i_size_write(inode, newsize); 3333 changed = 1; 3334 } 3335 if (newsize > EXT4_I(inode)->i_disksize) { 3336 ext4_update_i_disksize(inode, newsize); 3337 changed |= 2; 3338 } 3339 return changed; 3340 } 3341 3342 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3343 loff_t len); 3344 3345 struct ext4_group_info { 3346 unsigned long bb_state; 3347 #ifdef AGGRESSIVE_CHECK 3348 unsigned long bb_check_counter; 3349 #endif 3350 struct rb_root bb_free_root; 3351 ext4_grpblk_t bb_first_free; /* first free block */ 3352 ext4_grpblk_t bb_free; /* total free blocks */ 3353 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3354 int bb_avg_fragment_size_order; /* order of average 3355 fragment in BG */ 3356 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3357 ext4_group_t bb_group; /* Group number */ 3358 struct list_head bb_prealloc_list; 3359 #ifdef DOUBLE_CHECK 3360 void *bb_bitmap; 3361 #endif 3362 struct rw_semaphore alloc_sem; 3363 struct list_head bb_avg_fragment_size_node; 3364 struct list_head bb_largest_free_order_node; 3365 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3366 * regions, index is order. 3367 * bb_counters[3] = 5 means 3368 * 5 free 8-block regions. */ 3369 }; 3370 3371 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3372 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3373 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3374 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3375 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3376 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3377 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3378 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3379 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3380 3381 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3382 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3383 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3384 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3385 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3386 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3387 3388 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3389 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3390 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3391 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3392 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3393 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3394 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3395 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3396 3397 #define EXT4_MAX_CONTENTION 8 3398 #define EXT4_CONTENTION_THRESHOLD 2 3399 3400 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3401 ext4_group_t group) 3402 { 3403 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3404 } 3405 3406 /* 3407 * Returns true if the filesystem is busy enough that attempts to 3408 * access the block group locks has run into contention. 3409 */ 3410 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3411 { 3412 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3413 } 3414 3415 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3416 { 3417 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3418 if (spin_trylock(lock)) 3419 /* 3420 * We're able to grab the lock right away, so drop the 3421 * lock contention counter. 3422 */ 3423 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3424 else { 3425 /* 3426 * The lock is busy, so bump the contention counter, 3427 * and then wait on the spin lock. 3428 */ 3429 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3430 EXT4_MAX_CONTENTION); 3431 spin_lock(lock); 3432 } 3433 } 3434 3435 static inline void ext4_unlock_group(struct super_block *sb, 3436 ext4_group_t group) 3437 { 3438 spin_unlock(ext4_group_lock_ptr(sb, group)); 3439 } 3440 3441 #ifdef CONFIG_QUOTA 3442 static inline bool ext4_quota_capable(struct super_block *sb) 3443 { 3444 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3445 } 3446 3447 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3448 { 3449 struct ext4_sb_info *sbi = EXT4_SB(sb); 3450 3451 return (ext4_has_feature_quota(sb) || 3452 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3453 } 3454 int ext4_enable_quotas(struct super_block *sb); 3455 #endif 3456 3457 /* 3458 * Block validity checking 3459 */ 3460 #define ext4_check_indirect_blockref(inode, bh) \ 3461 ext4_check_blockref(__func__, __LINE__, inode, \ 3462 (__le32 *)(bh)->b_data, \ 3463 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3464 3465 #define ext4_ind_check_inode(inode) \ 3466 ext4_check_blockref(__func__, __LINE__, inode, \ 3467 EXT4_I(inode)->i_data, \ 3468 EXT4_NDIR_BLOCKS) 3469 3470 /* 3471 * Inodes and files operations 3472 */ 3473 3474 /* dir.c */ 3475 extern const struct file_operations ext4_dir_operations; 3476 3477 /* file.c */ 3478 extern const struct inode_operations ext4_file_inode_operations; 3479 extern const struct file_operations ext4_file_operations; 3480 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3481 3482 /* inline.c */ 3483 extern int ext4_get_max_inline_size(struct inode *inode); 3484 extern int ext4_find_inline_data_nolock(struct inode *inode); 3485 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3486 unsigned int len); 3487 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3488 3489 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3490 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3491 struct inode *inode, 3492 loff_t pos, unsigned len, 3493 struct page **pagep); 3494 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3495 unsigned copied, struct folio *folio); 3496 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3497 struct inode *inode, 3498 loff_t pos, unsigned len, 3499 struct page **pagep, 3500 void **fsdata); 3501 extern int ext4_try_add_inline_entry(handle_t *handle, 3502 struct ext4_filename *fname, 3503 struct inode *dir, struct inode *inode); 3504 extern int ext4_try_create_inline_dir(handle_t *handle, 3505 struct inode *parent, 3506 struct inode *inode); 3507 extern int ext4_read_inline_dir(struct file *filp, 3508 struct dir_context *ctx, 3509 int *has_inline_data); 3510 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3511 struct inode *dir, ext4_lblk_t block, 3512 struct dx_hash_info *hinfo, 3513 __u32 start_hash, __u32 start_minor_hash, 3514 int *has_inline_data); 3515 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3516 struct ext4_filename *fname, 3517 struct ext4_dir_entry_2 **res_dir, 3518 int *has_inline_data); 3519 extern int ext4_delete_inline_entry(handle_t *handle, 3520 struct inode *dir, 3521 struct ext4_dir_entry_2 *de_del, 3522 struct buffer_head *bh, 3523 int *has_inline_data); 3524 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3525 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3526 struct ext4_dir_entry_2 **parent_de, 3527 int *retval); 3528 extern void *ext4_read_inline_link(struct inode *inode); 3529 3530 struct iomap; 3531 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3532 3533 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3534 3535 extern int ext4_convert_inline_data(struct inode *inode); 3536 3537 static inline int ext4_has_inline_data(struct inode *inode) 3538 { 3539 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3540 EXT4_I(inode)->i_inline_off; 3541 } 3542 3543 /* namei.c */ 3544 extern const struct inode_operations ext4_dir_inode_operations; 3545 extern const struct inode_operations ext4_special_inode_operations; 3546 extern struct dentry *ext4_get_parent(struct dentry *child); 3547 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3548 struct ext4_dir_entry_2 *de, 3549 int blocksize, int csum_size, 3550 unsigned int parent_ino, int dotdot_real_len); 3551 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3552 unsigned int blocksize); 3553 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3554 struct buffer_head *bh); 3555 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3556 struct inode *inode, struct dentry *dentry); 3557 extern int __ext4_link(struct inode *dir, struct inode *inode, 3558 struct dentry *dentry); 3559 3560 #define S_SHIFT 12 3561 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3562 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3563 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3564 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3565 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3566 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3567 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3568 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3569 }; 3570 3571 static inline void ext4_set_de_type(struct super_block *sb, 3572 struct ext4_dir_entry_2 *de, 3573 umode_t mode) { 3574 if (ext4_has_feature_filetype(sb)) 3575 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3576 } 3577 3578 /* readpages.c */ 3579 extern int ext4_mpage_readpages(struct inode *inode, 3580 struct readahead_control *rac, struct folio *folio); 3581 extern int __init ext4_init_post_read_processing(void); 3582 extern void ext4_exit_post_read_processing(void); 3583 3584 /* symlink.c */ 3585 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3586 extern const struct inode_operations ext4_symlink_inode_operations; 3587 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3588 3589 /* sysfs.c */ 3590 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3591 extern int ext4_register_sysfs(struct super_block *sb); 3592 extern void ext4_unregister_sysfs(struct super_block *sb); 3593 extern int __init ext4_init_sysfs(void); 3594 extern void ext4_exit_sysfs(void); 3595 3596 /* block_validity */ 3597 extern void ext4_release_system_zone(struct super_block *sb); 3598 extern int ext4_setup_system_zone(struct super_block *sb); 3599 extern int __init ext4_init_system_zone(void); 3600 extern void ext4_exit_system_zone(void); 3601 extern int ext4_inode_block_valid(struct inode *inode, 3602 ext4_fsblk_t start_blk, 3603 unsigned int count); 3604 extern int ext4_check_blockref(const char *, unsigned int, 3605 struct inode *, __le32 *, unsigned int); 3606 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3607 ext4_fsblk_t start_blk, unsigned int count); 3608 3609 3610 /* extents.c */ 3611 struct ext4_ext_path; 3612 struct ext4_extent; 3613 3614 /* 3615 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3616 * __le32. 3617 */ 3618 #define EXT_MAX_BLOCKS 0xffffffff 3619 3620 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3621 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3622 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3623 struct ext4_map_blocks *map, int flags); 3624 extern int ext4_ext_truncate(handle_t *, struct inode *); 3625 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3626 ext4_lblk_t end); 3627 extern void ext4_ext_init(struct super_block *); 3628 extern void ext4_ext_release(struct super_block *); 3629 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3630 loff_t len); 3631 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3632 loff_t offset, ssize_t len); 3633 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3634 ext4_io_end_t *io_end); 3635 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3636 struct ext4_map_blocks *map, int flags); 3637 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3638 int num, 3639 struct ext4_ext_path *path); 3640 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3641 struct ext4_ext_path **, 3642 struct ext4_extent *, int); 3643 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3644 struct ext4_ext_path **, 3645 int flags); 3646 extern void ext4_free_ext_path(struct ext4_ext_path *); 3647 extern int ext4_ext_check_inode(struct inode *inode); 3648 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3649 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3650 __u64 start, __u64 len); 3651 extern int ext4_get_es_cache(struct inode *inode, 3652 struct fiemap_extent_info *fieinfo, 3653 __u64 start, __u64 len); 3654 extern int ext4_ext_precache(struct inode *inode); 3655 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3656 struct inode *inode2, ext4_lblk_t lblk1, 3657 ext4_lblk_t lblk2, ext4_lblk_t count, 3658 int mark_unwritten,int *err); 3659 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3660 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3661 int check_cred, int restart_cred, 3662 int revoke_cred); 3663 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3664 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3665 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3666 int len, int unwritten, ext4_fsblk_t pblk); 3667 extern int ext4_ext_clear_bb(struct inode *inode); 3668 3669 3670 /* move_extent.c */ 3671 extern void ext4_double_down_write_data_sem(struct inode *first, 3672 struct inode *second); 3673 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3674 struct inode *donor_inode); 3675 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3676 __u64 start_orig, __u64 start_donor, 3677 __u64 len, __u64 *moved_len); 3678 3679 /* page-io.c */ 3680 extern int __init ext4_init_pageio(void); 3681 extern void ext4_exit_pageio(void); 3682 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3683 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3684 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3685 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3686 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3687 struct writeback_control *wbc); 3688 extern void ext4_end_io_rsv_work(struct work_struct *work); 3689 extern void ext4_io_submit(struct ext4_io_submit *io); 3690 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3691 size_t len); 3692 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3693 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3694 3695 /* mmp.c */ 3696 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3697 3698 /* mmp.c */ 3699 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3700 3701 /* verity.c */ 3702 extern const struct fsverity_operations ext4_verityops; 3703 3704 /* orphan.c */ 3705 extern int ext4_orphan_add(handle_t *, struct inode *); 3706 extern int ext4_orphan_del(handle_t *, struct inode *); 3707 extern void ext4_orphan_cleanup(struct super_block *sb, 3708 struct ext4_super_block *es); 3709 extern void ext4_release_orphan_info(struct super_block *sb); 3710 extern int ext4_init_orphan_info(struct super_block *sb); 3711 extern int ext4_orphan_file_empty(struct super_block *sb); 3712 extern void ext4_orphan_file_block_trigger( 3713 struct jbd2_buffer_trigger_type *triggers, 3714 struct buffer_head *bh, 3715 void *data, size_t size); 3716 3717 /* 3718 * Add new method to test whether block and inode bitmaps are properly 3719 * initialized. With uninit_bg reading the block from disk is not enough 3720 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3721 */ 3722 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3723 3724 static inline int bitmap_uptodate(struct buffer_head *bh) 3725 { 3726 return (buffer_uptodate(bh) && 3727 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3728 } 3729 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3730 { 3731 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3732 } 3733 3734 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3735 3736 /* For ioend & aio unwritten conversion wait queues */ 3737 #define EXT4_WQ_HASH_SZ 37 3738 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3739 EXT4_WQ_HASH_SZ]) 3740 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3741 3742 extern int ext4_resize_begin(struct super_block *sb); 3743 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3744 3745 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3746 struct ext4_io_end *io_end) 3747 { 3748 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3749 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3750 atomic_inc(&EXT4_I(inode)->i_unwritten); 3751 } 3752 } 3753 3754 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3755 { 3756 struct inode *inode = io_end->inode; 3757 3758 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3759 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3760 /* Wake up anyone waiting on unwritten extent conversion */ 3761 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3762 wake_up_all(ext4_ioend_wq(inode)); 3763 } 3764 } 3765 3766 extern const struct iomap_ops ext4_iomap_ops; 3767 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3768 extern const struct iomap_ops ext4_iomap_report_ops; 3769 3770 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3771 { 3772 /* 3773 * If the buffer has the write error flag, we have failed 3774 * to write out data in the block. In this case, we don't 3775 * have to read the block because we may read the old data 3776 * successfully. 3777 */ 3778 if (buffer_write_io_error(bh)) 3779 set_buffer_uptodate(bh); 3780 return buffer_uptodate(bh); 3781 } 3782 3783 #endif /* __KERNEL__ */ 3784 3785 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3786 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3787 3788 #endif /* _EXT4_H */ 3789