1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/refcount.h> 21 #include <linux/types.h> 22 #include <linux/blkdev.h> 23 #include <linux/magic.h> 24 #include <linux/jbd2.h> 25 #include <linux/quota.h> 26 #include <linux/rwsem.h> 27 #include <linux/rbtree.h> 28 #include <linux/seqlock.h> 29 #include <linux/mutex.h> 30 #include <linux/timer.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 44 #include <linux/fscrypt.h> 45 #include <linux/fsverity.h> 46 47 #include <linux/compiler.h> 48 49 /* 50 * The fourth extended filesystem constants/structures 51 */ 52 53 /* 54 * with AGGRESSIVE_CHECK allocator runs consistency checks over 55 * structures. these checks slow things down a lot 56 */ 57 #define AGGRESSIVE_CHECK__ 58 59 /* 60 * with DOUBLE_CHECK defined mballoc creates persistent in-core 61 * bitmaps, maintains and uses them to check for double allocations 62 */ 63 #define DOUBLE_CHECK__ 64 65 /* 66 * Define EXT4FS_DEBUG to produce debug messages 67 */ 68 #undef EXT4FS_DEBUG 69 70 /* 71 * Debug code 72 */ 73 #ifdef EXT4FS_DEBUG 74 #define ext4_debug(f, a...) \ 75 do { \ 76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 77 __FILE__, __LINE__, __func__); \ 78 printk(KERN_DEBUG f, ## a); \ 79 } while (0) 80 #else 81 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 82 #endif 83 84 /* 85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 86 */ 87 #define EXT_DEBUG__ 88 89 /* 90 * Dynamic printk for controlled extents debugging. 91 */ 92 #ifdef CONFIG_EXT4_DEBUG 93 #define ext_debug(ino, fmt, ...) \ 94 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 95 current->comm, task_pid_nr(current), \ 96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 97 __func__, ##__VA_ARGS__) 98 #else 99 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 100 #endif 101 102 #define ASSERT(assert) \ 103 do { \ 104 if (unlikely(!(assert))) { \ 105 printk(KERN_EMERG \ 106 "Assertion failure in %s() at %s:%d: '%s'\n", \ 107 __func__, __FILE__, __LINE__, #assert); \ 108 BUG(); \ 109 } \ 110 } while (0) 111 112 /* data type for block offset of block group */ 113 typedef int ext4_grpblk_t; 114 115 /* data type for filesystem-wide blocks number */ 116 typedef unsigned long long ext4_fsblk_t; 117 118 /* data type for file logical block number */ 119 typedef __u32 ext4_lblk_t; 120 121 /* data type for block group number */ 122 typedef unsigned int ext4_group_t; 123 124 enum SHIFT_DIRECTION { 125 SHIFT_LEFT = 0, 126 SHIFT_RIGHT, 127 }; 128 129 /* 130 * Flags used in mballoc's allocation_context flags field. 131 * 132 * Also used to show what's going on for debugging purposes when the 133 * flag field is exported via the traceport interface 134 */ 135 136 /* prefer goal again. length */ 137 #define EXT4_MB_HINT_MERGE 0x0001 138 /* blocks already reserved */ 139 #define EXT4_MB_HINT_RESERVED 0x0002 140 /* metadata is being allocated */ 141 #define EXT4_MB_HINT_METADATA 0x0004 142 /* first blocks in the file */ 143 #define EXT4_MB_HINT_FIRST 0x0008 144 /* search for the best chunk */ 145 #define EXT4_MB_HINT_BEST 0x0010 146 /* data is being allocated */ 147 #define EXT4_MB_HINT_DATA 0x0020 148 /* don't preallocate (for tails) */ 149 #define EXT4_MB_HINT_NOPREALLOC 0x0040 150 /* allocate for locality group */ 151 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 152 /* allocate goal blocks or none */ 153 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 154 /* goal is meaningful */ 155 #define EXT4_MB_HINT_TRY_GOAL 0x0200 156 /* blocks already pre-reserved by delayed allocation */ 157 #define EXT4_MB_DELALLOC_RESERVED 0x0400 158 /* We are doing stream allocation */ 159 #define EXT4_MB_STREAM_ALLOC 0x0800 160 /* Use reserved root blocks if needed */ 161 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 162 /* Use blocks from reserved pool */ 163 #define EXT4_MB_USE_RESERVED 0x2000 164 /* Do strict check for free blocks while retrying block allocation */ 165 #define EXT4_MB_STRICT_CHECK 0x4000 166 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 167 #define EXT4_MB_CR0_OPTIMIZED 0x8000 168 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 169 #define EXT4_MB_CR1_OPTIMIZED 0x00010000 170 /* Perform linear traversal for one group */ 171 #define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000 172 struct ext4_allocation_request { 173 /* target inode for block we're allocating */ 174 struct inode *inode; 175 /* how many blocks we want to allocate */ 176 unsigned int len; 177 /* logical block in target inode */ 178 ext4_lblk_t logical; 179 /* the closest logical allocated block to the left */ 180 ext4_lblk_t lleft; 181 /* the closest logical allocated block to the right */ 182 ext4_lblk_t lright; 183 /* phys. target (a hint) */ 184 ext4_fsblk_t goal; 185 /* phys. block for the closest logical allocated block to the left */ 186 ext4_fsblk_t pleft; 187 /* phys. block for the closest logical allocated block to the right */ 188 ext4_fsblk_t pright; 189 /* flags. see above EXT4_MB_HINT_* */ 190 unsigned int flags; 191 }; 192 193 /* 194 * Logical to physical block mapping, used by ext4_map_blocks() 195 * 196 * This structure is used to pass requests into ext4_map_blocks() as 197 * well as to store the information returned by ext4_map_blocks(). It 198 * takes less room on the stack than a struct buffer_head. 199 */ 200 #define EXT4_MAP_NEW BIT(BH_New) 201 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 202 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 203 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 204 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 205 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 206 207 struct ext4_map_blocks { 208 ext4_fsblk_t m_pblk; 209 ext4_lblk_t m_lblk; 210 unsigned int m_len; 211 unsigned int m_flags; 212 }; 213 214 /* 215 * Block validity checking, system zone rbtree. 216 */ 217 struct ext4_system_blocks { 218 struct rb_root root; 219 struct rcu_head rcu; 220 }; 221 222 /* 223 * Flags for ext4_io_end->flags 224 */ 225 #define EXT4_IO_END_UNWRITTEN 0x0001 226 227 struct ext4_io_end_vec { 228 struct list_head list; /* list of io_end_vec */ 229 loff_t offset; /* offset in the file */ 230 ssize_t size; /* size of the extent */ 231 }; 232 233 /* 234 * For converting unwritten extents on a work queue. 'handle' is used for 235 * buffered writeback. 236 */ 237 typedef struct ext4_io_end { 238 struct list_head list; /* per-file finished IO list */ 239 handle_t *handle; /* handle reserved for extent 240 * conversion */ 241 struct inode *inode; /* file being written to */ 242 struct bio *bio; /* Linked list of completed 243 * bios covering the extent */ 244 unsigned int flag; /* unwritten or not */ 245 refcount_t count; /* reference counter */ 246 struct list_head list_vec; /* list of ext4_io_end_vec */ 247 } ext4_io_end_t; 248 249 struct ext4_io_submit { 250 struct writeback_control *io_wbc; 251 struct bio *io_bio; 252 ext4_io_end_t *io_end; 253 sector_t io_next_block; 254 }; 255 256 /* 257 * Special inodes numbers 258 */ 259 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 260 #define EXT4_ROOT_INO 2 /* Root inode */ 261 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 262 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 263 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 264 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 265 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 266 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 267 268 /* First non-reserved inode for old ext4 filesystems */ 269 #define EXT4_GOOD_OLD_FIRST_INO 11 270 271 /* 272 * Maximal count of links to a file 273 */ 274 #define EXT4_LINK_MAX 65000 275 276 /* 277 * Macro-instructions used to manage several block sizes 278 */ 279 #define EXT4_MIN_BLOCK_SIZE 1024 280 #define EXT4_MAX_BLOCK_SIZE 65536 281 #define EXT4_MIN_BLOCK_LOG_SIZE 10 282 #define EXT4_MAX_BLOCK_LOG_SIZE 16 283 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 284 #ifdef __KERNEL__ 285 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 286 #else 287 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 288 #endif 289 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 290 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 291 EXT4_SB(s)->s_cluster_bits) 292 #ifdef __KERNEL__ 293 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 294 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 295 #else 296 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 297 #endif 298 #ifdef __KERNEL__ 299 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 300 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 301 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 302 #else 303 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 304 EXT4_GOOD_OLD_INODE_SIZE : \ 305 (s)->s_inode_size) 306 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 307 EXT4_GOOD_OLD_FIRST_INO : \ 308 (s)->s_first_ino) 309 #endif 310 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 311 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 312 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 313 blkbits)) 314 315 /* Translate a block number to a cluster number */ 316 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 317 /* Translate a cluster number to a block number */ 318 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 319 /* Translate # of blks to # of clusters */ 320 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 321 (sbi)->s_cluster_bits) 322 /* Mask out the low bits to get the starting block of the cluster */ 323 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 324 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 325 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 326 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 327 /* Fill in the low bits to get the last block of the cluster */ 328 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 329 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 330 /* Get the cluster offset */ 331 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 332 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 333 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 334 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 335 336 /* 337 * Structure of a blocks group descriptor 338 */ 339 struct ext4_group_desc 340 { 341 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 342 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 343 __le32 bg_inode_table_lo; /* Inodes table block */ 344 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 345 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 346 __le16 bg_used_dirs_count_lo; /* Directories count */ 347 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 348 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 349 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 350 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 351 __le16 bg_itable_unused_lo; /* Unused inodes count */ 352 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 353 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 354 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 355 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 356 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 357 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 358 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 359 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 360 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 361 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 362 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 363 __u32 bg_reserved; 364 }; 365 366 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 367 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 368 sizeof(__le16)) 369 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 370 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 371 sizeof(__le16)) 372 373 /* 374 * Structure of a flex block group info 375 */ 376 377 struct flex_groups { 378 atomic64_t free_clusters; 379 atomic_t free_inodes; 380 atomic_t used_dirs; 381 }; 382 383 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 384 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 385 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 386 387 /* 388 * Macro-instructions used to manage group descriptors 389 */ 390 #define EXT4_MIN_DESC_SIZE 32 391 #define EXT4_MIN_DESC_SIZE_64BIT 64 392 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 393 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 394 #ifdef __KERNEL__ 395 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 396 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 397 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 398 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 399 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 400 #else 401 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 402 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 403 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 404 #endif 405 406 /* 407 * Constants relative to the data blocks 408 */ 409 #define EXT4_NDIR_BLOCKS 12 410 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 411 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 412 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 413 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 414 415 /* 416 * Inode flags 417 */ 418 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 419 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 420 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 421 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 422 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 423 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 424 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 425 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 426 /* Reserved for compression usage... */ 427 #define EXT4_DIRTY_FL 0x00000100 428 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 429 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 430 /* nb: was previously EXT2_ECOMPR_FL */ 431 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 432 /* End compression flags --- maybe not all used */ 433 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 434 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 435 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 436 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 437 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 438 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 439 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 440 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 441 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 442 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 443 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 444 445 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 446 447 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 448 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 449 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 450 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 451 452 /* User modifiable flags */ 453 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 454 EXT4_UNRM_FL | \ 455 EXT4_COMPR_FL | \ 456 EXT4_SYNC_FL | \ 457 EXT4_IMMUTABLE_FL | \ 458 EXT4_APPEND_FL | \ 459 EXT4_NODUMP_FL | \ 460 EXT4_NOATIME_FL | \ 461 EXT4_JOURNAL_DATA_FL | \ 462 EXT4_NOTAIL_FL | \ 463 EXT4_DIRSYNC_FL | \ 464 EXT4_TOPDIR_FL | \ 465 EXT4_EXTENTS_FL | \ 466 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 467 EXT4_DAX_FL | \ 468 EXT4_PROJINHERIT_FL | \ 469 EXT4_CASEFOLD_FL) 470 471 /* User visible flags */ 472 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 473 EXT4_DIRTY_FL | \ 474 EXT4_COMPRBLK_FL | \ 475 EXT4_NOCOMPR_FL | \ 476 EXT4_ENCRYPT_FL | \ 477 EXT4_INDEX_FL | \ 478 EXT4_VERITY_FL | \ 479 EXT4_INLINE_DATA_FL) 480 481 /* Flags that should be inherited by new inodes from their parent. */ 482 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 483 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 484 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 485 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 486 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 487 EXT4_DAX_FL) 488 489 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 490 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 491 EXT4_PROJINHERIT_FL)) 492 493 /* Flags that are appropriate for non-directories/regular files. */ 494 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 495 496 /* The only flags that should be swapped */ 497 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 498 499 /* Flags which are mutually exclusive to DAX */ 500 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 501 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 502 503 /* Mask out flags that are inappropriate for the given type of inode. */ 504 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 505 { 506 if (S_ISDIR(mode)) 507 return flags; 508 else if (S_ISREG(mode)) 509 return flags & EXT4_REG_FLMASK; 510 else 511 return flags & EXT4_OTHER_FLMASK; 512 } 513 514 /* 515 * Inode flags used for atomic set/get 516 */ 517 enum { 518 EXT4_INODE_SECRM = 0, /* Secure deletion */ 519 EXT4_INODE_UNRM = 1, /* Undelete */ 520 EXT4_INODE_COMPR = 2, /* Compress file */ 521 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 522 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 523 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 524 EXT4_INODE_NODUMP = 6, /* do not dump file */ 525 EXT4_INODE_NOATIME = 7, /* do not update atime */ 526 /* Reserved for compression usage... */ 527 EXT4_INODE_DIRTY = 8, 528 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 529 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 530 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 531 /* End compression flags --- maybe not all used */ 532 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 533 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 534 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 535 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 536 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 537 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 538 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 539 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 540 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 541 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 542 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 543 EXT4_INODE_DAX = 25, /* Inode is DAX */ 544 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 545 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 546 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 547 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 548 }; 549 550 /* 551 * Since it's pretty easy to mix up bit numbers and hex values, we use a 552 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 553 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 554 * any extra space in the compiled kernel image, otherwise, the build will fail. 555 * It's important that these values are the same, since we are using 556 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 557 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 558 * values found in ext2, ext3 and ext4 filesystems, and of course the values 559 * defined in e2fsprogs. 560 * 561 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 562 */ 563 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 564 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 565 566 static inline void ext4_check_flag_values(void) 567 { 568 CHECK_FLAG_VALUE(SECRM); 569 CHECK_FLAG_VALUE(UNRM); 570 CHECK_FLAG_VALUE(COMPR); 571 CHECK_FLAG_VALUE(SYNC); 572 CHECK_FLAG_VALUE(IMMUTABLE); 573 CHECK_FLAG_VALUE(APPEND); 574 CHECK_FLAG_VALUE(NODUMP); 575 CHECK_FLAG_VALUE(NOATIME); 576 CHECK_FLAG_VALUE(DIRTY); 577 CHECK_FLAG_VALUE(COMPRBLK); 578 CHECK_FLAG_VALUE(NOCOMPR); 579 CHECK_FLAG_VALUE(ENCRYPT); 580 CHECK_FLAG_VALUE(INDEX); 581 CHECK_FLAG_VALUE(IMAGIC); 582 CHECK_FLAG_VALUE(JOURNAL_DATA); 583 CHECK_FLAG_VALUE(NOTAIL); 584 CHECK_FLAG_VALUE(DIRSYNC); 585 CHECK_FLAG_VALUE(TOPDIR); 586 CHECK_FLAG_VALUE(HUGE_FILE); 587 CHECK_FLAG_VALUE(EXTENTS); 588 CHECK_FLAG_VALUE(VERITY); 589 CHECK_FLAG_VALUE(EA_INODE); 590 CHECK_FLAG_VALUE(INLINE_DATA); 591 CHECK_FLAG_VALUE(PROJINHERIT); 592 CHECK_FLAG_VALUE(CASEFOLD); 593 CHECK_FLAG_VALUE(RESERVED); 594 } 595 596 /* Used to pass group descriptor data when online resize is done */ 597 struct ext4_new_group_input { 598 __u32 group; /* Group number for this data */ 599 __u64 block_bitmap; /* Absolute block number of block bitmap */ 600 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 601 __u64 inode_table; /* Absolute block number of inode table start */ 602 __u32 blocks_count; /* Total number of blocks in this group */ 603 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 604 __u16 unused; 605 }; 606 607 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 608 struct compat_ext4_new_group_input { 609 u32 group; 610 compat_u64 block_bitmap; 611 compat_u64 inode_bitmap; 612 compat_u64 inode_table; 613 u32 blocks_count; 614 u16 reserved_blocks; 615 u16 unused; 616 }; 617 #endif 618 619 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 620 struct ext4_new_group_data { 621 __u32 group; 622 __u64 block_bitmap; 623 __u64 inode_bitmap; 624 __u64 inode_table; 625 __u32 blocks_count; 626 __u16 reserved_blocks; 627 __u16 mdata_blocks; 628 __u32 free_clusters_count; 629 }; 630 631 /* Indexes used to index group tables in ext4_new_group_data */ 632 enum { 633 BLOCK_BITMAP = 0, /* block bitmap */ 634 INODE_BITMAP, /* inode bitmap */ 635 INODE_TABLE, /* inode tables */ 636 GROUP_TABLE_COUNT, 637 }; 638 639 /* 640 * Flags used by ext4_map_blocks() 641 */ 642 /* Allocate any needed blocks and/or convert an unwritten 643 extent to be an initialized ext4 */ 644 #define EXT4_GET_BLOCKS_CREATE 0x0001 645 /* Request the creation of an unwritten extent */ 646 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 647 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 648 EXT4_GET_BLOCKS_CREATE) 649 /* Caller is from the delayed allocation writeout path 650 * finally doing the actual allocation of delayed blocks */ 651 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 652 /* caller is from the direct IO path, request to creation of an 653 unwritten extents if not allocated, split the unwritten 654 extent if blocks has been preallocated already*/ 655 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 656 #define EXT4_GET_BLOCKS_CONVERT 0x0010 657 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 658 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 659 /* Convert extent to initialized after IO complete */ 660 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 661 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 662 /* Eventual metadata allocation (due to growing extent tree) 663 * should not fail, so try to use reserved blocks for that.*/ 664 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 665 /* Don't normalize allocation size (used for fallocate) */ 666 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 667 /* Convert written extents to unwritten */ 668 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 669 /* Write zeros to newly created written extents */ 670 #define EXT4_GET_BLOCKS_ZERO 0x0200 671 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 672 EXT4_GET_BLOCKS_ZERO) 673 /* Caller will submit data before dropping transaction handle. This 674 * allows jbd2 to avoid submitting data before commit. */ 675 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 676 /* Caller is in the atomic contex, find extent if it has been cached */ 677 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 678 679 /* 680 * The bit position of these flags must not overlap with any of the 681 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 682 * read_extent_tree_block(), ext4_split_extent_at(), 683 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 684 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 685 * caching the extents when reading from the extent tree while a 686 * truncate or punch hole operation is in progress. 687 */ 688 #define EXT4_EX_NOCACHE 0x40000000 689 #define EXT4_EX_FORCE_CACHE 0x20000000 690 #define EXT4_EX_NOFAIL 0x10000000 691 692 /* 693 * Flags used by ext4_free_blocks 694 */ 695 #define EXT4_FREE_BLOCKS_METADATA 0x0001 696 #define EXT4_FREE_BLOCKS_FORGET 0x0002 697 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 698 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 699 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 700 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 701 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 702 703 /* 704 * ioctl commands 705 */ 706 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 707 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 708 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 709 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 710 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 711 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 712 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 713 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 714 #define EXT4_IOC_MIGRATE _IO('f', 9) 715 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 716 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 717 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 718 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 719 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 720 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 721 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 722 /* ioctl codes 19--39 are reserved for fscrypt */ 723 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 724 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 725 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 726 #define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32) 727 #define EXT4_IOC_GETFSUUID _IOR('f', 44, struct fsuuid) 728 #define EXT4_IOC_SETFSUUID _IOW('f', 44, struct fsuuid) 729 730 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 731 732 /* 733 * Flags for going down operation 734 */ 735 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 736 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 737 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 738 739 /* 740 * Flags returned by EXT4_IOC_GETSTATE 741 * 742 * We only expose to userspace a subset of the state flags in 743 * i_state_flags 744 */ 745 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 746 #define EXT4_STATE_FLAG_NEW 0x00000002 747 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 748 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 749 750 /* flags for ioctl EXT4_IOC_CHECKPOINT */ 751 #define EXT4_IOC_CHECKPOINT_FLAG_DISCARD 0x1 752 #define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT 0x2 753 #define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN 0x4 754 #define EXT4_IOC_CHECKPOINT_FLAG_VALID (EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \ 755 EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \ 756 EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN) 757 758 /* 759 * Structure for EXT4_IOC_GETFSUUID/EXT4_IOC_SETFSUUID 760 */ 761 struct fsuuid { 762 __u32 fsu_len; 763 __u32 fsu_flags; 764 __u8 fsu_uuid[]; 765 }; 766 767 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 768 /* 769 * ioctl commands in 32 bit emulation 770 */ 771 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 772 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 773 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 774 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 775 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 776 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 777 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 778 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 779 #endif 780 781 /* 782 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 783 * It indicates that the entry in extent status cache is for a hole. 784 */ 785 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 786 787 /* Max physical block we can address w/o extents */ 788 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 789 790 /* Max logical block we can support */ 791 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 792 793 /* 794 * Structure of an inode on the disk 795 */ 796 struct ext4_inode { 797 __le16 i_mode; /* File mode */ 798 __le16 i_uid; /* Low 16 bits of Owner Uid */ 799 __le32 i_size_lo; /* Size in bytes */ 800 __le32 i_atime; /* Access time */ 801 __le32 i_ctime; /* Inode Change time */ 802 __le32 i_mtime; /* Modification time */ 803 __le32 i_dtime; /* Deletion Time */ 804 __le16 i_gid; /* Low 16 bits of Group Id */ 805 __le16 i_links_count; /* Links count */ 806 __le32 i_blocks_lo; /* Blocks count */ 807 __le32 i_flags; /* File flags */ 808 union { 809 struct { 810 __le32 l_i_version; 811 } linux1; 812 struct { 813 __u32 h_i_translator; 814 } hurd1; 815 struct { 816 __u32 m_i_reserved1; 817 } masix1; 818 } osd1; /* OS dependent 1 */ 819 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 820 __le32 i_generation; /* File version (for NFS) */ 821 __le32 i_file_acl_lo; /* File ACL */ 822 __le32 i_size_high; 823 __le32 i_obso_faddr; /* Obsoleted fragment address */ 824 union { 825 struct { 826 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 827 __le16 l_i_file_acl_high; 828 __le16 l_i_uid_high; /* these 2 fields */ 829 __le16 l_i_gid_high; /* were reserved2[0] */ 830 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 831 __le16 l_i_reserved; 832 } linux2; 833 struct { 834 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 835 __u16 h_i_mode_high; 836 __u16 h_i_uid_high; 837 __u16 h_i_gid_high; 838 __u32 h_i_author; 839 } hurd2; 840 struct { 841 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 842 __le16 m_i_file_acl_high; 843 __u32 m_i_reserved2[2]; 844 } masix2; 845 } osd2; /* OS dependent 2 */ 846 __le16 i_extra_isize; 847 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 848 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 849 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 850 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 851 __le32 i_crtime; /* File Creation time */ 852 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 853 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 854 __le32 i_projid; /* Project ID */ 855 }; 856 857 struct move_extent { 858 __u32 reserved; /* should be zero */ 859 __u32 donor_fd; /* donor file descriptor */ 860 __u64 orig_start; /* logical start offset in block for orig */ 861 __u64 donor_start; /* logical start offset in block for donor */ 862 __u64 len; /* block length to be moved */ 863 __u64 moved_len; /* moved block length */ 864 }; 865 866 #define EXT4_EPOCH_BITS 2 867 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 868 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 869 870 /* 871 * Extended fields will fit into an inode if the filesystem was formatted 872 * with large inodes (-I 256 or larger) and there are not currently any EAs 873 * consuming all of the available space. For new inodes we always reserve 874 * enough space for the kernel's known extended fields, but for inodes 875 * created with an old kernel this might not have been the case. None of 876 * the extended inode fields is critical for correct filesystem operation. 877 * This macro checks if a certain field fits in the inode. Note that 878 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 879 */ 880 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 881 ((offsetof(typeof(*ext4_inode), field) + \ 882 sizeof((ext4_inode)->field)) \ 883 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 884 (einode)->i_extra_isize)) \ 885 886 /* 887 * We use an encoding that preserves the times for extra epoch "00": 888 * 889 * extra msb of adjust for signed 890 * epoch 32-bit 32-bit tv_sec to 891 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 892 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 893 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 894 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 895 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 896 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 897 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 898 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 899 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 900 * 901 * Note that previous versions of the kernel on 64-bit systems would 902 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 903 * 1970. e2fsck will correct this, assuming that it is run on the 904 * affected filesystem before 2242. 905 */ 906 907 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 908 { 909 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 910 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 911 } 912 913 static inline void ext4_decode_extra_time(struct timespec64 *time, 914 __le32 extra) 915 { 916 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 917 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 918 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 919 } 920 921 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 922 do { \ 923 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 924 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 925 (raw_inode)->xtime ## _extra = \ 926 ext4_encode_extra_time(&(inode)->xtime); \ 927 } \ 928 else \ 929 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 930 } while (0) 931 932 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 933 do { \ 934 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 935 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 936 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 937 (raw_inode)->xtime ## _extra = \ 938 ext4_encode_extra_time(&(einode)->xtime); \ 939 } while (0) 940 941 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 942 do { \ 943 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 944 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 945 ext4_decode_extra_time(&(inode)->xtime, \ 946 raw_inode->xtime ## _extra); \ 947 } \ 948 else \ 949 (inode)->xtime.tv_nsec = 0; \ 950 } while (0) 951 952 953 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 954 do { \ 955 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 956 (einode)->xtime.tv_sec = \ 957 (signed)le32_to_cpu((raw_inode)->xtime); \ 958 else \ 959 (einode)->xtime.tv_sec = 0; \ 960 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 961 ext4_decode_extra_time(&(einode)->xtime, \ 962 raw_inode->xtime ## _extra); \ 963 else \ 964 (einode)->xtime.tv_nsec = 0; \ 965 } while (0) 966 967 #define i_disk_version osd1.linux1.l_i_version 968 969 #if defined(__KERNEL__) || defined(__linux__) 970 #define i_reserved1 osd1.linux1.l_i_reserved1 971 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 972 #define i_blocks_high osd2.linux2.l_i_blocks_high 973 #define i_uid_low i_uid 974 #define i_gid_low i_gid 975 #define i_uid_high osd2.linux2.l_i_uid_high 976 #define i_gid_high osd2.linux2.l_i_gid_high 977 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 978 979 #elif defined(__GNU__) 980 981 #define i_translator osd1.hurd1.h_i_translator 982 #define i_uid_high osd2.hurd2.h_i_uid_high 983 #define i_gid_high osd2.hurd2.h_i_gid_high 984 #define i_author osd2.hurd2.h_i_author 985 986 #elif defined(__masix__) 987 988 #define i_reserved1 osd1.masix1.m_i_reserved1 989 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 990 #define i_reserved2 osd2.masix2.m_i_reserved2 991 992 #endif /* defined(__KERNEL__) || defined(__linux__) */ 993 994 #include "extents_status.h" 995 #include "fast_commit.h" 996 997 /* 998 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 999 * 1000 * These are needed to avoid lockdep false positives when we need to 1001 * allocate blocks to the quota inode during ext4_map_blocks(), while 1002 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1003 * do quota tracking for the quota inode, this avoids deadlock (as 1004 * well as infinite recursion, since it isn't turtles all the way 1005 * down...) 1006 * 1007 * I_DATA_SEM_NORMAL - Used for most inodes 1008 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1009 * where the second inode has larger inode number 1010 * than the first 1011 * I_DATA_SEM_QUOTA - Used for quota inodes only 1012 */ 1013 enum { 1014 I_DATA_SEM_NORMAL = 0, 1015 I_DATA_SEM_OTHER, 1016 I_DATA_SEM_QUOTA, 1017 }; 1018 1019 1020 /* 1021 * fourth extended file system inode data in memory 1022 */ 1023 struct ext4_inode_info { 1024 __le32 i_data[15]; /* unconverted */ 1025 __u32 i_dtime; 1026 ext4_fsblk_t i_file_acl; 1027 1028 /* 1029 * i_block_group is the number of the block group which contains 1030 * this file's inode. Constant across the lifetime of the inode, 1031 * it is used for making block allocation decisions - we try to 1032 * place a file's data blocks near its inode block, and new inodes 1033 * near to their parent directory's inode. 1034 */ 1035 ext4_group_t i_block_group; 1036 ext4_lblk_t i_dir_start_lookup; 1037 #if (BITS_PER_LONG < 64) 1038 unsigned long i_state_flags; /* Dynamic state flags */ 1039 #endif 1040 unsigned long i_flags; 1041 1042 /* 1043 * Extended attributes can be read independently of the main file 1044 * data. Taking i_rwsem even when reading would cause contention 1045 * between readers of EAs and writers of regular file data, so 1046 * instead we synchronize on xattr_sem when reading or changing 1047 * EAs. 1048 */ 1049 struct rw_semaphore xattr_sem; 1050 1051 /* 1052 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1053 * i_orphan is used. 1054 */ 1055 union { 1056 struct list_head i_orphan; /* unlinked but open inodes */ 1057 unsigned int i_orphan_idx; /* Index in orphan file */ 1058 }; 1059 1060 /* Fast commit related info */ 1061 1062 /* For tracking dentry create updates */ 1063 struct list_head i_fc_dilist; 1064 struct list_head i_fc_list; /* 1065 * inodes that need fast commit 1066 * protected by sbi->s_fc_lock. 1067 */ 1068 1069 /* Start of lblk range that needs to be committed in this fast commit */ 1070 ext4_lblk_t i_fc_lblk_start; 1071 1072 /* End of lblk range that needs to be committed in this fast commit */ 1073 ext4_lblk_t i_fc_lblk_len; 1074 1075 /* Number of ongoing updates on this inode */ 1076 atomic_t i_fc_updates; 1077 1078 /* Fast commit wait queue for this inode */ 1079 wait_queue_head_t i_fc_wait; 1080 1081 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1082 struct mutex i_fc_lock; 1083 1084 /* 1085 * i_disksize keeps track of what the inode size is ON DISK, not 1086 * in memory. During truncate, i_size is set to the new size by 1087 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1088 * set i_disksize to 0 until the truncate is actually under way. 1089 * 1090 * The intent is that i_disksize always represents the blocks which 1091 * are used by this file. This allows recovery to restart truncate 1092 * on orphans if we crash during truncate. We actually write i_disksize 1093 * into the on-disk inode when writing inodes out, instead of i_size. 1094 * 1095 * The only time when i_disksize and i_size may be different is when 1096 * a truncate is in progress. The only things which change i_disksize 1097 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1098 */ 1099 loff_t i_disksize; 1100 1101 /* 1102 * i_data_sem is for serialising ext4_truncate() against 1103 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1104 * data tree are chopped off during truncate. We can't do that in 1105 * ext4 because whenever we perform intermediate commits during 1106 * truncate, the inode and all the metadata blocks *must* be in a 1107 * consistent state which allows truncation of the orphans to restart 1108 * during recovery. Hence we must fix the get_block-vs-truncate race 1109 * by other means, so we have i_data_sem. 1110 */ 1111 struct rw_semaphore i_data_sem; 1112 struct inode vfs_inode; 1113 struct jbd2_inode *jinode; 1114 1115 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1116 1117 /* 1118 * File creation time. Its function is same as that of 1119 * struct timespec64 i_{a,c,m}time in the generic inode. 1120 */ 1121 struct timespec64 i_crtime; 1122 1123 /* mballoc */ 1124 atomic_t i_prealloc_active; 1125 struct list_head i_prealloc_list; 1126 spinlock_t i_prealloc_lock; 1127 1128 /* extents status tree */ 1129 struct ext4_es_tree i_es_tree; 1130 rwlock_t i_es_lock; 1131 struct list_head i_es_list; 1132 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1133 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1134 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1135 extents to shrink. Protected by 1136 i_es_lock */ 1137 1138 /* ialloc */ 1139 ext4_group_t i_last_alloc_group; 1140 1141 /* allocation reservation info for delalloc */ 1142 /* In case of bigalloc, this refer to clusters rather than blocks */ 1143 unsigned int i_reserved_data_blocks; 1144 1145 /* pending cluster reservations for bigalloc file systems */ 1146 struct ext4_pending_tree i_pending_tree; 1147 1148 /* on-disk additional length */ 1149 __u16 i_extra_isize; 1150 1151 /* Indicate the inline data space. */ 1152 u16 i_inline_off; 1153 u16 i_inline_size; 1154 1155 #ifdef CONFIG_QUOTA 1156 /* quota space reservation, managed internally by quota code */ 1157 qsize_t i_reserved_quota; 1158 #endif 1159 1160 /* Lock protecting lists below */ 1161 spinlock_t i_completed_io_lock; 1162 /* 1163 * Completed IOs that need unwritten extents handling and have 1164 * transaction reserved 1165 */ 1166 struct list_head i_rsv_conversion_list; 1167 struct work_struct i_rsv_conversion_work; 1168 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1169 1170 spinlock_t i_block_reservation_lock; 1171 1172 /* 1173 * Transactions that contain inode's metadata needed to complete 1174 * fsync and fdatasync, respectively. 1175 */ 1176 tid_t i_sync_tid; 1177 tid_t i_datasync_tid; 1178 1179 #ifdef CONFIG_QUOTA 1180 struct dquot *i_dquot[MAXQUOTAS]; 1181 #endif 1182 1183 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1184 __u32 i_csum_seed; 1185 1186 kprojid_t i_projid; 1187 }; 1188 1189 /* 1190 * File system states 1191 */ 1192 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1193 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1194 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1195 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1196 1197 /* 1198 * Misc. filesystem flags 1199 */ 1200 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1201 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1202 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1203 1204 /* 1205 * Mount flags set via mount options or defaults 1206 */ 1207 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1208 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1209 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1210 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1211 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1212 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1213 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1214 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1215 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1216 #ifdef CONFIG_FS_DAX 1217 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1218 #else 1219 #define EXT4_MOUNT_DAX_ALWAYS 0 1220 #endif 1221 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1222 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1223 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1224 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1225 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1226 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1227 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1228 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1229 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1230 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1231 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1232 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1233 * enable enforcement for hidden 1234 * quota files */ 1235 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1236 * enforcement for hidden quota 1237 * files */ 1238 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1239 * enforcement */ 1240 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1241 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1242 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1243 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1244 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1245 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1246 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1247 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1248 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1249 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1250 1251 /* 1252 * Mount flags set either automatically (could not be set by mount option) 1253 * based on per file system feature or property or in special cases such as 1254 * distinguishing between explicit mount option definition and default. 1255 */ 1256 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1257 specified delalloc */ 1258 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1259 size of blocksize * 8 1260 blocks */ 1261 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1262 file systems */ 1263 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1264 specified journal checksum */ 1265 1266 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1267 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1268 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1269 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1270 * scanning in mballoc 1271 */ 1272 1273 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1274 ~EXT4_MOUNT_##opt 1275 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1276 EXT4_MOUNT_##opt 1277 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1278 EXT4_MOUNT_##opt) 1279 1280 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1281 ~EXT4_MOUNT2_##opt 1282 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1283 EXT4_MOUNT2_##opt 1284 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1285 EXT4_MOUNT2_##opt) 1286 1287 #define ext4_test_and_set_bit __test_and_set_bit_le 1288 #define ext4_set_bit __set_bit_le 1289 #define ext4_set_bit_atomic ext2_set_bit_atomic 1290 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1291 #define ext4_clear_bit __clear_bit_le 1292 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1293 #define ext4_test_bit test_bit_le 1294 #define ext4_find_next_zero_bit find_next_zero_bit_le 1295 #define ext4_find_next_bit find_next_bit_le 1296 1297 extern void mb_set_bits(void *bm, int cur, int len); 1298 1299 /* 1300 * Maximal mount counts between two filesystem checks 1301 */ 1302 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1303 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1304 1305 /* 1306 * Behaviour when detecting errors 1307 */ 1308 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1309 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1310 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1311 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1312 1313 /* Metadata checksum algorithm codes */ 1314 #define EXT4_CRC32C_CHKSUM 1 1315 1316 #define EXT4_LABEL_MAX 16 1317 1318 /* 1319 * Structure of the super block 1320 */ 1321 struct ext4_super_block { 1322 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1323 __le32 s_blocks_count_lo; /* Blocks count */ 1324 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1325 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1326 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1327 __le32 s_first_data_block; /* First Data Block */ 1328 __le32 s_log_block_size; /* Block size */ 1329 __le32 s_log_cluster_size; /* Allocation cluster size */ 1330 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1331 __le32 s_clusters_per_group; /* # Clusters per group */ 1332 __le32 s_inodes_per_group; /* # Inodes per group */ 1333 __le32 s_mtime; /* Mount time */ 1334 /*30*/ __le32 s_wtime; /* Write time */ 1335 __le16 s_mnt_count; /* Mount count */ 1336 __le16 s_max_mnt_count; /* Maximal mount count */ 1337 __le16 s_magic; /* Magic signature */ 1338 __le16 s_state; /* File system state */ 1339 __le16 s_errors; /* Behaviour when detecting errors */ 1340 __le16 s_minor_rev_level; /* minor revision level */ 1341 /*40*/ __le32 s_lastcheck; /* time of last check */ 1342 __le32 s_checkinterval; /* max. time between checks */ 1343 __le32 s_creator_os; /* OS */ 1344 __le32 s_rev_level; /* Revision level */ 1345 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1346 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1347 /* 1348 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1349 * 1350 * Note: the difference between the compatible feature set and 1351 * the incompatible feature set is that if there is a bit set 1352 * in the incompatible feature set that the kernel doesn't 1353 * know about, it should refuse to mount the filesystem. 1354 * 1355 * e2fsck's requirements are more strict; if it doesn't know 1356 * about a feature in either the compatible or incompatible 1357 * feature set, it must abort and not try to meddle with 1358 * things it doesn't understand... 1359 */ 1360 __le32 s_first_ino; /* First non-reserved inode */ 1361 __le16 s_inode_size; /* size of inode structure */ 1362 __le16 s_block_group_nr; /* block group # of this superblock */ 1363 __le32 s_feature_compat; /* compatible feature set */ 1364 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1365 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1366 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1367 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1368 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1369 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1370 /* 1371 * Performance hints. Directory preallocation should only 1372 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1373 */ 1374 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1375 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1376 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1377 /* 1378 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1379 */ 1380 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1381 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1382 __le32 s_journal_dev; /* device number of journal file */ 1383 __le32 s_last_orphan; /* start of list of inodes to delete */ 1384 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1385 __u8 s_def_hash_version; /* Default hash version to use */ 1386 __u8 s_jnl_backup_type; 1387 __le16 s_desc_size; /* size of group descriptor */ 1388 /*100*/ __le32 s_default_mount_opts; 1389 __le32 s_first_meta_bg; /* First metablock block group */ 1390 __le32 s_mkfs_time; /* When the filesystem was created */ 1391 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1392 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1393 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1394 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1395 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1396 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1397 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1398 __le32 s_flags; /* Miscellaneous flags */ 1399 __le16 s_raid_stride; /* RAID stride */ 1400 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1401 __le64 s_mmp_block; /* Block for multi-mount protection */ 1402 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1403 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1404 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1405 __u8 s_encryption_level; /* versioning level for encryption */ 1406 __u8 s_reserved_pad; /* Padding to next 32bits */ 1407 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1408 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1409 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1410 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1411 snapshot's future use */ 1412 __le32 s_snapshot_list; /* inode number of the head of the 1413 on-disk snapshot list */ 1414 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1415 __le32 s_error_count; /* number of fs errors */ 1416 __le32 s_first_error_time; /* first time an error happened */ 1417 __le32 s_first_error_ino; /* inode involved in first error */ 1418 __le64 s_first_error_block; /* block involved of first error */ 1419 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1420 __le32 s_first_error_line; /* line number where error happened */ 1421 __le32 s_last_error_time; /* most recent time of an error */ 1422 __le32 s_last_error_ino; /* inode involved in last error */ 1423 __le32 s_last_error_line; /* line number where error happened */ 1424 __le64 s_last_error_block; /* block involved of last error */ 1425 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1426 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1427 __u8 s_mount_opts[64]; 1428 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1429 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1430 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1431 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1432 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1433 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1434 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1435 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1436 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1437 __u8 s_wtime_hi; 1438 __u8 s_mtime_hi; 1439 __u8 s_mkfs_time_hi; 1440 __u8 s_lastcheck_hi; 1441 __u8 s_first_error_time_hi; 1442 __u8 s_last_error_time_hi; 1443 __u8 s_first_error_errcode; 1444 __u8 s_last_error_errcode; 1445 __le16 s_encoding; /* Filename charset encoding */ 1446 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1447 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1448 __le32 s_reserved[94]; /* Padding to the end of the block */ 1449 __le32 s_checksum; /* crc32c(superblock) */ 1450 }; 1451 1452 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1453 1454 #ifdef __KERNEL__ 1455 1456 /* Number of quota types we support */ 1457 #define EXT4_MAXQUOTAS 3 1458 1459 #define EXT4_ENC_UTF8_12_1 1 1460 1461 /* Types of ext4 journal triggers */ 1462 enum ext4_journal_trigger_type { 1463 EXT4_JTR_ORPHAN_FILE, 1464 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1465 }; 1466 1467 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1468 1469 struct ext4_journal_trigger { 1470 struct jbd2_buffer_trigger_type tr_triggers; 1471 struct super_block *sb; 1472 }; 1473 1474 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1475 struct jbd2_buffer_trigger_type *trigger) 1476 { 1477 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1478 } 1479 1480 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1481 1482 /* Structure at the tail of orphan block */ 1483 struct ext4_orphan_block_tail { 1484 __le32 ob_magic; 1485 __le32 ob_checksum; 1486 }; 1487 1488 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1489 { 1490 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1491 sizeof(u32); 1492 } 1493 1494 struct ext4_orphan_block { 1495 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1496 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1497 }; 1498 1499 /* 1500 * Info about orphan file. 1501 */ 1502 struct ext4_orphan_info { 1503 int of_blocks; /* Number of orphan blocks in a file */ 1504 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1505 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1506 * file blocks */ 1507 }; 1508 1509 /* 1510 * fourth extended-fs super-block data in memory 1511 */ 1512 struct ext4_sb_info { 1513 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1514 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1515 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1516 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1517 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1518 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1519 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1520 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1521 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1522 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1523 unsigned long s_overhead; /* # of fs overhead clusters */ 1524 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1525 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1526 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1527 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1528 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1529 struct buffer_head * __rcu *s_group_desc; 1530 unsigned int s_mount_opt; 1531 unsigned int s_mount_opt2; 1532 unsigned long s_mount_flags; 1533 unsigned int s_def_mount_opt; 1534 ext4_fsblk_t s_sb_block; 1535 atomic64_t s_resv_clusters; 1536 kuid_t s_resuid; 1537 kgid_t s_resgid; 1538 unsigned short s_mount_state; 1539 unsigned short s_pad; 1540 int s_addr_per_block_bits; 1541 int s_desc_per_block_bits; 1542 int s_inode_size; 1543 int s_first_ino; 1544 unsigned int s_inode_readahead_blks; 1545 unsigned int s_inode_goal; 1546 u32 s_hash_seed[4]; 1547 int s_def_hash_version; 1548 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1549 struct percpu_counter s_freeclusters_counter; 1550 struct percpu_counter s_freeinodes_counter; 1551 struct percpu_counter s_dirs_counter; 1552 struct percpu_counter s_dirtyclusters_counter; 1553 struct percpu_counter s_sra_exceeded_retry_limit; 1554 struct blockgroup_lock *s_blockgroup_lock; 1555 struct proc_dir_entry *s_proc; 1556 struct kobject s_kobj; 1557 struct completion s_kobj_unregister; 1558 struct super_block *s_sb; 1559 struct buffer_head *s_mmp_bh; 1560 1561 /* Journaling */ 1562 struct journal_s *s_journal; 1563 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1564 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1565 struct list_head s_orphan; /* List of orphaned inodes in on disk 1566 list */ 1567 struct ext4_orphan_info s_orphan_info; 1568 unsigned long s_commit_interval; 1569 u32 s_max_batch_time; 1570 u32 s_min_batch_time; 1571 struct block_device *s_journal_bdev; 1572 #ifdef CONFIG_QUOTA 1573 /* Names of quota files with journalled quota */ 1574 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1575 int s_jquota_fmt; /* Format of quota to use */ 1576 #endif 1577 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1578 struct ext4_system_blocks __rcu *s_system_blks; 1579 1580 #ifdef EXTENTS_STATS 1581 /* ext4 extents stats */ 1582 unsigned long s_ext_min; 1583 unsigned long s_ext_max; 1584 unsigned long s_depth_max; 1585 spinlock_t s_ext_stats_lock; 1586 unsigned long s_ext_blocks; 1587 unsigned long s_ext_extents; 1588 #endif 1589 1590 /* for buddy allocator */ 1591 struct ext4_group_info ** __rcu *s_group_info; 1592 struct inode *s_buddy_cache; 1593 spinlock_t s_md_lock; 1594 unsigned short *s_mb_offsets; 1595 unsigned int *s_mb_maxs; 1596 unsigned int s_group_info_size; 1597 unsigned int s_mb_free_pending; 1598 struct list_head s_freed_data_list; /* List of blocks to be freed 1599 after commit completed */ 1600 struct list_head s_discard_list; 1601 struct work_struct s_discard_work; 1602 atomic_t s_retry_alloc_pending; 1603 struct rb_root s_mb_avg_fragment_size_root; 1604 rwlock_t s_mb_rb_lock; 1605 struct list_head *s_mb_largest_free_orders; 1606 rwlock_t *s_mb_largest_free_orders_locks; 1607 1608 /* tunables */ 1609 unsigned long s_stripe; 1610 unsigned int s_mb_max_linear_groups; 1611 unsigned int s_mb_stream_request; 1612 unsigned int s_mb_max_to_scan; 1613 unsigned int s_mb_min_to_scan; 1614 unsigned int s_mb_stats; 1615 unsigned int s_mb_order2_reqs; 1616 unsigned int s_mb_group_prealloc; 1617 unsigned int s_mb_max_inode_prealloc; 1618 unsigned int s_max_dir_size_kb; 1619 /* where last allocation was done - for stream allocation */ 1620 unsigned long s_mb_last_group; 1621 unsigned long s_mb_last_start; 1622 unsigned int s_mb_prefetch; 1623 unsigned int s_mb_prefetch_limit; 1624 1625 /* stats for buddy allocator */ 1626 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1627 atomic_t s_bal_success; /* we found long enough chunks */ 1628 atomic_t s_bal_allocated; /* in blocks */ 1629 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1630 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1631 atomic_t s_bal_goals; /* goal hits */ 1632 atomic_t s_bal_breaks; /* too long searches */ 1633 atomic_t s_bal_2orders; /* 2^order hits */ 1634 atomic_t s_bal_cr0_bad_suggestions; 1635 atomic_t s_bal_cr1_bad_suggestions; 1636 atomic64_t s_bal_cX_groups_considered[4]; 1637 atomic64_t s_bal_cX_hits[4]; 1638 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */ 1639 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1640 atomic64_t s_mb_generation_time; 1641 atomic_t s_mb_lost_chunks; 1642 atomic_t s_mb_preallocated; 1643 atomic_t s_mb_discarded; 1644 atomic_t s_lock_busy; 1645 1646 /* locality groups */ 1647 struct ext4_locality_group __percpu *s_locality_groups; 1648 1649 /* for write statistics */ 1650 unsigned long s_sectors_written_start; 1651 u64 s_kbytes_written; 1652 1653 /* the size of zero-out chunk */ 1654 unsigned int s_extent_max_zeroout_kb; 1655 1656 unsigned int s_log_groups_per_flex; 1657 struct flex_groups * __rcu *s_flex_groups; 1658 ext4_group_t s_flex_groups_allocated; 1659 1660 /* workqueue for reserved extent conversions (buffered io) */ 1661 struct workqueue_struct *rsv_conversion_wq; 1662 1663 /* timer for periodic error stats printing */ 1664 struct timer_list s_err_report; 1665 1666 /* Lazy inode table initialization info */ 1667 struct ext4_li_request *s_li_request; 1668 /* Wait multiplier for lazy initialization thread */ 1669 unsigned int s_li_wait_mult; 1670 1671 /* Kernel thread for multiple mount protection */ 1672 struct task_struct *s_mmp_tsk; 1673 1674 /* record the last minlen when FITRIM is called. */ 1675 unsigned long s_last_trim_minblks; 1676 1677 /* Reference to checksum algorithm driver via cryptoapi */ 1678 struct crypto_shash *s_chksum_driver; 1679 1680 /* Precomputed FS UUID checksum for seeding other checksums */ 1681 __u32 s_csum_seed; 1682 1683 /* Reclaim extents from extent status tree */ 1684 struct shrinker s_es_shrinker; 1685 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1686 long s_es_nr_inode; 1687 struct ext4_es_stats s_es_stats; 1688 struct mb_cache *s_ea_block_cache; 1689 struct mb_cache *s_ea_inode_cache; 1690 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1691 1692 /* Journal triggers for checksum computation */ 1693 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1694 1695 /* Ratelimit ext4 messages. */ 1696 struct ratelimit_state s_err_ratelimit_state; 1697 struct ratelimit_state s_warning_ratelimit_state; 1698 struct ratelimit_state s_msg_ratelimit_state; 1699 atomic_t s_warning_count; 1700 atomic_t s_msg_count; 1701 1702 /* Encryption policy for '-o test_dummy_encryption' */ 1703 struct fscrypt_dummy_policy s_dummy_enc_policy; 1704 1705 /* 1706 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1707 * or EXTENTS flag. 1708 */ 1709 struct percpu_rw_semaphore s_writepages_rwsem; 1710 struct dax_device *s_daxdev; 1711 u64 s_dax_part_off; 1712 #ifdef CONFIG_EXT4_DEBUG 1713 unsigned long s_simulate_fail; 1714 #endif 1715 /* Record the errseq of the backing block device */ 1716 errseq_t s_bdev_wb_err; 1717 spinlock_t s_bdev_wb_lock; 1718 1719 /* Information about errors that happened during this mount */ 1720 spinlock_t s_error_lock; 1721 int s_add_error_count; 1722 int s_first_error_code; 1723 __u32 s_first_error_line; 1724 __u32 s_first_error_ino; 1725 __u64 s_first_error_block; 1726 const char *s_first_error_func; 1727 time64_t s_first_error_time; 1728 int s_last_error_code; 1729 __u32 s_last_error_line; 1730 __u32 s_last_error_ino; 1731 __u64 s_last_error_block; 1732 const char *s_last_error_func; 1733 time64_t s_last_error_time; 1734 /* 1735 * If we are in a context where we cannot update error information in 1736 * the on-disk superblock, we queue this work to do it. 1737 */ 1738 struct work_struct s_error_work; 1739 1740 /* Ext4 fast commit sub transaction ID */ 1741 atomic_t s_fc_subtid; 1742 1743 /* 1744 * After commit starts, the main queue gets locked, and the further 1745 * updates get added in the staging queue. 1746 */ 1747 #define FC_Q_MAIN 0 1748 #define FC_Q_STAGING 1 1749 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1750 * that have data changes in them. 1751 */ 1752 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1753 unsigned int s_fc_bytes; 1754 /* 1755 * Main fast commit lock. This lock protects accesses to the 1756 * following fields: 1757 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1758 */ 1759 spinlock_t s_fc_lock; 1760 struct buffer_head *s_fc_bh; 1761 struct ext4_fc_stats s_fc_stats; 1762 tid_t s_fc_ineligible_tid; 1763 #ifdef CONFIG_EXT4_DEBUG 1764 int s_fc_debug_max_replay; 1765 #endif 1766 struct ext4_fc_replay_state s_fc_replay_state; 1767 }; 1768 1769 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1770 { 1771 return sb->s_fs_info; 1772 } 1773 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1774 { 1775 return container_of(inode, struct ext4_inode_info, vfs_inode); 1776 } 1777 1778 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1779 { 1780 return ino == EXT4_ROOT_INO || 1781 (ino >= EXT4_FIRST_INO(sb) && 1782 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1783 } 1784 1785 /* 1786 * Returns: sbi->field[index] 1787 * Used to access an array element from the following sbi fields which require 1788 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1789 * - s_group_desc 1790 * - s_group_info 1791 * - s_flex_group 1792 */ 1793 #define sbi_array_rcu_deref(sbi, field, index) \ 1794 ({ \ 1795 typeof(*((sbi)->field)) _v; \ 1796 rcu_read_lock(); \ 1797 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1798 rcu_read_unlock(); \ 1799 _v; \ 1800 }) 1801 1802 /* 1803 * run-time mount flags 1804 */ 1805 enum { 1806 EXT4_MF_MNTDIR_SAMPLED, 1807 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1808 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1809 }; 1810 1811 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1812 { 1813 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1814 } 1815 1816 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1817 { 1818 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1819 } 1820 1821 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1822 { 1823 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1824 } 1825 1826 1827 /* 1828 * Simulate_fail codes 1829 */ 1830 #define EXT4_SIM_BBITMAP_EIO 1 1831 #define EXT4_SIM_BBITMAP_CRC 2 1832 #define EXT4_SIM_IBITMAP_EIO 3 1833 #define EXT4_SIM_IBITMAP_CRC 4 1834 #define EXT4_SIM_INODE_EIO 5 1835 #define EXT4_SIM_INODE_CRC 6 1836 #define EXT4_SIM_DIRBLOCK_EIO 7 1837 #define EXT4_SIM_DIRBLOCK_CRC 8 1838 1839 static inline bool ext4_simulate_fail(struct super_block *sb, 1840 unsigned long code) 1841 { 1842 #ifdef CONFIG_EXT4_DEBUG 1843 struct ext4_sb_info *sbi = EXT4_SB(sb); 1844 1845 if (unlikely(sbi->s_simulate_fail == code)) { 1846 sbi->s_simulate_fail = 0; 1847 return true; 1848 } 1849 #endif 1850 return false; 1851 } 1852 1853 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1854 struct buffer_head *bh, 1855 unsigned long code) 1856 { 1857 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1858 clear_buffer_uptodate(bh); 1859 } 1860 1861 /* 1862 * Error number codes for s_{first,last}_error_errno 1863 * 1864 * Linux errno numbers are architecture specific, so we need to translate 1865 * them into something which is architecture independent. We don't define 1866 * codes for all errno's; just the ones which are most likely to be the cause 1867 * of an ext4_error() call. 1868 */ 1869 #define EXT4_ERR_UNKNOWN 1 1870 #define EXT4_ERR_EIO 2 1871 #define EXT4_ERR_ENOMEM 3 1872 #define EXT4_ERR_EFSBADCRC 4 1873 #define EXT4_ERR_EFSCORRUPTED 5 1874 #define EXT4_ERR_ENOSPC 6 1875 #define EXT4_ERR_ENOKEY 7 1876 #define EXT4_ERR_EROFS 8 1877 #define EXT4_ERR_EFBIG 9 1878 #define EXT4_ERR_EEXIST 10 1879 #define EXT4_ERR_ERANGE 11 1880 #define EXT4_ERR_EOVERFLOW 12 1881 #define EXT4_ERR_EBUSY 13 1882 #define EXT4_ERR_ENOTDIR 14 1883 #define EXT4_ERR_ENOTEMPTY 15 1884 #define EXT4_ERR_ESHUTDOWN 16 1885 #define EXT4_ERR_EFAULT 17 1886 1887 /* 1888 * Inode dynamic state flags 1889 */ 1890 enum { 1891 EXT4_STATE_JDATA, /* journaled data exists */ 1892 EXT4_STATE_NEW, /* inode is newly created */ 1893 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1894 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1895 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1896 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1897 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1898 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1899 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1900 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1901 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1902 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1903 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1904 }; 1905 1906 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1907 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1908 { \ 1909 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1910 } \ 1911 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1912 { \ 1913 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1914 } \ 1915 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1916 { \ 1917 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1918 } 1919 1920 /* Add these declarations here only so that these functions can be 1921 * found by name. Otherwise, they are very hard to locate. */ 1922 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1923 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1924 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1925 EXT4_INODE_BIT_FNS(flag, flags, 0) 1926 1927 /* Add these declarations here only so that these functions can be 1928 * found by name. Otherwise, they are very hard to locate. */ 1929 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1930 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1931 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1932 #if (BITS_PER_LONG < 64) 1933 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1934 1935 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1936 { 1937 (ei)->i_state_flags = 0; 1938 } 1939 #else 1940 EXT4_INODE_BIT_FNS(state, flags, 32) 1941 1942 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1943 { 1944 /* We depend on the fact that callers will set i_flags */ 1945 } 1946 #endif 1947 #else 1948 /* Assume that user mode programs are passing in an ext4fs superblock, not 1949 * a kernel struct super_block. This will allow us to call the feature-test 1950 * macros from user land. */ 1951 #define EXT4_SB(sb) (sb) 1952 #endif 1953 1954 static inline bool ext4_verity_in_progress(struct inode *inode) 1955 { 1956 return IS_ENABLED(CONFIG_FS_VERITY) && 1957 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1958 } 1959 1960 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1961 1962 /* 1963 * Codes for operating systems 1964 */ 1965 #define EXT4_OS_LINUX 0 1966 #define EXT4_OS_HURD 1 1967 #define EXT4_OS_MASIX 2 1968 #define EXT4_OS_FREEBSD 3 1969 #define EXT4_OS_LITES 4 1970 1971 /* 1972 * Revision levels 1973 */ 1974 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1975 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1976 1977 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1978 1979 #define EXT4_GOOD_OLD_INODE_SIZE 128 1980 1981 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1982 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1983 #define EXT4_TIMESTAMP_MIN S32_MIN 1984 1985 /* 1986 * Feature set definitions 1987 */ 1988 1989 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1990 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1991 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1992 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1993 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1994 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1995 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1996 /* 1997 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1998 * incompatible only if fast commit blocks are present in the FS. Since we 1999 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2000 * incompatible. We also have a JBD2 incompat feature, which gets set when 2001 * there are fast commit blocks present in the journal. 2002 */ 2003 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2004 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2005 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2006 2007 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2008 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2009 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2010 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2011 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2012 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2013 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2014 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2015 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2016 /* 2017 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2018 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2019 * all other data structures' checksums. However, the METADATA_CSUM and 2020 * GDT_CSUM bits are mutually exclusive. 2021 */ 2022 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2023 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2024 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2025 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2026 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2027 non-empty */ 2028 2029 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2030 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2031 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2032 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2033 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2034 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2035 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2036 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2037 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2038 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2039 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2040 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2041 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2042 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2043 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2044 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2045 2046 extern void ext4_update_dynamic_rev(struct super_block *sb); 2047 2048 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2049 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2050 { \ 2051 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2052 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2053 } \ 2054 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2055 { \ 2056 ext4_update_dynamic_rev(sb); \ 2057 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2058 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2059 } \ 2060 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2061 { \ 2062 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2063 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2064 } 2065 2066 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2067 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2068 { \ 2069 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2070 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2071 } \ 2072 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2073 { \ 2074 ext4_update_dynamic_rev(sb); \ 2075 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2076 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2077 } \ 2078 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2079 { \ 2080 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2081 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2082 } 2083 2084 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2085 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2086 { \ 2087 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2088 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2089 } \ 2090 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2091 { \ 2092 ext4_update_dynamic_rev(sb); \ 2093 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2094 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2095 } \ 2096 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2097 { \ 2098 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2099 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2100 } 2101 2102 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2103 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2104 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2105 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2106 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2107 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2108 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2109 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2110 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2111 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2112 2113 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2114 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2115 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2116 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2117 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2118 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2119 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2120 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2121 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2122 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2123 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2124 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2125 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2126 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2127 2128 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2129 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2130 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2131 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2132 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2133 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2134 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2135 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2136 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2137 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2138 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2139 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2140 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2141 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2142 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2143 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2144 2145 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2146 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2147 EXT4_FEATURE_INCOMPAT_META_BG) 2148 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2149 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2150 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2151 2152 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2153 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2154 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2155 EXT4_FEATURE_INCOMPAT_META_BG) 2156 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2157 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2158 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2159 2160 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2161 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2162 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2163 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2164 EXT4_FEATURE_INCOMPAT_META_BG| \ 2165 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2166 EXT4_FEATURE_INCOMPAT_64BIT| \ 2167 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2168 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2169 EXT4_FEATURE_INCOMPAT_MMP | \ 2170 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2171 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2172 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2173 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2174 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2175 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2176 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2177 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2178 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2179 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2180 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2181 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2182 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2183 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2184 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2185 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2186 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2187 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2188 2189 #define EXTN_FEATURE_FUNCS(ver) \ 2190 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2191 { \ 2192 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2193 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2194 } \ 2195 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2196 { \ 2197 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2198 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2199 } \ 2200 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2201 { \ 2202 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2203 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2204 } 2205 2206 EXTN_FEATURE_FUNCS(2) 2207 EXTN_FEATURE_FUNCS(3) 2208 EXTN_FEATURE_FUNCS(4) 2209 2210 static inline bool ext4_has_compat_features(struct super_block *sb) 2211 { 2212 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2213 } 2214 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2215 { 2216 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2217 } 2218 static inline bool ext4_has_incompat_features(struct super_block *sb) 2219 { 2220 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2221 } 2222 2223 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2224 2225 /* 2226 * Superblock flags 2227 */ 2228 #define EXT4_FLAGS_RESIZING 0 2229 #define EXT4_FLAGS_SHUTDOWN 1 2230 #define EXT4_FLAGS_BDEV_IS_DAX 2 2231 2232 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2233 { 2234 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2235 } 2236 2237 /* 2238 * Default values for user and/or group using reserved blocks 2239 */ 2240 #define EXT4_DEF_RESUID 0 2241 #define EXT4_DEF_RESGID 0 2242 2243 /* 2244 * Default project ID 2245 */ 2246 #define EXT4_DEF_PROJID 0 2247 2248 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2249 2250 /* 2251 * Default mount options 2252 */ 2253 #define EXT4_DEFM_DEBUG 0x0001 2254 #define EXT4_DEFM_BSDGROUPS 0x0002 2255 #define EXT4_DEFM_XATTR_USER 0x0004 2256 #define EXT4_DEFM_ACL 0x0008 2257 #define EXT4_DEFM_UID16 0x0010 2258 #define EXT4_DEFM_JMODE 0x0060 2259 #define EXT4_DEFM_JMODE_DATA 0x0020 2260 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2261 #define EXT4_DEFM_JMODE_WBACK 0x0060 2262 #define EXT4_DEFM_NOBARRIER 0x0100 2263 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2264 #define EXT4_DEFM_DISCARD 0x0400 2265 #define EXT4_DEFM_NODELALLOC 0x0800 2266 2267 /* 2268 * Default journal batch times 2269 */ 2270 #define EXT4_DEF_MIN_BATCH_TIME 0 2271 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2272 2273 /* 2274 * Minimum number of groups in a flexgroup before we separate out 2275 * directories into the first block group of a flexgroup 2276 */ 2277 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2278 2279 /* 2280 * Structure of a directory entry 2281 */ 2282 #define EXT4_NAME_LEN 255 2283 /* 2284 * Base length of the ext4 directory entry excluding the name length 2285 */ 2286 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2287 2288 struct ext4_dir_entry { 2289 __le32 inode; /* Inode number */ 2290 __le16 rec_len; /* Directory entry length */ 2291 __le16 name_len; /* Name length */ 2292 char name[EXT4_NAME_LEN]; /* File name */ 2293 }; 2294 2295 2296 /* 2297 * Encrypted Casefolded entries require saving the hash on disk. This structure 2298 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2299 * boundary. 2300 */ 2301 struct ext4_dir_entry_hash { 2302 __le32 hash; 2303 __le32 minor_hash; 2304 }; 2305 2306 /* 2307 * The new version of the directory entry. Since EXT4 structures are 2308 * stored in intel byte order, and the name_len field could never be 2309 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2310 * file_type field. 2311 */ 2312 struct ext4_dir_entry_2 { 2313 __le32 inode; /* Inode number */ 2314 __le16 rec_len; /* Directory entry length */ 2315 __u8 name_len; /* Name length */ 2316 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2317 char name[EXT4_NAME_LEN]; /* File name */ 2318 }; 2319 2320 /* 2321 * Access the hashes at the end of ext4_dir_entry_2 2322 */ 2323 #define EXT4_DIRENT_HASHES(entry) \ 2324 ((struct ext4_dir_entry_hash *) \ 2325 (((void *)(entry)) + \ 2326 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2327 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2328 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2329 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2330 2331 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2332 { 2333 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2334 } 2335 2336 /* 2337 * This is a bogus directory entry at the end of each leaf block that 2338 * records checksums. 2339 */ 2340 struct ext4_dir_entry_tail { 2341 __le32 det_reserved_zero1; /* Pretend to be unused */ 2342 __le16 det_rec_len; /* 12 */ 2343 __u8 det_reserved_zero2; /* Zero name length */ 2344 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2345 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2346 }; 2347 2348 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2349 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2350 ((blocksize) - \ 2351 sizeof(struct ext4_dir_entry_tail)))) 2352 2353 /* 2354 * Ext4 directory file types. Only the low 3 bits are used. The 2355 * other bits are reserved for now. 2356 */ 2357 #define EXT4_FT_UNKNOWN 0 2358 #define EXT4_FT_REG_FILE 1 2359 #define EXT4_FT_DIR 2 2360 #define EXT4_FT_CHRDEV 3 2361 #define EXT4_FT_BLKDEV 4 2362 #define EXT4_FT_FIFO 5 2363 #define EXT4_FT_SOCK 6 2364 #define EXT4_FT_SYMLINK 7 2365 2366 #define EXT4_FT_MAX 8 2367 2368 #define EXT4_FT_DIR_CSUM 0xDE 2369 2370 /* 2371 * EXT4_DIR_PAD defines the directory entries boundaries 2372 * 2373 * NOTE: It must be a multiple of 4 2374 */ 2375 #define EXT4_DIR_PAD 4 2376 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2377 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2378 2379 /* 2380 * The rec_len is dependent on the type of directory. Directories that are 2381 * casefolded and encrypted need to store the hash as well, so we add room for 2382 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2383 * pass NULL for dir, as those entries do not use the extra fields. 2384 */ 2385 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2386 const struct inode *dir) 2387 { 2388 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2389 2390 if (dir && ext4_hash_in_dirent(dir)) 2391 rec_len += sizeof(struct ext4_dir_entry_hash); 2392 return (rec_len & ~EXT4_DIR_ROUND); 2393 } 2394 2395 /* 2396 * If we ever get support for fs block sizes > page_size, we'll need 2397 * to remove the #if statements in the next two functions... 2398 */ 2399 static inline unsigned int 2400 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2401 { 2402 unsigned len = le16_to_cpu(dlen); 2403 2404 #if (PAGE_SIZE >= 65536) 2405 if (len == EXT4_MAX_REC_LEN || len == 0) 2406 return blocksize; 2407 return (len & 65532) | ((len & 3) << 16); 2408 #else 2409 return len; 2410 #endif 2411 } 2412 2413 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2414 { 2415 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2416 #if (PAGE_SIZE >= 65536) 2417 if (len < 65536) 2418 return cpu_to_le16(len); 2419 if (len == blocksize) { 2420 if (blocksize == 65536) 2421 return cpu_to_le16(EXT4_MAX_REC_LEN); 2422 else 2423 return cpu_to_le16(0); 2424 } 2425 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2426 #else 2427 return cpu_to_le16(len); 2428 #endif 2429 } 2430 2431 /* 2432 * Hash Tree Directory indexing 2433 * (c) Daniel Phillips, 2001 2434 */ 2435 2436 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2437 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2438 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2439 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2440 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2441 2442 /* Legal values for the dx_root hash_version field: */ 2443 2444 #define DX_HASH_LEGACY 0 2445 #define DX_HASH_HALF_MD4 1 2446 #define DX_HASH_TEA 2 2447 #define DX_HASH_LEGACY_UNSIGNED 3 2448 #define DX_HASH_HALF_MD4_UNSIGNED 4 2449 #define DX_HASH_TEA_UNSIGNED 5 2450 #define DX_HASH_SIPHASH 6 2451 2452 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2453 const void *address, unsigned int length) 2454 { 2455 struct { 2456 struct shash_desc shash; 2457 char ctx[4]; 2458 } desc; 2459 2460 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2461 2462 desc.shash.tfm = sbi->s_chksum_driver; 2463 *(u32 *)desc.ctx = crc; 2464 2465 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2466 2467 return *(u32 *)desc.ctx; 2468 } 2469 2470 #ifdef __KERNEL__ 2471 2472 /* hash info structure used by the directory hash */ 2473 struct dx_hash_info 2474 { 2475 u32 hash; 2476 u32 minor_hash; 2477 int hash_version; 2478 u32 *seed; 2479 }; 2480 2481 2482 /* 32 and 64 bit signed EOF for dx directories */ 2483 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2484 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2485 2486 2487 /* 2488 * Control parameters used by ext4_htree_next_block 2489 */ 2490 #define HASH_NB_ALWAYS 1 2491 2492 struct ext4_filename { 2493 const struct qstr *usr_fname; 2494 struct fscrypt_str disk_name; 2495 struct dx_hash_info hinfo; 2496 #ifdef CONFIG_FS_ENCRYPTION 2497 struct fscrypt_str crypto_buf; 2498 #endif 2499 #if IS_ENABLED(CONFIG_UNICODE) 2500 struct fscrypt_str cf_name; 2501 #endif 2502 }; 2503 2504 #define fname_name(p) ((p)->disk_name.name) 2505 #define fname_usr_name(p) ((p)->usr_fname->name) 2506 #define fname_len(p) ((p)->disk_name.len) 2507 2508 /* 2509 * Describe an inode's exact location on disk and in memory 2510 */ 2511 struct ext4_iloc 2512 { 2513 struct buffer_head *bh; 2514 unsigned long offset; 2515 ext4_group_t block_group; 2516 }; 2517 2518 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2519 { 2520 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2521 } 2522 2523 static inline bool ext4_is_quota_file(struct inode *inode) 2524 { 2525 return IS_NOQUOTA(inode) && 2526 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2527 } 2528 2529 /* 2530 * This structure is stuffed into the struct file's private_data field 2531 * for directories. It is where we put information so that we can do 2532 * readdir operations in hash tree order. 2533 */ 2534 struct dir_private_info { 2535 struct rb_root root; 2536 struct rb_node *curr_node; 2537 struct fname *extra_fname; 2538 loff_t last_pos; 2539 __u32 curr_hash; 2540 __u32 curr_minor_hash; 2541 __u32 next_hash; 2542 }; 2543 2544 /* calculate the first block number of the group */ 2545 static inline ext4_fsblk_t 2546 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2547 { 2548 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2549 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2550 } 2551 2552 /* 2553 * Special error return code only used by dx_probe() and its callers. 2554 */ 2555 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2556 2557 /* htree levels for ext4 */ 2558 #define EXT4_HTREE_LEVEL_COMPAT 2 2559 #define EXT4_HTREE_LEVEL 3 2560 2561 static inline int ext4_dir_htree_level(struct super_block *sb) 2562 { 2563 return ext4_has_feature_largedir(sb) ? 2564 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2565 } 2566 2567 /* 2568 * Timeout and state flag for lazy initialization inode thread. 2569 */ 2570 #define EXT4_DEF_LI_WAIT_MULT 10 2571 #define EXT4_DEF_LI_MAX_START_DELAY 5 2572 #define EXT4_LAZYINIT_QUIT 0x0001 2573 #define EXT4_LAZYINIT_RUNNING 0x0002 2574 2575 /* 2576 * Lazy inode table initialization info 2577 */ 2578 struct ext4_lazy_init { 2579 unsigned long li_state; 2580 struct list_head li_request_list; 2581 struct mutex li_list_mtx; 2582 }; 2583 2584 enum ext4_li_mode { 2585 EXT4_LI_MODE_PREFETCH_BBITMAP, 2586 EXT4_LI_MODE_ITABLE, 2587 }; 2588 2589 struct ext4_li_request { 2590 struct super_block *lr_super; 2591 enum ext4_li_mode lr_mode; 2592 ext4_group_t lr_first_not_zeroed; 2593 ext4_group_t lr_next_group; 2594 struct list_head lr_request; 2595 unsigned long lr_next_sched; 2596 unsigned long lr_timeout; 2597 }; 2598 2599 struct ext4_features { 2600 struct kobject f_kobj; 2601 struct completion f_kobj_unregister; 2602 }; 2603 2604 /* 2605 * This structure will be used for multiple mount protection. It will be 2606 * written into the block number saved in the s_mmp_block field in the 2607 * superblock. Programs that check MMP should assume that if 2608 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2609 * to use the filesystem, regardless of how old the timestamp is. 2610 */ 2611 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2612 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2613 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2614 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2615 2616 struct mmp_struct { 2617 __le32 mmp_magic; /* Magic number for MMP */ 2618 __le32 mmp_seq; /* Sequence no. updated periodically */ 2619 2620 /* 2621 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2622 * purposes and do not affect the correctness of the algorithm 2623 */ 2624 __le64 mmp_time; /* Time last updated */ 2625 char mmp_nodename[64]; /* Node which last updated MMP block */ 2626 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2627 2628 /* 2629 * mmp_check_interval is used to verify if the MMP block has been 2630 * updated on the block device. The value is updated based on the 2631 * maximum time to write the MMP block during an update cycle. 2632 */ 2633 __le16 mmp_check_interval; 2634 2635 __le16 mmp_pad1; 2636 __le32 mmp_pad2[226]; 2637 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2638 }; 2639 2640 /* arguments passed to the mmp thread */ 2641 struct mmpd_data { 2642 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2643 struct super_block *sb; /* super block of the fs */ 2644 }; 2645 2646 /* 2647 * Check interval multiplier 2648 * The MMP block is written every update interval and initially checked every 2649 * update interval x the multiplier (the value is then adapted based on the 2650 * write latency). The reason is that writes can be delayed under load and we 2651 * don't want readers to incorrectly assume that the filesystem is no longer 2652 * in use. 2653 */ 2654 #define EXT4_MMP_CHECK_MULT 2UL 2655 2656 /* 2657 * Minimum interval for MMP checking in seconds. 2658 */ 2659 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2660 2661 /* 2662 * Maximum interval for MMP checking in seconds. 2663 */ 2664 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2665 2666 /* 2667 * Function prototypes 2668 */ 2669 2670 /* 2671 * Ok, these declarations are also in <linux/kernel.h> but none of the 2672 * ext4 source programs needs to include it so they are duplicated here. 2673 */ 2674 # define NORET_TYPE /**/ 2675 # define ATTRIB_NORET __attribute__((noreturn)) 2676 # define NORET_AND noreturn, 2677 2678 /* bitmap.c */ 2679 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2680 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2681 struct ext4_group_desc *gdp, 2682 struct buffer_head *bh, int sz); 2683 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2684 struct ext4_group_desc *gdp, 2685 struct buffer_head *bh, int sz); 2686 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2687 struct ext4_group_desc *gdp, 2688 struct buffer_head *bh); 2689 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2690 struct ext4_group_desc *gdp, 2691 struct buffer_head *bh); 2692 2693 /* balloc.c */ 2694 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2695 ext4_fsblk_t blocknr, 2696 ext4_group_t *blockgrpp, 2697 ext4_grpblk_t *offsetp); 2698 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2699 ext4_fsblk_t block); 2700 2701 extern unsigned int ext4_block_group(struct super_block *sb, 2702 ext4_fsblk_t blocknr); 2703 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2704 ext4_fsblk_t blocknr); 2705 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2706 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2707 ext4_group_t group); 2708 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2709 ext4_fsblk_t goal, 2710 unsigned int flags, 2711 unsigned long *count, 2712 int *errp); 2713 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2714 s64 nclusters, unsigned int flags); 2715 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2716 extern void ext4_check_blocks_bitmap(struct super_block *); 2717 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2718 ext4_group_t block_group, 2719 struct buffer_head ** bh); 2720 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2721 2722 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2723 ext4_group_t block_group, 2724 bool ignore_locked); 2725 extern int ext4_wait_block_bitmap(struct super_block *sb, 2726 ext4_group_t block_group, 2727 struct buffer_head *bh); 2728 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2729 ext4_group_t block_group); 2730 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2731 ext4_group_t block_group, 2732 struct ext4_group_desc *gdp); 2733 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2734 2735 #if IS_ENABLED(CONFIG_UNICODE) 2736 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2737 const struct qstr *iname, 2738 struct ext4_filename *fname); 2739 #endif 2740 2741 /* ext4 encryption related stuff goes here crypto.c */ 2742 #ifdef CONFIG_FS_ENCRYPTION 2743 extern const struct fscrypt_operations ext4_cryptops; 2744 2745 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2746 int lookup, struct ext4_filename *fname); 2747 2748 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2749 struct ext4_filename *fname); 2750 2751 void ext4_fname_free_filename(struct ext4_filename *fname); 2752 2753 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2754 2755 #else /* !CONFIG_FS_ENCRYPTION */ 2756 static inline int ext4_fname_setup_filename(struct inode *dir, 2757 const struct qstr *iname, 2758 int lookup, 2759 struct ext4_filename *fname) 2760 { 2761 int err = 0; 2762 fname->usr_fname = iname; 2763 fname->disk_name.name = (unsigned char *) iname->name; 2764 fname->disk_name.len = iname->len; 2765 2766 #if IS_ENABLED(CONFIG_UNICODE) 2767 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2768 #endif 2769 2770 return err; 2771 } 2772 2773 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2774 struct dentry *dentry, 2775 struct ext4_filename *fname) 2776 { 2777 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2778 } 2779 2780 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2781 { 2782 #if IS_ENABLED(CONFIG_UNICODE) 2783 kfree(fname->cf_name.name); 2784 fname->cf_name.name = NULL; 2785 #endif 2786 } 2787 2788 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2789 void __user *arg) 2790 { 2791 return -EOPNOTSUPP; 2792 } 2793 #endif /* !CONFIG_FS_ENCRYPTION */ 2794 2795 /* dir.c */ 2796 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2797 struct file *, 2798 struct ext4_dir_entry_2 *, 2799 struct buffer_head *, char *, int, 2800 unsigned int); 2801 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2802 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2803 (de), (bh), (buf), (size), (offset))) 2804 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2805 __u32 minor_hash, 2806 struct ext4_dir_entry_2 *dirent, 2807 struct fscrypt_str *ent_name); 2808 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2809 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2810 struct buffer_head *bh, 2811 void *buf, int buf_size, 2812 struct ext4_filename *fname, 2813 struct ext4_dir_entry_2 **dest_de); 2814 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2815 struct ext4_dir_entry_2 *de, 2816 int buf_size, 2817 struct ext4_filename *fname); 2818 static inline void ext4_update_dx_flag(struct inode *inode) 2819 { 2820 if (!ext4_has_feature_dir_index(inode->i_sb) && 2821 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2822 /* ext4_iget() should have caught this... */ 2823 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2824 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2825 } 2826 } 2827 static const unsigned char ext4_filetype_table[] = { 2828 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2829 }; 2830 2831 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2832 { 2833 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2834 return DT_UNKNOWN; 2835 2836 return ext4_filetype_table[filetype]; 2837 } 2838 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2839 void *buf, int buf_size); 2840 2841 /* fsync.c */ 2842 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2843 2844 /* hash.c */ 2845 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2846 struct dx_hash_info *hinfo); 2847 2848 /* ialloc.c */ 2849 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2850 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *, 2851 struct inode *, umode_t, 2852 const struct qstr *qstr, __u32 goal, 2853 uid_t *owner, __u32 i_flags, 2854 int handle_type, unsigned int line_no, 2855 int nblocks); 2856 2857 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2858 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \ 2859 (goal), (owner), i_flags, 0, 0, 0) 2860 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \ 2861 type, nblocks) \ 2862 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2863 0, (type), __LINE__, (nblocks)) 2864 2865 2866 extern void ext4_free_inode(handle_t *, struct inode *); 2867 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2868 extern unsigned long ext4_count_free_inodes(struct super_block *); 2869 extern unsigned long ext4_count_dirs(struct super_block *); 2870 extern void ext4_check_inodes_bitmap(struct super_block *); 2871 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2872 extern int ext4_init_inode_table(struct super_block *sb, 2873 ext4_group_t group, int barrier); 2874 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2875 2876 /* fast_commit.c */ 2877 int ext4_fc_info_show(struct seq_file *seq, void *v); 2878 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2879 void ext4_fc_init_inode(struct inode *inode); 2880 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2881 ext4_lblk_t end); 2882 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2883 struct dentry *dentry); 2884 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2885 struct dentry *dentry); 2886 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2887 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2888 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2889 struct dentry *dentry); 2890 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2891 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2892 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2893 void ext4_fc_start_update(struct inode *inode); 2894 void ext4_fc_stop_update(struct inode *inode); 2895 void ext4_fc_del(struct inode *inode); 2896 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2897 void ext4_fc_replay_cleanup(struct super_block *sb); 2898 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2899 int __init ext4_fc_init_dentry_cache(void); 2900 void ext4_fc_destroy_dentry_cache(void); 2901 int ext4_fc_record_regions(struct super_block *sb, int ino, 2902 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2903 int len, int replay); 2904 2905 /* mballoc.c */ 2906 extern const struct seq_operations ext4_mb_seq_groups_ops; 2907 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2908 extern long ext4_mb_stats; 2909 extern long ext4_mb_max_to_scan; 2910 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2911 extern int ext4_mb_init(struct super_block *); 2912 extern int ext4_mb_release(struct super_block *); 2913 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2914 struct ext4_allocation_request *, int *); 2915 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2916 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2917 extern int __init ext4_init_mballoc(void); 2918 extern void ext4_exit_mballoc(void); 2919 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2920 ext4_group_t group, 2921 unsigned int nr, int *cnt); 2922 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2923 unsigned int nr); 2924 2925 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2926 struct buffer_head *bh, ext4_fsblk_t block, 2927 unsigned long count, int flags); 2928 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2929 ext4_group_t ngroups); 2930 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2931 ext4_group_t i, struct ext4_group_desc *desc); 2932 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2933 ext4_fsblk_t block, unsigned long count); 2934 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2935 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2936 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2937 int len, int state); 2938 2939 /* inode.c */ 2940 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2941 struct ext4_inode_info *ei); 2942 int ext4_inode_is_fast_symlink(struct inode *inode); 2943 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2944 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2945 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2946 bool wait, struct buffer_head **bhs); 2947 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2948 struct buffer_head *bh_result, int create); 2949 int ext4_get_block(struct inode *inode, sector_t iblock, 2950 struct buffer_head *bh_result, int create); 2951 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2952 struct buffer_head *bh, int create); 2953 int ext4_walk_page_buffers(handle_t *handle, 2954 struct inode *inode, 2955 struct buffer_head *head, 2956 unsigned from, 2957 unsigned to, 2958 int *partial, 2959 int (*fn)(handle_t *handle, struct inode *inode, 2960 struct buffer_head *bh)); 2961 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2962 struct buffer_head *bh); 2963 #define FALL_BACK_TO_NONDELALLOC 1 2964 #define CONVERT_INLINE_DATA 2 2965 2966 typedef enum { 2967 EXT4_IGET_NORMAL = 0, 2968 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2969 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2970 } ext4_iget_flags; 2971 2972 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2973 ext4_iget_flags flags, const char *function, 2974 unsigned int line); 2975 2976 #define ext4_iget(sb, ino, flags) \ 2977 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2978 2979 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2980 extern int ext4_setattr(struct user_namespace *, struct dentry *, 2981 struct iattr *); 2982 extern int ext4_getattr(struct user_namespace *, const struct path *, 2983 struct kstat *, u32, unsigned int); 2984 extern void ext4_evict_inode(struct inode *); 2985 extern void ext4_clear_inode(struct inode *); 2986 extern int ext4_file_getattr(struct user_namespace *, const struct path *, 2987 struct kstat *, u32, unsigned int); 2988 extern int ext4_sync_inode(handle_t *, struct inode *); 2989 extern void ext4_dirty_inode(struct inode *, int); 2990 extern int ext4_change_inode_journal_flag(struct inode *, int); 2991 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2992 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2993 struct ext4_iloc *iloc); 2994 extern int ext4_inode_attach_jinode(struct inode *inode); 2995 extern int ext4_can_truncate(struct inode *inode); 2996 extern int ext4_truncate(struct inode *); 2997 extern int ext4_break_layouts(struct inode *); 2998 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2999 extern void ext4_set_inode_flags(struct inode *, bool init); 3000 extern int ext4_alloc_da_blocks(struct inode *inode); 3001 extern void ext4_set_aops(struct inode *inode); 3002 extern int ext4_writepage_trans_blocks(struct inode *); 3003 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3004 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3005 loff_t lstart, loff_t lend); 3006 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3007 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3008 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3009 extern void ext4_da_release_space(struct inode *inode, int to_free); 3010 extern void ext4_da_update_reserve_space(struct inode *inode, 3011 int used, int quota_claim); 3012 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3013 ext4_fsblk_t pblk, ext4_lblk_t len); 3014 3015 /* indirect.c */ 3016 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3017 struct ext4_map_blocks *map, int flags); 3018 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3019 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3020 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3021 ext4_lblk_t start, ext4_lblk_t end); 3022 3023 /* ioctl.c */ 3024 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3025 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3026 int ext4_fileattr_set(struct user_namespace *mnt_userns, 3027 struct dentry *dentry, struct fileattr *fa); 3028 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3029 extern void ext4_reset_inode_seed(struct inode *inode); 3030 int ext4_update_overhead(struct super_block *sb, bool force); 3031 3032 /* migrate.c */ 3033 extern int ext4_ext_migrate(struct inode *); 3034 extern int ext4_ind_migrate(struct inode *inode); 3035 3036 /* namei.c */ 3037 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3038 struct inode *inode); 3039 extern int ext4_dirblock_csum_verify(struct inode *inode, 3040 struct buffer_head *bh); 3041 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3042 __u32 start_minor_hash, __u32 *next_hash); 3043 extern int ext4_search_dir(struct buffer_head *bh, 3044 char *search_buf, 3045 int buf_size, 3046 struct inode *dir, 3047 struct ext4_filename *fname, 3048 unsigned int offset, 3049 struct ext4_dir_entry_2 **res_dir); 3050 extern int ext4_generic_delete_entry(struct inode *dir, 3051 struct ext4_dir_entry_2 *de_del, 3052 struct buffer_head *bh, 3053 void *entry_buf, 3054 int buf_size, 3055 int csum_size); 3056 extern bool ext4_empty_dir(struct inode *inode); 3057 3058 /* resize.c */ 3059 extern void ext4_kvfree_array_rcu(void *to_free); 3060 extern int ext4_group_add(struct super_block *sb, 3061 struct ext4_new_group_data *input); 3062 extern int ext4_group_extend(struct super_block *sb, 3063 struct ext4_super_block *es, 3064 ext4_fsblk_t n_blocks_count); 3065 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3066 extern unsigned int ext4_list_backups(struct super_block *sb, 3067 unsigned int *three, unsigned int *five, 3068 unsigned int *seven); 3069 3070 /* super.c */ 3071 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3072 sector_t block, blk_opf_t op_flags); 3073 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3074 sector_t block); 3075 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3076 bh_end_io_t *end_io); 3077 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3078 bh_end_io_t *end_io); 3079 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3080 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3081 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3082 extern int ext4_calculate_overhead(struct super_block *sb); 3083 extern __le32 ext4_superblock_csum(struct super_block *sb, 3084 struct ext4_super_block *es); 3085 extern void ext4_superblock_csum_set(struct super_block *sb); 3086 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3087 ext4_group_t ngroup); 3088 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3089 char nbuf[16]); 3090 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3091 ext4_group_t block_group, 3092 unsigned int flags); 3093 3094 extern __printf(7, 8) 3095 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3096 int, __u64, const char *, ...); 3097 extern __printf(6, 7) 3098 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3099 ext4_fsblk_t, int, const char *, ...); 3100 extern __printf(5, 6) 3101 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3102 const char *, ...); 3103 extern void __ext4_std_error(struct super_block *, const char *, 3104 unsigned int, int); 3105 extern __printf(4, 5) 3106 void __ext4_warning(struct super_block *, const char *, unsigned int, 3107 const char *, ...); 3108 extern __printf(4, 5) 3109 void __ext4_warning_inode(const struct inode *inode, const char *function, 3110 unsigned int line, const char *fmt, ...); 3111 extern __printf(3, 4) 3112 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3113 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3114 const char *, unsigned int, const char *); 3115 extern __printf(7, 8) 3116 void __ext4_grp_locked_error(const char *, unsigned int, 3117 struct super_block *, ext4_group_t, 3118 unsigned long, ext4_fsblk_t, 3119 const char *, ...); 3120 3121 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3122 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3123 3124 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3125 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3126 3127 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3128 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3129 (fmt), ## a) 3130 3131 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3132 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3133 3134 #define ext4_abort(sb, err, fmt, a...) \ 3135 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3136 3137 #ifdef CONFIG_PRINTK 3138 3139 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3140 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3141 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3142 __ext4_error_inode((inode), (func), (line), (block), \ 3143 (err), (fmt), ##__VA_ARGS__) 3144 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3145 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3146 #define ext4_error(sb, fmt, ...) \ 3147 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3148 ##__VA_ARGS__) 3149 #define ext4_error_err(sb, err, fmt, ...) \ 3150 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3151 ##__VA_ARGS__) 3152 #define ext4_warning(sb, fmt, ...) \ 3153 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3154 #define ext4_warning_inode(inode, fmt, ...) \ 3155 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3156 #define ext4_msg(sb, level, fmt, ...) \ 3157 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3158 #define dump_mmp_msg(sb, mmp, msg) \ 3159 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3160 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3161 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3162 fmt, ##__VA_ARGS__) 3163 3164 #else 3165 3166 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3167 do { \ 3168 no_printk(fmt, ##__VA_ARGS__); \ 3169 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3170 } while (0) 3171 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3172 do { \ 3173 no_printk(fmt, ##__VA_ARGS__); \ 3174 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3175 } while (0) 3176 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3177 do { \ 3178 no_printk(fmt, ##__VA_ARGS__); \ 3179 __ext4_error_file(file, "", 0, block, " "); \ 3180 } while (0) 3181 #define ext4_error(sb, fmt, ...) \ 3182 do { \ 3183 no_printk(fmt, ##__VA_ARGS__); \ 3184 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3185 } while (0) 3186 #define ext4_error_err(sb, err, fmt, ...) \ 3187 do { \ 3188 no_printk(fmt, ##__VA_ARGS__); \ 3189 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3190 } while (0) 3191 #define ext4_warning(sb, fmt, ...) \ 3192 do { \ 3193 no_printk(fmt, ##__VA_ARGS__); \ 3194 __ext4_warning(sb, "", 0, " "); \ 3195 } while (0) 3196 #define ext4_warning_inode(inode, fmt, ...) \ 3197 do { \ 3198 no_printk(fmt, ##__VA_ARGS__); \ 3199 __ext4_warning_inode(inode, "", 0, " "); \ 3200 } while (0) 3201 #define ext4_msg(sb, level, fmt, ...) \ 3202 do { \ 3203 no_printk(fmt, ##__VA_ARGS__); \ 3204 __ext4_msg(sb, "", " "); \ 3205 } while (0) 3206 #define dump_mmp_msg(sb, mmp, msg) \ 3207 __dump_mmp_msg(sb, mmp, "", 0, "") 3208 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3209 do { \ 3210 no_printk(fmt, ##__VA_ARGS__); \ 3211 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3212 } while (0) 3213 3214 #endif 3215 3216 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3217 struct ext4_group_desc *bg); 3218 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3219 struct ext4_group_desc *bg); 3220 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3221 struct ext4_group_desc *bg); 3222 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3223 struct ext4_group_desc *bg); 3224 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3225 struct ext4_group_desc *bg); 3226 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3227 struct ext4_group_desc *bg); 3228 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3229 struct ext4_group_desc *bg); 3230 extern void ext4_block_bitmap_set(struct super_block *sb, 3231 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3232 extern void ext4_inode_bitmap_set(struct super_block *sb, 3233 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3234 extern void ext4_inode_table_set(struct super_block *sb, 3235 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3236 extern void ext4_free_group_clusters_set(struct super_block *sb, 3237 struct ext4_group_desc *bg, 3238 __u32 count); 3239 extern void ext4_free_inodes_set(struct super_block *sb, 3240 struct ext4_group_desc *bg, __u32 count); 3241 extern void ext4_used_dirs_set(struct super_block *sb, 3242 struct ext4_group_desc *bg, __u32 count); 3243 extern void ext4_itable_unused_set(struct super_block *sb, 3244 struct ext4_group_desc *bg, __u32 count); 3245 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3246 struct ext4_group_desc *gdp); 3247 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3248 struct ext4_group_desc *gdp); 3249 extern int ext4_register_li_request(struct super_block *sb, 3250 ext4_group_t first_not_zeroed); 3251 3252 static inline int ext4_has_metadata_csum(struct super_block *sb) 3253 { 3254 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3255 !EXT4_SB(sb)->s_chksum_driver); 3256 3257 return ext4_has_feature_metadata_csum(sb) && 3258 (EXT4_SB(sb)->s_chksum_driver != NULL); 3259 } 3260 3261 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3262 { 3263 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3264 } 3265 3266 #define ext4_read_incompat_64bit_val(es, name) \ 3267 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3268 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3269 le32_to_cpu(es->name##_lo)) 3270 3271 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3272 { 3273 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3274 } 3275 3276 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3277 { 3278 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3279 } 3280 3281 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3282 { 3283 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3284 } 3285 3286 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3287 ext4_fsblk_t blk) 3288 { 3289 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3290 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3291 } 3292 3293 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3294 ext4_fsblk_t blk) 3295 { 3296 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3297 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3298 } 3299 3300 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3301 ext4_fsblk_t blk) 3302 { 3303 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3304 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3305 } 3306 3307 static inline loff_t ext4_isize(struct super_block *sb, 3308 struct ext4_inode *raw_inode) 3309 { 3310 if (ext4_has_feature_largedir(sb) || 3311 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3312 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3313 le32_to_cpu(raw_inode->i_size_lo); 3314 3315 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3316 } 3317 3318 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3319 { 3320 raw_inode->i_size_lo = cpu_to_le32(i_size); 3321 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3322 } 3323 3324 static inline 3325 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 3326 ext4_group_t group) 3327 { 3328 struct ext4_group_info **grp_info; 3329 long indexv, indexh; 3330 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 3331 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 3332 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 3333 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv); 3334 return grp_info[indexh]; 3335 } 3336 3337 /* 3338 * Reading s_groups_count requires using smp_rmb() afterwards. See 3339 * the locking protocol documented in the comments of ext4_group_add() 3340 * in resize.c 3341 */ 3342 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3343 { 3344 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3345 3346 smp_rmb(); 3347 return ngroups; 3348 } 3349 3350 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3351 ext4_group_t block_group) 3352 { 3353 return block_group >> sbi->s_log_groups_per_flex; 3354 } 3355 3356 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3357 { 3358 return 1 << sbi->s_log_groups_per_flex; 3359 } 3360 3361 #define ext4_std_error(sb, errno) \ 3362 do { \ 3363 if ((errno)) \ 3364 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3365 } while (0) 3366 3367 #ifdef CONFIG_SMP 3368 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3369 * counters. So we need to make sure we have free clusters more 3370 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3371 */ 3372 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3373 #else 3374 #define EXT4_FREECLUSTERS_WATERMARK 0 3375 #endif 3376 3377 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3378 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3379 { 3380 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3381 !inode_is_locked(inode)); 3382 down_write(&EXT4_I(inode)->i_data_sem); 3383 if (newsize > EXT4_I(inode)->i_disksize) 3384 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3385 up_write(&EXT4_I(inode)->i_data_sem); 3386 } 3387 3388 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3389 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3390 { 3391 int changed = 0; 3392 3393 if (newsize > inode->i_size) { 3394 i_size_write(inode, newsize); 3395 changed = 1; 3396 } 3397 if (newsize > EXT4_I(inode)->i_disksize) { 3398 ext4_update_i_disksize(inode, newsize); 3399 changed |= 2; 3400 } 3401 return changed; 3402 } 3403 3404 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3405 loff_t len); 3406 3407 struct ext4_group_info { 3408 unsigned long bb_state; 3409 #ifdef AGGRESSIVE_CHECK 3410 unsigned long bb_check_counter; 3411 #endif 3412 struct rb_root bb_free_root; 3413 ext4_grpblk_t bb_first_free; /* first free block */ 3414 ext4_grpblk_t bb_free; /* total free blocks */ 3415 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3416 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3417 ext4_group_t bb_group; /* Group number */ 3418 struct list_head bb_prealloc_list; 3419 #ifdef DOUBLE_CHECK 3420 void *bb_bitmap; 3421 #endif 3422 struct rw_semaphore alloc_sem; 3423 struct rb_node bb_avg_fragment_size_rb; 3424 struct list_head bb_largest_free_order_node; 3425 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3426 * regions, index is order. 3427 * bb_counters[3] = 5 means 3428 * 5 free 8-block regions. */ 3429 }; 3430 3431 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3432 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3433 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3434 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3435 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3436 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3437 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3438 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3439 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3440 3441 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3442 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3443 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3444 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3445 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3446 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3447 3448 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3449 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3450 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3451 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3452 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3453 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3454 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3455 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3456 3457 #define EXT4_MAX_CONTENTION 8 3458 #define EXT4_CONTENTION_THRESHOLD 2 3459 3460 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3461 ext4_group_t group) 3462 { 3463 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3464 } 3465 3466 /* 3467 * Returns true if the filesystem is busy enough that attempts to 3468 * access the block group locks has run into contention. 3469 */ 3470 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3471 { 3472 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3473 } 3474 3475 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3476 { 3477 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3478 if (spin_trylock(lock)) 3479 /* 3480 * We're able to grab the lock right away, so drop the 3481 * lock contention counter. 3482 */ 3483 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3484 else { 3485 /* 3486 * The lock is busy, so bump the contention counter, 3487 * and then wait on the spin lock. 3488 */ 3489 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3490 EXT4_MAX_CONTENTION); 3491 spin_lock(lock); 3492 } 3493 } 3494 3495 static inline void ext4_unlock_group(struct super_block *sb, 3496 ext4_group_t group) 3497 { 3498 spin_unlock(ext4_group_lock_ptr(sb, group)); 3499 } 3500 3501 #ifdef CONFIG_QUOTA 3502 static inline bool ext4_quota_capable(struct super_block *sb) 3503 { 3504 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3505 } 3506 3507 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3508 { 3509 struct ext4_sb_info *sbi = EXT4_SB(sb); 3510 3511 return (ext4_has_feature_quota(sb) || 3512 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3513 } 3514 int ext4_enable_quotas(struct super_block *sb); 3515 #endif 3516 3517 /* 3518 * Block validity checking 3519 */ 3520 #define ext4_check_indirect_blockref(inode, bh) \ 3521 ext4_check_blockref(__func__, __LINE__, inode, \ 3522 (__le32 *)(bh)->b_data, \ 3523 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3524 3525 #define ext4_ind_check_inode(inode) \ 3526 ext4_check_blockref(__func__, __LINE__, inode, \ 3527 EXT4_I(inode)->i_data, \ 3528 EXT4_NDIR_BLOCKS) 3529 3530 /* 3531 * Inodes and files operations 3532 */ 3533 3534 /* dir.c */ 3535 extern const struct file_operations ext4_dir_operations; 3536 3537 /* file.c */ 3538 extern const struct inode_operations ext4_file_inode_operations; 3539 extern const struct file_operations ext4_file_operations; 3540 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3541 3542 /* inline.c */ 3543 extern int ext4_get_max_inline_size(struct inode *inode); 3544 extern int ext4_find_inline_data_nolock(struct inode *inode); 3545 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3546 unsigned int len); 3547 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3548 3549 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3550 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3551 struct inode *inode, 3552 loff_t pos, unsigned len, 3553 struct page **pagep); 3554 extern int ext4_write_inline_data_end(struct inode *inode, 3555 loff_t pos, unsigned len, 3556 unsigned copied, 3557 struct page *page); 3558 extern struct buffer_head * 3559 ext4_journalled_write_inline_data(struct inode *inode, 3560 unsigned len, 3561 struct page *page); 3562 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3563 struct inode *inode, 3564 loff_t pos, unsigned len, 3565 struct page **pagep, 3566 void **fsdata); 3567 extern int ext4_try_add_inline_entry(handle_t *handle, 3568 struct ext4_filename *fname, 3569 struct inode *dir, struct inode *inode); 3570 extern int ext4_try_create_inline_dir(handle_t *handle, 3571 struct inode *parent, 3572 struct inode *inode); 3573 extern int ext4_read_inline_dir(struct file *filp, 3574 struct dir_context *ctx, 3575 int *has_inline_data); 3576 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3577 struct inode *dir, ext4_lblk_t block, 3578 struct dx_hash_info *hinfo, 3579 __u32 start_hash, __u32 start_minor_hash, 3580 int *has_inline_data); 3581 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3582 struct ext4_filename *fname, 3583 struct ext4_dir_entry_2 **res_dir, 3584 int *has_inline_data); 3585 extern int ext4_delete_inline_entry(handle_t *handle, 3586 struct inode *dir, 3587 struct ext4_dir_entry_2 *de_del, 3588 struct buffer_head *bh, 3589 int *has_inline_data); 3590 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3591 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3592 struct ext4_dir_entry_2 **parent_de, 3593 int *retval); 3594 extern int ext4_inline_data_fiemap(struct inode *inode, 3595 struct fiemap_extent_info *fieinfo, 3596 int *has_inline, __u64 start, __u64 len); 3597 extern void *ext4_read_inline_link(struct inode *inode); 3598 3599 struct iomap; 3600 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3601 3602 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3603 3604 extern int ext4_convert_inline_data(struct inode *inode); 3605 3606 static inline int ext4_has_inline_data(struct inode *inode) 3607 { 3608 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3609 EXT4_I(inode)->i_inline_off; 3610 } 3611 3612 /* namei.c */ 3613 extern const struct inode_operations ext4_dir_inode_operations; 3614 extern const struct inode_operations ext4_special_inode_operations; 3615 extern struct dentry *ext4_get_parent(struct dentry *child); 3616 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3617 struct ext4_dir_entry_2 *de, 3618 int blocksize, int csum_size, 3619 unsigned int parent_ino, int dotdot_real_len); 3620 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3621 unsigned int blocksize); 3622 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3623 struct buffer_head *bh); 3624 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name, 3625 struct inode *inode); 3626 extern int __ext4_link(struct inode *dir, struct inode *inode, 3627 struct dentry *dentry); 3628 3629 #define S_SHIFT 12 3630 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3631 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3632 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3633 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3634 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3635 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3636 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3637 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3638 }; 3639 3640 static inline void ext4_set_de_type(struct super_block *sb, 3641 struct ext4_dir_entry_2 *de, 3642 umode_t mode) { 3643 if (ext4_has_feature_filetype(sb)) 3644 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3645 } 3646 3647 /* readpages.c */ 3648 extern int ext4_mpage_readpages(struct inode *inode, 3649 struct readahead_control *rac, struct page *page); 3650 extern int __init ext4_init_post_read_processing(void); 3651 extern void ext4_exit_post_read_processing(void); 3652 3653 /* symlink.c */ 3654 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3655 extern const struct inode_operations ext4_symlink_inode_operations; 3656 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3657 3658 /* sysfs.c */ 3659 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3660 extern int ext4_register_sysfs(struct super_block *sb); 3661 extern void ext4_unregister_sysfs(struct super_block *sb); 3662 extern int __init ext4_init_sysfs(void); 3663 extern void ext4_exit_sysfs(void); 3664 3665 /* block_validity */ 3666 extern void ext4_release_system_zone(struct super_block *sb); 3667 extern int ext4_setup_system_zone(struct super_block *sb); 3668 extern int __init ext4_init_system_zone(void); 3669 extern void ext4_exit_system_zone(void); 3670 extern int ext4_inode_block_valid(struct inode *inode, 3671 ext4_fsblk_t start_blk, 3672 unsigned int count); 3673 extern int ext4_check_blockref(const char *, unsigned int, 3674 struct inode *, __le32 *, unsigned int); 3675 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3676 ext4_fsblk_t start_blk, unsigned int count); 3677 3678 3679 /* extents.c */ 3680 struct ext4_ext_path; 3681 struct ext4_extent; 3682 3683 /* 3684 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3685 * __le32. 3686 */ 3687 #define EXT_MAX_BLOCKS 0xffffffff 3688 3689 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3690 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3691 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3692 struct ext4_map_blocks *map, int flags); 3693 extern int ext4_ext_truncate(handle_t *, struct inode *); 3694 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3695 ext4_lblk_t end); 3696 extern void ext4_ext_init(struct super_block *); 3697 extern void ext4_ext_release(struct super_block *); 3698 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3699 loff_t len); 3700 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3701 loff_t offset, ssize_t len); 3702 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3703 ext4_io_end_t *io_end); 3704 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3705 struct ext4_map_blocks *map, int flags); 3706 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3707 int num, 3708 struct ext4_ext_path *path); 3709 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3710 struct ext4_ext_path **, 3711 struct ext4_extent *, int); 3712 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3713 struct ext4_ext_path **, 3714 int flags); 3715 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3716 extern int ext4_ext_check_inode(struct inode *inode); 3717 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3718 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3719 __u64 start, __u64 len); 3720 extern int ext4_get_es_cache(struct inode *inode, 3721 struct fiemap_extent_info *fieinfo, 3722 __u64 start, __u64 len); 3723 extern int ext4_ext_precache(struct inode *inode); 3724 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3725 struct inode *inode2, ext4_lblk_t lblk1, 3726 ext4_lblk_t lblk2, ext4_lblk_t count, 3727 int mark_unwritten,int *err); 3728 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3729 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3730 int check_cred, int restart_cred, 3731 int revoke_cred); 3732 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3733 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3734 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3735 int len, int unwritten, ext4_fsblk_t pblk); 3736 extern int ext4_ext_clear_bb(struct inode *inode); 3737 3738 3739 /* move_extent.c */ 3740 extern void ext4_double_down_write_data_sem(struct inode *first, 3741 struct inode *second); 3742 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3743 struct inode *donor_inode); 3744 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3745 __u64 start_orig, __u64 start_donor, 3746 __u64 len, __u64 *moved_len); 3747 3748 /* page-io.c */ 3749 extern int __init ext4_init_pageio(void); 3750 extern void ext4_exit_pageio(void); 3751 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3752 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3753 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3754 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3755 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3756 struct writeback_control *wbc); 3757 extern void ext4_end_io_rsv_work(struct work_struct *work); 3758 extern void ext4_io_submit(struct ext4_io_submit *io); 3759 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3760 struct page *page, 3761 int len, 3762 bool keep_towrite); 3763 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3764 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3765 3766 /* mmp.c */ 3767 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3768 3769 /* mmp.c */ 3770 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3771 3772 /* verity.c */ 3773 extern const struct fsverity_operations ext4_verityops; 3774 3775 /* orphan.c */ 3776 extern int ext4_orphan_add(handle_t *, struct inode *); 3777 extern int ext4_orphan_del(handle_t *, struct inode *); 3778 extern void ext4_orphan_cleanup(struct super_block *sb, 3779 struct ext4_super_block *es); 3780 extern void ext4_release_orphan_info(struct super_block *sb); 3781 extern int ext4_init_orphan_info(struct super_block *sb); 3782 extern int ext4_orphan_file_empty(struct super_block *sb); 3783 extern void ext4_orphan_file_block_trigger( 3784 struct jbd2_buffer_trigger_type *triggers, 3785 struct buffer_head *bh, 3786 void *data, size_t size); 3787 3788 /* 3789 * Add new method to test whether block and inode bitmaps are properly 3790 * initialized. With uninit_bg reading the block from disk is not enough 3791 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3792 */ 3793 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3794 3795 static inline int bitmap_uptodate(struct buffer_head *bh) 3796 { 3797 return (buffer_uptodate(bh) && 3798 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3799 } 3800 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3801 { 3802 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3803 } 3804 3805 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3806 3807 /* For ioend & aio unwritten conversion wait queues */ 3808 #define EXT4_WQ_HASH_SZ 37 3809 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3810 EXT4_WQ_HASH_SZ]) 3811 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3812 3813 extern int ext4_resize_begin(struct super_block *sb); 3814 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3815 3816 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3817 struct ext4_io_end *io_end) 3818 { 3819 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3820 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3821 atomic_inc(&EXT4_I(inode)->i_unwritten); 3822 } 3823 } 3824 3825 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3826 { 3827 struct inode *inode = io_end->inode; 3828 3829 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3830 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3831 /* Wake up anyone waiting on unwritten extent conversion */ 3832 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3833 wake_up_all(ext4_ioend_wq(inode)); 3834 } 3835 } 3836 3837 extern const struct iomap_ops ext4_iomap_ops; 3838 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3839 extern const struct iomap_ops ext4_iomap_report_ops; 3840 3841 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3842 { 3843 /* 3844 * If the buffer has the write error flag, we have failed 3845 * to write out data in the block. In this case, we don't 3846 * have to read the block because we may read the old data 3847 * successfully. 3848 */ 3849 if (buffer_write_io_error(bh)) 3850 set_buffer_uptodate(bh); 3851 return buffer_uptodate(bh); 3852 } 3853 3854 #endif /* __KERNEL__ */ 3855 3856 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3857 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3858 3859 #endif /* _EXT4_H */ 3860