1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/refcount.h> 21 #include <linux/types.h> 22 #include <linux/blkdev.h> 23 #include <linux/magic.h> 24 #include <linux/jbd2.h> 25 #include <linux/quota.h> 26 #include <linux/rwsem.h> 27 #include <linux/rbtree.h> 28 #include <linux/seqlock.h> 29 #include <linux/mutex.h> 30 #include <linux/timer.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #include <linux/fiemap.h> 40 #ifdef __KERNEL__ 41 #include <linux/compat.h> 42 #endif 43 #include <uapi/linux/ext4.h> 44 45 #include <linux/fscrypt.h> 46 #include <linux/fsverity.h> 47 48 #include <linux/compiler.h> 49 50 /* 51 * The fourth extended filesystem constants/structures 52 */ 53 54 /* 55 * with AGGRESSIVE_CHECK allocator runs consistency checks over 56 * structures. these checks slow things down a lot 57 */ 58 #define AGGRESSIVE_CHECK__ 59 60 /* 61 * with DOUBLE_CHECK defined mballoc creates persistent in-core 62 * bitmaps, maintains and uses them to check for double allocations 63 */ 64 #define DOUBLE_CHECK__ 65 66 /* 67 * Define EXT4FS_DEBUG to produce debug messages 68 */ 69 #undef EXT4FS_DEBUG 70 71 /* 72 * Debug code 73 */ 74 #ifdef EXT4FS_DEBUG 75 #define ext4_debug(f, a...) \ 76 do { \ 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 78 __FILE__, __LINE__, __func__); \ 79 printk(KERN_DEBUG f, ## a); \ 80 } while (0) 81 #else 82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 83 #endif 84 85 /* 86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 87 */ 88 #define EXT_DEBUG__ 89 90 /* 91 * Dynamic printk for controlled extents debugging. 92 */ 93 #ifdef CONFIG_EXT4_DEBUG 94 #define ext_debug(ino, fmt, ...) \ 95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 96 current->comm, task_pid_nr(current), \ 97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 98 __func__, ##__VA_ARGS__) 99 #else 100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 101 #endif 102 103 #define ASSERT(assert) \ 104 do { \ 105 if (unlikely(!(assert))) { \ 106 printk(KERN_EMERG \ 107 "Assertion failure in %s() at %s:%d: '%s'\n", \ 108 __func__, __FILE__, __LINE__, #assert); \ 109 BUG(); \ 110 } \ 111 } while (0) 112 113 /* data type for block offset of block group */ 114 typedef int ext4_grpblk_t; 115 116 /* data type for filesystem-wide blocks number */ 117 typedef unsigned long long ext4_fsblk_t; 118 119 /* data type for file logical block number */ 120 typedef __u32 ext4_lblk_t; 121 122 /* data type for block group number */ 123 typedef unsigned int ext4_group_t; 124 125 enum SHIFT_DIRECTION { 126 SHIFT_LEFT = 0, 127 SHIFT_RIGHT, 128 }; 129 130 /* 131 * Number of criterias defined. For each criteria, mballoc has slightly 132 * different way of finding the required blocks nad usually, higher the 133 * criteria the slower the allocation. We start at lower criterias and keep 134 * falling back to higher ones if we are not able to find any blocks. 135 */ 136 #define EXT4_MB_NUM_CRS 5 137 /* 138 * All possible allocation criterias for mballoc. Lower are faster. 139 */ 140 enum criteria { 141 /* 142 * Used when number of blocks needed is a power of 2. This doesn't 143 * trigger any disk IO except prefetch and is the fastest criteria. 144 */ 145 CR_POWER2_ALIGNED, 146 147 /* 148 * Tries to lookup in-memory data structures to find the most suitable 149 * group that satisfies goal request. No disk IO except block prefetch. 150 */ 151 CR_GOAL_LEN_FAST, 152 153 /* 154 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal length to 155 * the best available length for faster allocation. 156 */ 157 CR_BEST_AVAIL_LEN, 158 159 /* 160 * Reads each block group sequentially, performing disk IO if necessary, to 161 * find find_suitable block group. Tries to allocate goal length but might trim 162 * the request if nothing is found after enough tries. 163 */ 164 CR_GOAL_LEN_SLOW, 165 166 /* 167 * Finds the first free set of blocks and allocates those. This is only 168 * used in rare cases when CR_GOAL_LEN_SLOW also fails to allocate 169 * anything. 170 */ 171 CR_ANY_FREE, 172 }; 173 174 /* criteria below which we use fast block scanning and avoid unnecessary IO */ 175 #define CR_FAST CR_GOAL_LEN_SLOW 176 177 /* 178 * Flags used in mballoc's allocation_context flags field. 179 * 180 * Also used to show what's going on for debugging purposes when the 181 * flag field is exported via the traceport interface 182 */ 183 184 /* prefer goal again. length */ 185 #define EXT4_MB_HINT_MERGE 0x0001 186 /* blocks already reserved */ 187 #define EXT4_MB_HINT_RESERVED 0x0002 188 /* metadata is being allocated */ 189 #define EXT4_MB_HINT_METADATA 0x0004 190 /* first blocks in the file */ 191 #define EXT4_MB_HINT_FIRST 0x0008 192 /* search for the best chunk */ 193 #define EXT4_MB_HINT_BEST 0x0010 194 /* data is being allocated */ 195 #define EXT4_MB_HINT_DATA 0x0020 196 /* don't preallocate (for tails) */ 197 #define EXT4_MB_HINT_NOPREALLOC 0x0040 198 /* allocate for locality group */ 199 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 200 /* allocate goal blocks or none */ 201 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 202 /* goal is meaningful */ 203 #define EXT4_MB_HINT_TRY_GOAL 0x0200 204 /* blocks already pre-reserved by delayed allocation */ 205 #define EXT4_MB_DELALLOC_RESERVED 0x0400 206 /* We are doing stream allocation */ 207 #define EXT4_MB_STREAM_ALLOC 0x0800 208 /* Use reserved root blocks if needed */ 209 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 210 /* Use blocks from reserved pool */ 211 #define EXT4_MB_USE_RESERVED 0x2000 212 /* Do strict check for free blocks while retrying block allocation */ 213 #define EXT4_MB_STRICT_CHECK 0x4000 214 /* Large fragment size list lookup succeeded at least once for cr = 0 */ 215 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000 216 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */ 217 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000 218 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */ 219 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000 220 221 struct ext4_allocation_request { 222 /* target inode for block we're allocating */ 223 struct inode *inode; 224 /* how many blocks we want to allocate */ 225 unsigned int len; 226 /* logical block in target inode */ 227 ext4_lblk_t logical; 228 /* the closest logical allocated block to the left */ 229 ext4_lblk_t lleft; 230 /* the closest logical allocated block to the right */ 231 ext4_lblk_t lright; 232 /* phys. target (a hint) */ 233 ext4_fsblk_t goal; 234 /* phys. block for the closest logical allocated block to the left */ 235 ext4_fsblk_t pleft; 236 /* phys. block for the closest logical allocated block to the right */ 237 ext4_fsblk_t pright; 238 /* flags. see above EXT4_MB_HINT_* */ 239 unsigned int flags; 240 }; 241 242 /* 243 * Logical to physical block mapping, used by ext4_map_blocks() 244 * 245 * This structure is used to pass requests into ext4_map_blocks() as 246 * well as to store the information returned by ext4_map_blocks(). It 247 * takes less room on the stack than a struct buffer_head. 248 */ 249 #define EXT4_MAP_NEW BIT(BH_New) 250 #define EXT4_MAP_MAPPED BIT(BH_Mapped) 251 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 252 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 253 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 254 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 255 256 struct ext4_map_blocks { 257 ext4_fsblk_t m_pblk; 258 ext4_lblk_t m_lblk; 259 unsigned int m_len; 260 unsigned int m_flags; 261 }; 262 263 /* 264 * Block validity checking, system zone rbtree. 265 */ 266 struct ext4_system_blocks { 267 struct rb_root root; 268 struct rcu_head rcu; 269 }; 270 271 /* 272 * Flags for ext4_io_end->flags 273 */ 274 #define EXT4_IO_END_UNWRITTEN 0x0001 275 276 struct ext4_io_end_vec { 277 struct list_head list; /* list of io_end_vec */ 278 loff_t offset; /* offset in the file */ 279 ssize_t size; /* size of the extent */ 280 }; 281 282 /* 283 * For converting unwritten extents on a work queue. 'handle' is used for 284 * buffered writeback. 285 */ 286 typedef struct ext4_io_end { 287 struct list_head list; /* per-file finished IO list */ 288 handle_t *handle; /* handle reserved for extent 289 * conversion */ 290 struct inode *inode; /* file being written to */ 291 struct bio *bio; /* Linked list of completed 292 * bios covering the extent */ 293 unsigned int flag; /* unwritten or not */ 294 refcount_t count; /* reference counter */ 295 struct list_head list_vec; /* list of ext4_io_end_vec */ 296 } ext4_io_end_t; 297 298 struct ext4_io_submit { 299 struct writeback_control *io_wbc; 300 struct bio *io_bio; 301 ext4_io_end_t *io_end; 302 sector_t io_next_block; 303 }; 304 305 /* 306 * Special inodes numbers 307 */ 308 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 309 #define EXT4_ROOT_INO 2 /* Root inode */ 310 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 311 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 312 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 313 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 314 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 315 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 316 317 /* First non-reserved inode for old ext4 filesystems */ 318 #define EXT4_GOOD_OLD_FIRST_INO 11 319 320 /* 321 * Maximal count of links to a file 322 */ 323 #define EXT4_LINK_MAX 65000 324 325 /* 326 * Macro-instructions used to manage several block sizes 327 */ 328 #define EXT4_MIN_BLOCK_SIZE 1024 329 #define EXT4_MAX_BLOCK_SIZE 65536 330 #define EXT4_MIN_BLOCK_LOG_SIZE 10 331 #define EXT4_MAX_BLOCK_LOG_SIZE 16 332 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 333 #ifdef __KERNEL__ 334 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 335 #else 336 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 337 #endif 338 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 339 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 340 EXT4_SB(s)->s_cluster_bits) 341 #ifdef __KERNEL__ 342 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 343 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 344 #else 345 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 346 #endif 347 #ifdef __KERNEL__ 348 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 349 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 350 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 351 #else 352 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 353 EXT4_GOOD_OLD_INODE_SIZE : \ 354 (s)->s_inode_size) 355 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 356 EXT4_GOOD_OLD_FIRST_INO : \ 357 (s)->s_first_ino) 358 #endif 359 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 360 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 361 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 362 blkbits)) 363 364 /* Translate a block number to a cluster number */ 365 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 366 /* Translate a cluster number to a block number */ 367 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 368 /* Translate # of blks to # of clusters */ 369 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 370 (sbi)->s_cluster_bits) 371 /* Mask out the low bits to get the starting block of the cluster */ 372 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 373 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 374 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 375 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 376 /* Fill in the low bits to get the last block of the cluster */ 377 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 378 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 379 /* Get the cluster offset */ 380 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 381 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 382 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 383 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 384 385 /* 386 * Structure of a blocks group descriptor 387 */ 388 struct ext4_group_desc 389 { 390 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 391 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 392 __le32 bg_inode_table_lo; /* Inodes table block */ 393 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 394 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 395 __le16 bg_used_dirs_count_lo; /* Directories count */ 396 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 397 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 398 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 399 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 400 __le16 bg_itable_unused_lo; /* Unused inodes count */ 401 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 402 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 403 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 404 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 405 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 406 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 407 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 408 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 409 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 410 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 411 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 412 __u32 bg_reserved; 413 }; 414 415 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 416 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 417 sizeof(__le16)) 418 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 419 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 420 sizeof(__le16)) 421 422 /* 423 * Structure of a flex block group info 424 */ 425 426 struct flex_groups { 427 atomic64_t free_clusters; 428 atomic_t free_inodes; 429 atomic_t used_dirs; 430 }; 431 432 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 433 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 434 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 435 436 /* 437 * Macro-instructions used to manage group descriptors 438 */ 439 #define EXT4_MIN_DESC_SIZE 32 440 #define EXT4_MIN_DESC_SIZE_64BIT 64 441 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 442 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 443 #ifdef __KERNEL__ 444 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 445 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 446 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 447 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 448 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 449 #else 450 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 451 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 452 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 453 #endif 454 455 /* 456 * Constants relative to the data blocks 457 */ 458 #define EXT4_NDIR_BLOCKS 12 459 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 460 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 461 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 462 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 463 464 /* 465 * Inode flags 466 */ 467 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 468 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 469 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 470 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 471 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 472 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 473 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 474 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 475 /* Reserved for compression usage... */ 476 #define EXT4_DIRTY_FL 0x00000100 477 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 478 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 479 /* nb: was previously EXT2_ECOMPR_FL */ 480 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 481 /* End compression flags --- maybe not all used */ 482 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 483 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 484 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 485 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 486 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 487 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 488 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 489 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 490 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 491 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 492 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 493 494 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 495 496 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 497 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 498 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 499 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 500 501 /* User modifiable flags */ 502 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 503 EXT4_UNRM_FL | \ 504 EXT4_COMPR_FL | \ 505 EXT4_SYNC_FL | \ 506 EXT4_IMMUTABLE_FL | \ 507 EXT4_APPEND_FL | \ 508 EXT4_NODUMP_FL | \ 509 EXT4_NOATIME_FL | \ 510 EXT4_JOURNAL_DATA_FL | \ 511 EXT4_NOTAIL_FL | \ 512 EXT4_DIRSYNC_FL | \ 513 EXT4_TOPDIR_FL | \ 514 EXT4_EXTENTS_FL | \ 515 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 516 EXT4_DAX_FL | \ 517 EXT4_PROJINHERIT_FL | \ 518 EXT4_CASEFOLD_FL) 519 520 /* User visible flags */ 521 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 522 EXT4_DIRTY_FL | \ 523 EXT4_COMPRBLK_FL | \ 524 EXT4_NOCOMPR_FL | \ 525 EXT4_ENCRYPT_FL | \ 526 EXT4_INDEX_FL | \ 527 EXT4_VERITY_FL | \ 528 EXT4_INLINE_DATA_FL) 529 530 /* Flags that should be inherited by new inodes from their parent. */ 531 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 532 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 533 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 534 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 535 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 536 EXT4_DAX_FL) 537 538 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 539 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 540 EXT4_PROJINHERIT_FL)) 541 542 /* Flags that are appropriate for non-directories/regular files. */ 543 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 544 545 /* The only flags that should be swapped */ 546 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 547 548 /* Flags which are mutually exclusive to DAX */ 549 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 550 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 551 552 /* Mask out flags that are inappropriate for the given type of inode. */ 553 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 554 { 555 if (S_ISDIR(mode)) 556 return flags; 557 else if (S_ISREG(mode)) 558 return flags & EXT4_REG_FLMASK; 559 else 560 return flags & EXT4_OTHER_FLMASK; 561 } 562 563 /* 564 * Inode flags used for atomic set/get 565 */ 566 enum { 567 EXT4_INODE_SECRM = 0, /* Secure deletion */ 568 EXT4_INODE_UNRM = 1, /* Undelete */ 569 EXT4_INODE_COMPR = 2, /* Compress file */ 570 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 571 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 572 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 573 EXT4_INODE_NODUMP = 6, /* do not dump file */ 574 EXT4_INODE_NOATIME = 7, /* do not update atime */ 575 /* Reserved for compression usage... */ 576 EXT4_INODE_DIRTY = 8, 577 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 578 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 579 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 580 /* End compression flags --- maybe not all used */ 581 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 582 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 583 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 584 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 585 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 586 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 587 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 588 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 589 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 590 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 591 /* 22 was formerly EXT4_INODE_EOFBLOCKS */ 592 EXT4_INODE_DAX = 25, /* Inode is DAX */ 593 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 594 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 595 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 596 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 597 }; 598 599 /* 600 * Since it's pretty easy to mix up bit numbers and hex values, we use a 601 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 602 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 603 * any extra space in the compiled kernel image, otherwise, the build will fail. 604 * It's important that these values are the same, since we are using 605 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 606 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 607 * values found in ext2, ext3 and ext4 filesystems, and of course the values 608 * defined in e2fsprogs. 609 * 610 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 611 */ 612 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 613 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 614 615 static inline void ext4_check_flag_values(void) 616 { 617 CHECK_FLAG_VALUE(SECRM); 618 CHECK_FLAG_VALUE(UNRM); 619 CHECK_FLAG_VALUE(COMPR); 620 CHECK_FLAG_VALUE(SYNC); 621 CHECK_FLAG_VALUE(IMMUTABLE); 622 CHECK_FLAG_VALUE(APPEND); 623 CHECK_FLAG_VALUE(NODUMP); 624 CHECK_FLAG_VALUE(NOATIME); 625 CHECK_FLAG_VALUE(DIRTY); 626 CHECK_FLAG_VALUE(COMPRBLK); 627 CHECK_FLAG_VALUE(NOCOMPR); 628 CHECK_FLAG_VALUE(ENCRYPT); 629 CHECK_FLAG_VALUE(INDEX); 630 CHECK_FLAG_VALUE(IMAGIC); 631 CHECK_FLAG_VALUE(JOURNAL_DATA); 632 CHECK_FLAG_VALUE(NOTAIL); 633 CHECK_FLAG_VALUE(DIRSYNC); 634 CHECK_FLAG_VALUE(TOPDIR); 635 CHECK_FLAG_VALUE(HUGE_FILE); 636 CHECK_FLAG_VALUE(EXTENTS); 637 CHECK_FLAG_VALUE(VERITY); 638 CHECK_FLAG_VALUE(EA_INODE); 639 CHECK_FLAG_VALUE(INLINE_DATA); 640 CHECK_FLAG_VALUE(PROJINHERIT); 641 CHECK_FLAG_VALUE(CASEFOLD); 642 CHECK_FLAG_VALUE(RESERVED); 643 } 644 645 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 646 struct compat_ext4_new_group_input { 647 u32 group; 648 compat_u64 block_bitmap; 649 compat_u64 inode_bitmap; 650 compat_u64 inode_table; 651 u32 blocks_count; 652 u16 reserved_blocks; 653 u16 unused; 654 }; 655 #endif 656 657 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 658 struct ext4_new_group_data { 659 __u32 group; 660 __u64 block_bitmap; 661 __u64 inode_bitmap; 662 __u64 inode_table; 663 __u32 blocks_count; 664 __u16 reserved_blocks; 665 __u16 mdata_blocks; 666 __u32 free_clusters_count; 667 }; 668 669 /* Indexes used to index group tables in ext4_new_group_data */ 670 enum { 671 BLOCK_BITMAP = 0, /* block bitmap */ 672 INODE_BITMAP, /* inode bitmap */ 673 INODE_TABLE, /* inode tables */ 674 GROUP_TABLE_COUNT, 675 }; 676 677 /* 678 * Flags used by ext4_map_blocks() 679 */ 680 /* Allocate any needed blocks and/or convert an unwritten 681 extent to be an initialized ext4 */ 682 #define EXT4_GET_BLOCKS_CREATE 0x0001 683 /* Request the creation of an unwritten extent */ 684 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 685 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 686 EXT4_GET_BLOCKS_CREATE) 687 /* Caller is from the delayed allocation writeout path 688 * finally doing the actual allocation of delayed blocks */ 689 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 690 /* caller is from the direct IO path, request to creation of an 691 unwritten extents if not allocated, split the unwritten 692 extent if blocks has been preallocated already*/ 693 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 694 #define EXT4_GET_BLOCKS_CONVERT 0x0010 695 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 696 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 697 /* Convert extent to initialized after IO complete */ 698 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 699 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 700 /* Eventual metadata allocation (due to growing extent tree) 701 * should not fail, so try to use reserved blocks for that.*/ 702 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 703 /* Don't normalize allocation size (used for fallocate) */ 704 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 705 /* Convert written extents to unwritten */ 706 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 707 /* Write zeros to newly created written extents */ 708 #define EXT4_GET_BLOCKS_ZERO 0x0200 709 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 710 EXT4_GET_BLOCKS_ZERO) 711 /* Caller will submit data before dropping transaction handle. This 712 * allows jbd2 to avoid submitting data before commit. */ 713 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 714 /* Caller is in the atomic contex, find extent if it has been cached */ 715 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 716 717 /* 718 * The bit position of these flags must not overlap with any of the 719 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 720 * read_extent_tree_block(), ext4_split_extent_at(), 721 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 722 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 723 * caching the extents when reading from the extent tree while a 724 * truncate or punch hole operation is in progress. 725 */ 726 #define EXT4_EX_NOCACHE 0x40000000 727 #define EXT4_EX_FORCE_CACHE 0x20000000 728 #define EXT4_EX_NOFAIL 0x10000000 729 730 /* 731 * Flags used by ext4_free_blocks 732 */ 733 #define EXT4_FREE_BLOCKS_METADATA 0x0001 734 #define EXT4_FREE_BLOCKS_FORGET 0x0002 735 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 736 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 737 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 738 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 739 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 740 741 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 742 /* 743 * ioctl commands in 32 bit emulation 744 */ 745 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 746 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 747 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 748 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 749 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 750 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 751 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 752 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 753 #endif 754 755 /* Max physical block we can address w/o extents */ 756 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 757 758 /* Max logical block we can support */ 759 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 760 761 /* 762 * Structure of an inode on the disk 763 */ 764 struct ext4_inode { 765 __le16 i_mode; /* File mode */ 766 __le16 i_uid; /* Low 16 bits of Owner Uid */ 767 __le32 i_size_lo; /* Size in bytes */ 768 __le32 i_atime; /* Access time */ 769 __le32 i_ctime; /* Inode Change time */ 770 __le32 i_mtime; /* Modification time */ 771 __le32 i_dtime; /* Deletion Time */ 772 __le16 i_gid; /* Low 16 bits of Group Id */ 773 __le16 i_links_count; /* Links count */ 774 __le32 i_blocks_lo; /* Blocks count */ 775 __le32 i_flags; /* File flags */ 776 union { 777 struct { 778 __le32 l_i_version; 779 } linux1; 780 struct { 781 __u32 h_i_translator; 782 } hurd1; 783 struct { 784 __u32 m_i_reserved1; 785 } masix1; 786 } osd1; /* OS dependent 1 */ 787 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 788 __le32 i_generation; /* File version (for NFS) */ 789 __le32 i_file_acl_lo; /* File ACL */ 790 __le32 i_size_high; 791 __le32 i_obso_faddr; /* Obsoleted fragment address */ 792 union { 793 struct { 794 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 795 __le16 l_i_file_acl_high; 796 __le16 l_i_uid_high; /* these 2 fields */ 797 __le16 l_i_gid_high; /* were reserved2[0] */ 798 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 799 __le16 l_i_reserved; 800 } linux2; 801 struct { 802 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 803 __u16 h_i_mode_high; 804 __u16 h_i_uid_high; 805 __u16 h_i_gid_high; 806 __u32 h_i_author; 807 } hurd2; 808 struct { 809 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 810 __le16 m_i_file_acl_high; 811 __u32 m_i_reserved2[2]; 812 } masix2; 813 } osd2; /* OS dependent 2 */ 814 __le16 i_extra_isize; 815 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 816 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 817 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 818 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 819 __le32 i_crtime; /* File Creation time */ 820 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 821 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 822 __le32 i_projid; /* Project ID */ 823 }; 824 825 #define EXT4_EPOCH_BITS 2 826 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 827 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 828 829 /* 830 * Extended fields will fit into an inode if the filesystem was formatted 831 * with large inodes (-I 256 or larger) and there are not currently any EAs 832 * consuming all of the available space. For new inodes we always reserve 833 * enough space for the kernel's known extended fields, but for inodes 834 * created with an old kernel this might not have been the case. None of 835 * the extended inode fields is critical for correct filesystem operation. 836 * This macro checks if a certain field fits in the inode. Note that 837 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 838 */ 839 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 840 ((offsetof(typeof(*ext4_inode), field) + \ 841 sizeof((ext4_inode)->field)) \ 842 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 843 (einode)->i_extra_isize)) \ 844 845 /* 846 * We use an encoding that preserves the times for extra epoch "00": 847 * 848 * extra msb of adjust for signed 849 * epoch 32-bit 32-bit tv_sec to 850 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 851 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 852 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 853 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 854 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 855 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 856 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 857 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 858 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 859 * 860 * Note that previous versions of the kernel on 64-bit systems would 861 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 862 * 1970. e2fsck will correct this, assuming that it is run on the 863 * affected filesystem before 2242. 864 */ 865 866 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 867 { 868 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 869 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 870 } 871 872 static inline void ext4_decode_extra_time(struct timespec64 *time, 873 __le32 extra) 874 { 875 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 876 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 877 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 878 } 879 880 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 881 do { \ 882 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 883 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 884 (raw_inode)->xtime ## _extra = \ 885 ext4_encode_extra_time(&(inode)->xtime); \ 886 } \ 887 else \ 888 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \ 889 } while (0) 890 891 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 892 do { \ 893 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 894 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 895 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 896 (raw_inode)->xtime ## _extra = \ 897 ext4_encode_extra_time(&(einode)->xtime); \ 898 } while (0) 899 900 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 901 do { \ 902 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 903 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 904 ext4_decode_extra_time(&(inode)->xtime, \ 905 raw_inode->xtime ## _extra); \ 906 } \ 907 else \ 908 (inode)->xtime.tv_nsec = 0; \ 909 } while (0) 910 911 912 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 913 do { \ 914 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 915 (einode)->xtime.tv_sec = \ 916 (signed)le32_to_cpu((raw_inode)->xtime); \ 917 else \ 918 (einode)->xtime.tv_sec = 0; \ 919 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 920 ext4_decode_extra_time(&(einode)->xtime, \ 921 raw_inode->xtime ## _extra); \ 922 else \ 923 (einode)->xtime.tv_nsec = 0; \ 924 } while (0) 925 926 #define i_disk_version osd1.linux1.l_i_version 927 928 #if defined(__KERNEL__) || defined(__linux__) 929 #define i_reserved1 osd1.linux1.l_i_reserved1 930 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 931 #define i_blocks_high osd2.linux2.l_i_blocks_high 932 #define i_uid_low i_uid 933 #define i_gid_low i_gid 934 #define i_uid_high osd2.linux2.l_i_uid_high 935 #define i_gid_high osd2.linux2.l_i_gid_high 936 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 937 938 #elif defined(__GNU__) 939 940 #define i_translator osd1.hurd1.h_i_translator 941 #define i_uid_high osd2.hurd2.h_i_uid_high 942 #define i_gid_high osd2.hurd2.h_i_gid_high 943 #define i_author osd2.hurd2.h_i_author 944 945 #elif defined(__masix__) 946 947 #define i_reserved1 osd1.masix1.m_i_reserved1 948 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 949 #define i_reserved2 osd2.masix2.m_i_reserved2 950 951 #endif /* defined(__KERNEL__) || defined(__linux__) */ 952 953 #include "extents_status.h" 954 #include "fast_commit.h" 955 956 /* 957 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 958 * 959 * These are needed to avoid lockdep false positives when we need to 960 * allocate blocks to the quota inode during ext4_map_blocks(), while 961 * holding i_data_sem for a normal (non-quota) inode. Since we don't 962 * do quota tracking for the quota inode, this avoids deadlock (as 963 * well as infinite recursion, since it isn't turtles all the way 964 * down...) 965 * 966 * I_DATA_SEM_NORMAL - Used for most inodes 967 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 968 * where the second inode has larger inode number 969 * than the first 970 * I_DATA_SEM_QUOTA - Used for quota inodes only 971 * I_DATA_SEM_EA - Used for ea_inodes only 972 */ 973 enum { 974 I_DATA_SEM_NORMAL = 0, 975 I_DATA_SEM_OTHER, 976 I_DATA_SEM_QUOTA, 977 I_DATA_SEM_EA 978 }; 979 980 981 /* 982 * fourth extended file system inode data in memory 983 */ 984 struct ext4_inode_info { 985 __le32 i_data[15]; /* unconverted */ 986 __u32 i_dtime; 987 ext4_fsblk_t i_file_acl; 988 989 /* 990 * i_block_group is the number of the block group which contains 991 * this file's inode. Constant across the lifetime of the inode, 992 * it is used for making block allocation decisions - we try to 993 * place a file's data blocks near its inode block, and new inodes 994 * near to their parent directory's inode. 995 */ 996 ext4_group_t i_block_group; 997 ext4_lblk_t i_dir_start_lookup; 998 #if (BITS_PER_LONG < 64) 999 unsigned long i_state_flags; /* Dynamic state flags */ 1000 #endif 1001 unsigned long i_flags; 1002 1003 /* 1004 * Extended attributes can be read independently of the main file 1005 * data. Taking i_rwsem even when reading would cause contention 1006 * between readers of EAs and writers of regular file data, so 1007 * instead we synchronize on xattr_sem when reading or changing 1008 * EAs. 1009 */ 1010 struct rw_semaphore xattr_sem; 1011 1012 /* 1013 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1014 * i_orphan is used. 1015 */ 1016 union { 1017 struct list_head i_orphan; /* unlinked but open inodes */ 1018 unsigned int i_orphan_idx; /* Index in orphan file */ 1019 }; 1020 1021 /* Fast commit related info */ 1022 1023 /* For tracking dentry create updates */ 1024 struct list_head i_fc_dilist; 1025 struct list_head i_fc_list; /* 1026 * inodes that need fast commit 1027 * protected by sbi->s_fc_lock. 1028 */ 1029 1030 /* Start of lblk range that needs to be committed in this fast commit */ 1031 ext4_lblk_t i_fc_lblk_start; 1032 1033 /* End of lblk range that needs to be committed in this fast commit */ 1034 ext4_lblk_t i_fc_lblk_len; 1035 1036 /* Number of ongoing updates on this inode */ 1037 atomic_t i_fc_updates; 1038 1039 /* Fast commit wait queue for this inode */ 1040 wait_queue_head_t i_fc_wait; 1041 1042 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */ 1043 struct mutex i_fc_lock; 1044 1045 /* 1046 * i_disksize keeps track of what the inode size is ON DISK, not 1047 * in memory. During truncate, i_size is set to the new size by 1048 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1049 * set i_disksize to 0 until the truncate is actually under way. 1050 * 1051 * The intent is that i_disksize always represents the blocks which 1052 * are used by this file. This allows recovery to restart truncate 1053 * on orphans if we crash during truncate. We actually write i_disksize 1054 * into the on-disk inode when writing inodes out, instead of i_size. 1055 * 1056 * The only time when i_disksize and i_size may be different is when 1057 * a truncate is in progress. The only things which change i_disksize 1058 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1059 */ 1060 loff_t i_disksize; 1061 1062 /* 1063 * i_data_sem is for serialising ext4_truncate() against 1064 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1065 * data tree are chopped off during truncate. We can't do that in 1066 * ext4 because whenever we perform intermediate commits during 1067 * truncate, the inode and all the metadata blocks *must* be in a 1068 * consistent state which allows truncation of the orphans to restart 1069 * during recovery. Hence we must fix the get_block-vs-truncate race 1070 * by other means, so we have i_data_sem. 1071 */ 1072 struct rw_semaphore i_data_sem; 1073 struct inode vfs_inode; 1074 struct jbd2_inode *jinode; 1075 1076 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1077 1078 /* 1079 * File creation time. Its function is same as that of 1080 * struct timespec64 i_{a,c,m}time in the generic inode. 1081 */ 1082 struct timespec64 i_crtime; 1083 1084 /* mballoc */ 1085 atomic_t i_prealloc_active; 1086 struct rb_root i_prealloc_node; 1087 rwlock_t i_prealloc_lock; 1088 1089 /* extents status tree */ 1090 struct ext4_es_tree i_es_tree; 1091 rwlock_t i_es_lock; 1092 struct list_head i_es_list; 1093 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1094 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1095 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1096 extents to shrink. Protected by 1097 i_es_lock */ 1098 1099 /* ialloc */ 1100 ext4_group_t i_last_alloc_group; 1101 1102 /* allocation reservation info for delalloc */ 1103 /* In case of bigalloc, this refer to clusters rather than blocks */ 1104 unsigned int i_reserved_data_blocks; 1105 1106 /* pending cluster reservations for bigalloc file systems */ 1107 struct ext4_pending_tree i_pending_tree; 1108 1109 /* on-disk additional length */ 1110 __u16 i_extra_isize; 1111 1112 /* Indicate the inline data space. */ 1113 u16 i_inline_off; 1114 u16 i_inline_size; 1115 1116 #ifdef CONFIG_QUOTA 1117 /* quota space reservation, managed internally by quota code */ 1118 qsize_t i_reserved_quota; 1119 #endif 1120 1121 /* Lock protecting lists below */ 1122 spinlock_t i_completed_io_lock; 1123 /* 1124 * Completed IOs that need unwritten extents handling and have 1125 * transaction reserved 1126 */ 1127 struct list_head i_rsv_conversion_list; 1128 struct work_struct i_rsv_conversion_work; 1129 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1130 1131 spinlock_t i_block_reservation_lock; 1132 1133 /* 1134 * Transactions that contain inode's metadata needed to complete 1135 * fsync and fdatasync, respectively. 1136 */ 1137 tid_t i_sync_tid; 1138 tid_t i_datasync_tid; 1139 1140 #ifdef CONFIG_QUOTA 1141 struct dquot *i_dquot[MAXQUOTAS]; 1142 #endif 1143 1144 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1145 __u32 i_csum_seed; 1146 1147 kprojid_t i_projid; 1148 }; 1149 1150 /* 1151 * File system states 1152 */ 1153 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1154 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1155 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1156 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1157 1158 /* 1159 * Misc. filesystem flags 1160 */ 1161 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1162 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1163 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1164 1165 /* 1166 * Mount flags set via mount options or defaults 1167 */ 1168 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1169 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1170 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1171 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1172 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1173 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1174 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1175 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1176 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1177 #ifdef CONFIG_FS_DAX 1178 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1179 #else 1180 #define EXT4_MOUNT_DAX_ALWAYS 0 1181 #endif 1182 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1183 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1184 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1185 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1186 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1187 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1188 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1189 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1190 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1191 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1192 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1193 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1194 * enable enforcement for hidden 1195 * quota files */ 1196 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1197 * enforcement for hidden quota 1198 * files */ 1199 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1200 * enforcement */ 1201 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1202 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1203 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1204 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1205 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1206 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1207 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1208 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1209 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1210 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1211 1212 /* 1213 * Mount flags set either automatically (could not be set by mount option) 1214 * based on per file system feature or property or in special cases such as 1215 * distinguishing between explicit mount option definition and default. 1216 */ 1217 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1218 specified delalloc */ 1219 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1220 size of blocksize * 8 1221 blocks */ 1222 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1223 file systems */ 1224 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1225 specified journal checksum */ 1226 1227 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1228 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1229 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1230 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1231 * scanning in mballoc 1232 */ 1233 1234 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1235 ~EXT4_MOUNT_##opt 1236 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1237 EXT4_MOUNT_##opt 1238 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1239 EXT4_MOUNT_##opt) 1240 1241 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1242 ~EXT4_MOUNT2_##opt 1243 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1244 EXT4_MOUNT2_##opt 1245 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1246 EXT4_MOUNT2_##opt) 1247 1248 #define ext4_test_and_set_bit __test_and_set_bit_le 1249 #define ext4_set_bit __set_bit_le 1250 #define ext4_set_bit_atomic ext2_set_bit_atomic 1251 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1252 #define ext4_clear_bit __clear_bit_le 1253 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1254 #define ext4_test_bit test_bit_le 1255 #define ext4_find_next_zero_bit find_next_zero_bit_le 1256 #define ext4_find_next_bit find_next_bit_le 1257 1258 extern void mb_set_bits(void *bm, int cur, int len); 1259 1260 /* 1261 * Maximal mount counts between two filesystem checks 1262 */ 1263 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1264 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1265 1266 /* 1267 * Behaviour when detecting errors 1268 */ 1269 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1270 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1271 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1272 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1273 1274 /* Metadata checksum algorithm codes */ 1275 #define EXT4_CRC32C_CHKSUM 1 1276 1277 #define EXT4_LABEL_MAX 16 1278 1279 /* 1280 * Structure of the super block 1281 */ 1282 struct ext4_super_block { 1283 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1284 __le32 s_blocks_count_lo; /* Blocks count */ 1285 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1286 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1287 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1288 __le32 s_first_data_block; /* First Data Block */ 1289 __le32 s_log_block_size; /* Block size */ 1290 __le32 s_log_cluster_size; /* Allocation cluster size */ 1291 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1292 __le32 s_clusters_per_group; /* # Clusters per group */ 1293 __le32 s_inodes_per_group; /* # Inodes per group */ 1294 __le32 s_mtime; /* Mount time */ 1295 /*30*/ __le32 s_wtime; /* Write time */ 1296 __le16 s_mnt_count; /* Mount count */ 1297 __le16 s_max_mnt_count; /* Maximal mount count */ 1298 __le16 s_magic; /* Magic signature */ 1299 __le16 s_state; /* File system state */ 1300 __le16 s_errors; /* Behaviour when detecting errors */ 1301 __le16 s_minor_rev_level; /* minor revision level */ 1302 /*40*/ __le32 s_lastcheck; /* time of last check */ 1303 __le32 s_checkinterval; /* max. time between checks */ 1304 __le32 s_creator_os; /* OS */ 1305 __le32 s_rev_level; /* Revision level */ 1306 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1307 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1308 /* 1309 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1310 * 1311 * Note: the difference between the compatible feature set and 1312 * the incompatible feature set is that if there is a bit set 1313 * in the incompatible feature set that the kernel doesn't 1314 * know about, it should refuse to mount the filesystem. 1315 * 1316 * e2fsck's requirements are more strict; if it doesn't know 1317 * about a feature in either the compatible or incompatible 1318 * feature set, it must abort and not try to meddle with 1319 * things it doesn't understand... 1320 */ 1321 __le32 s_first_ino; /* First non-reserved inode */ 1322 __le16 s_inode_size; /* size of inode structure */ 1323 __le16 s_block_group_nr; /* block group # of this superblock */ 1324 __le32 s_feature_compat; /* compatible feature set */ 1325 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1326 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1327 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1328 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */ 1329 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1330 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1331 /* 1332 * Performance hints. Directory preallocation should only 1333 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1334 */ 1335 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1336 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1337 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1338 /* 1339 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1340 */ 1341 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1342 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1343 __le32 s_journal_dev; /* device number of journal file */ 1344 __le32 s_last_orphan; /* start of list of inodes to delete */ 1345 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1346 __u8 s_def_hash_version; /* Default hash version to use */ 1347 __u8 s_jnl_backup_type; 1348 __le16 s_desc_size; /* size of group descriptor */ 1349 /*100*/ __le32 s_default_mount_opts; 1350 __le32 s_first_meta_bg; /* First metablock block group */ 1351 __le32 s_mkfs_time; /* When the filesystem was created */ 1352 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1353 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1354 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1355 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1356 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1357 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1358 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1359 __le32 s_flags; /* Miscellaneous flags */ 1360 __le16 s_raid_stride; /* RAID stride */ 1361 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1362 __le64 s_mmp_block; /* Block for multi-mount protection */ 1363 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1364 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1365 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1366 __u8 s_encryption_level; /* versioning level for encryption */ 1367 __u8 s_reserved_pad; /* Padding to next 32bits */ 1368 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1369 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1370 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1371 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1372 snapshot's future use */ 1373 __le32 s_snapshot_list; /* inode number of the head of the 1374 on-disk snapshot list */ 1375 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1376 __le32 s_error_count; /* number of fs errors */ 1377 __le32 s_first_error_time; /* first time an error happened */ 1378 __le32 s_first_error_ino; /* inode involved in first error */ 1379 __le64 s_first_error_block; /* block involved of first error */ 1380 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1381 __le32 s_first_error_line; /* line number where error happened */ 1382 __le32 s_last_error_time; /* most recent time of an error */ 1383 __le32 s_last_error_ino; /* inode involved in last error */ 1384 __le32 s_last_error_line; /* line number where error happened */ 1385 __le64 s_last_error_block; /* block involved of last error */ 1386 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1387 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1388 __u8 s_mount_opts[64]; 1389 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1390 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1391 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1392 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1393 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1394 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1395 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1396 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1397 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1398 __u8 s_wtime_hi; 1399 __u8 s_mtime_hi; 1400 __u8 s_mkfs_time_hi; 1401 __u8 s_lastcheck_hi; 1402 __u8 s_first_error_time_hi; 1403 __u8 s_last_error_time_hi; 1404 __u8 s_first_error_errcode; 1405 __u8 s_last_error_errcode; 1406 __le16 s_encoding; /* Filename charset encoding */ 1407 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1408 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1409 __le32 s_reserved[94]; /* Padding to the end of the block */ 1410 __le32 s_checksum; /* crc32c(superblock) */ 1411 }; 1412 1413 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1414 1415 #ifdef __KERNEL__ 1416 1417 /* Number of quota types we support */ 1418 #define EXT4_MAXQUOTAS 3 1419 1420 #define EXT4_ENC_UTF8_12_1 1 1421 1422 /* Types of ext4 journal triggers */ 1423 enum ext4_journal_trigger_type { 1424 EXT4_JTR_ORPHAN_FILE, 1425 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1426 }; 1427 1428 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1429 1430 struct ext4_journal_trigger { 1431 struct jbd2_buffer_trigger_type tr_triggers; 1432 struct super_block *sb; 1433 }; 1434 1435 static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1436 struct jbd2_buffer_trigger_type *trigger) 1437 { 1438 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1439 } 1440 1441 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1442 1443 /* Structure at the tail of orphan block */ 1444 struct ext4_orphan_block_tail { 1445 __le32 ob_magic; 1446 __le32 ob_checksum; 1447 }; 1448 1449 static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1450 { 1451 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1452 sizeof(u32); 1453 } 1454 1455 struct ext4_orphan_block { 1456 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1457 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1458 }; 1459 1460 /* 1461 * Info about orphan file. 1462 */ 1463 struct ext4_orphan_info { 1464 int of_blocks; /* Number of orphan blocks in a file */ 1465 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1466 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1467 * file blocks */ 1468 }; 1469 1470 /* 1471 * fourth extended-fs super-block data in memory 1472 */ 1473 struct ext4_sb_info { 1474 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1475 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1476 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1477 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1478 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1479 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1480 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1481 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1482 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1483 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1484 unsigned long s_overhead; /* # of fs overhead clusters */ 1485 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1486 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1487 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1488 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1489 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1490 struct buffer_head * __rcu *s_group_desc; 1491 unsigned int s_mount_opt; 1492 unsigned int s_mount_opt2; 1493 unsigned long s_mount_flags; 1494 unsigned int s_def_mount_opt; 1495 unsigned int s_def_mount_opt2; 1496 ext4_fsblk_t s_sb_block; 1497 atomic64_t s_resv_clusters; 1498 kuid_t s_resuid; 1499 kgid_t s_resgid; 1500 unsigned short s_mount_state; 1501 unsigned short s_pad; 1502 int s_addr_per_block_bits; 1503 int s_desc_per_block_bits; 1504 int s_inode_size; 1505 int s_first_ino; 1506 unsigned int s_inode_readahead_blks; 1507 unsigned int s_inode_goal; 1508 u32 s_hash_seed[4]; 1509 int s_def_hash_version; 1510 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1511 struct percpu_counter s_freeclusters_counter; 1512 struct percpu_counter s_freeinodes_counter; 1513 struct percpu_counter s_dirs_counter; 1514 struct percpu_counter s_dirtyclusters_counter; 1515 struct percpu_counter s_sra_exceeded_retry_limit; 1516 struct blockgroup_lock *s_blockgroup_lock; 1517 struct proc_dir_entry *s_proc; 1518 struct kobject s_kobj; 1519 struct completion s_kobj_unregister; 1520 struct super_block *s_sb; 1521 struct buffer_head *s_mmp_bh; 1522 1523 /* Journaling */ 1524 struct journal_s *s_journal; 1525 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1526 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1527 struct list_head s_orphan; /* List of orphaned inodes in on disk 1528 list */ 1529 struct ext4_orphan_info s_orphan_info; 1530 unsigned long s_commit_interval; 1531 u32 s_max_batch_time; 1532 u32 s_min_batch_time; 1533 struct block_device *s_journal_bdev; 1534 #ifdef CONFIG_QUOTA 1535 /* Names of quota files with journalled quota */ 1536 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1537 int s_jquota_fmt; /* Format of quota to use */ 1538 #endif 1539 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1540 struct ext4_system_blocks __rcu *s_system_blks; 1541 1542 #ifdef EXTENTS_STATS 1543 /* ext4 extents stats */ 1544 unsigned long s_ext_min; 1545 unsigned long s_ext_max; 1546 unsigned long s_depth_max; 1547 spinlock_t s_ext_stats_lock; 1548 unsigned long s_ext_blocks; 1549 unsigned long s_ext_extents; 1550 #endif 1551 1552 /* for buddy allocator */ 1553 struct ext4_group_info ** __rcu *s_group_info; 1554 struct inode *s_buddy_cache; 1555 spinlock_t s_md_lock; 1556 unsigned short *s_mb_offsets; 1557 unsigned int *s_mb_maxs; 1558 unsigned int s_group_info_size; 1559 unsigned int s_mb_free_pending; 1560 struct list_head s_freed_data_list; /* List of blocks to be freed 1561 after commit completed */ 1562 struct list_head s_discard_list; 1563 struct work_struct s_discard_work; 1564 atomic_t s_retry_alloc_pending; 1565 struct list_head *s_mb_avg_fragment_size; 1566 rwlock_t *s_mb_avg_fragment_size_locks; 1567 struct list_head *s_mb_largest_free_orders; 1568 rwlock_t *s_mb_largest_free_orders_locks; 1569 1570 /* tunables */ 1571 unsigned long s_stripe; 1572 unsigned int s_mb_max_linear_groups; 1573 unsigned int s_mb_stream_request; 1574 unsigned int s_mb_max_to_scan; 1575 unsigned int s_mb_min_to_scan; 1576 unsigned int s_mb_stats; 1577 unsigned int s_mb_order2_reqs; 1578 unsigned int s_mb_group_prealloc; 1579 unsigned int s_max_dir_size_kb; 1580 /* where last allocation was done - for stream allocation */ 1581 unsigned long s_mb_last_group; 1582 unsigned long s_mb_last_start; 1583 unsigned int s_mb_prefetch; 1584 unsigned int s_mb_prefetch_limit; 1585 unsigned int s_mb_best_avail_max_trim_order; 1586 1587 /* stats for buddy allocator */ 1588 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1589 atomic_t s_bal_success; /* we found long enough chunks */ 1590 atomic_t s_bal_allocated; /* in blocks */ 1591 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1592 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1593 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1594 atomic_t s_bal_goals; /* goal hits */ 1595 atomic_t s_bal_len_goals; /* len goal hits */ 1596 atomic_t s_bal_breaks; /* too long searches */ 1597 atomic_t s_bal_2orders; /* 2^order hits */ 1598 atomic_t s_bal_p2_aligned_bad_suggestions; 1599 atomic_t s_bal_goal_fast_bad_suggestions; 1600 atomic_t s_bal_best_avail_bad_suggestions; 1601 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1602 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1603 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1604 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1605 atomic64_t s_mb_generation_time; 1606 atomic_t s_mb_lost_chunks; 1607 atomic_t s_mb_preallocated; 1608 atomic_t s_mb_discarded; 1609 atomic_t s_lock_busy; 1610 1611 /* locality groups */ 1612 struct ext4_locality_group __percpu *s_locality_groups; 1613 1614 /* for write statistics */ 1615 unsigned long s_sectors_written_start; 1616 u64 s_kbytes_written; 1617 1618 /* the size of zero-out chunk */ 1619 unsigned int s_extent_max_zeroout_kb; 1620 1621 unsigned int s_log_groups_per_flex; 1622 struct flex_groups * __rcu *s_flex_groups; 1623 ext4_group_t s_flex_groups_allocated; 1624 1625 /* workqueue for reserved extent conversions (buffered io) */ 1626 struct workqueue_struct *rsv_conversion_wq; 1627 1628 /* timer for periodic error stats printing */ 1629 struct timer_list s_err_report; 1630 1631 /* Lazy inode table initialization info */ 1632 struct ext4_li_request *s_li_request; 1633 /* Wait multiplier for lazy initialization thread */ 1634 unsigned int s_li_wait_mult; 1635 1636 /* Kernel thread for multiple mount protection */ 1637 struct task_struct *s_mmp_tsk; 1638 1639 /* record the last minlen when FITRIM is called. */ 1640 unsigned long s_last_trim_minblks; 1641 1642 /* Reference to checksum algorithm driver via cryptoapi */ 1643 struct crypto_shash *s_chksum_driver; 1644 1645 /* Precomputed FS UUID checksum for seeding other checksums */ 1646 __u32 s_csum_seed; 1647 1648 /* Reclaim extents from extent status tree */ 1649 struct shrinker s_es_shrinker; 1650 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1651 long s_es_nr_inode; 1652 struct ext4_es_stats s_es_stats; 1653 struct mb_cache *s_ea_block_cache; 1654 struct mb_cache *s_ea_inode_cache; 1655 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1656 1657 /* Journal triggers for checksum computation */ 1658 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1659 1660 /* Ratelimit ext4 messages. */ 1661 struct ratelimit_state s_err_ratelimit_state; 1662 struct ratelimit_state s_warning_ratelimit_state; 1663 struct ratelimit_state s_msg_ratelimit_state; 1664 atomic_t s_warning_count; 1665 atomic_t s_msg_count; 1666 1667 /* Encryption policy for '-o test_dummy_encryption' */ 1668 struct fscrypt_dummy_policy s_dummy_enc_policy; 1669 1670 /* 1671 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1672 * or EXTENTS flag. 1673 */ 1674 struct percpu_rw_semaphore s_writepages_rwsem; 1675 struct dax_device *s_daxdev; 1676 u64 s_dax_part_off; 1677 #ifdef CONFIG_EXT4_DEBUG 1678 unsigned long s_simulate_fail; 1679 #endif 1680 /* Record the errseq of the backing block device */ 1681 errseq_t s_bdev_wb_err; 1682 spinlock_t s_bdev_wb_lock; 1683 1684 /* Information about errors that happened during this mount */ 1685 spinlock_t s_error_lock; 1686 int s_add_error_count; 1687 int s_first_error_code; 1688 __u32 s_first_error_line; 1689 __u32 s_first_error_ino; 1690 __u64 s_first_error_block; 1691 const char *s_first_error_func; 1692 time64_t s_first_error_time; 1693 int s_last_error_code; 1694 __u32 s_last_error_line; 1695 __u32 s_last_error_ino; 1696 __u64 s_last_error_block; 1697 const char *s_last_error_func; 1698 time64_t s_last_error_time; 1699 /* 1700 * If we are in a context where we cannot update error information in 1701 * the on-disk superblock, we queue this work to do it. 1702 */ 1703 struct work_struct s_error_work; 1704 1705 /* Ext4 fast commit sub transaction ID */ 1706 atomic_t s_fc_subtid; 1707 1708 /* 1709 * After commit starts, the main queue gets locked, and the further 1710 * updates get added in the staging queue. 1711 */ 1712 #define FC_Q_MAIN 0 1713 #define FC_Q_STAGING 1 1714 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1715 * that have data changes in them. 1716 */ 1717 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1718 unsigned int s_fc_bytes; 1719 /* 1720 * Main fast commit lock. This lock protects accesses to the 1721 * following fields: 1722 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1723 */ 1724 spinlock_t s_fc_lock; 1725 struct buffer_head *s_fc_bh; 1726 struct ext4_fc_stats s_fc_stats; 1727 tid_t s_fc_ineligible_tid; 1728 #ifdef CONFIG_EXT4_DEBUG 1729 int s_fc_debug_max_replay; 1730 #endif 1731 struct ext4_fc_replay_state s_fc_replay_state; 1732 }; 1733 1734 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1735 { 1736 return sb->s_fs_info; 1737 } 1738 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1739 { 1740 return container_of(inode, struct ext4_inode_info, vfs_inode); 1741 } 1742 1743 static inline int ext4_writepages_down_read(struct super_block *sb) 1744 { 1745 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1746 return memalloc_nofs_save(); 1747 } 1748 1749 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1750 { 1751 memalloc_nofs_restore(ctx); 1752 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1753 } 1754 1755 static inline int ext4_writepages_down_write(struct super_block *sb) 1756 { 1757 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1758 return memalloc_nofs_save(); 1759 } 1760 1761 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1762 { 1763 memalloc_nofs_restore(ctx); 1764 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1765 } 1766 1767 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1768 { 1769 return ino == EXT4_ROOT_INO || 1770 (ino >= EXT4_FIRST_INO(sb) && 1771 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1772 } 1773 1774 /* 1775 * Returns: sbi->field[index] 1776 * Used to access an array element from the following sbi fields which require 1777 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1778 * - s_group_desc 1779 * - s_group_info 1780 * - s_flex_group 1781 */ 1782 #define sbi_array_rcu_deref(sbi, field, index) \ 1783 ({ \ 1784 typeof(*((sbi)->field)) _v; \ 1785 rcu_read_lock(); \ 1786 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1787 rcu_read_unlock(); \ 1788 _v; \ 1789 }) 1790 1791 /* 1792 * run-time mount flags 1793 */ 1794 enum { 1795 EXT4_MF_MNTDIR_SAMPLED, 1796 EXT4_MF_FS_ABORTED, /* Fatal error detected */ 1797 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */ 1798 }; 1799 1800 static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1801 { 1802 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1803 } 1804 1805 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1806 { 1807 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1808 } 1809 1810 static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1811 { 1812 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1813 } 1814 1815 1816 /* 1817 * Simulate_fail codes 1818 */ 1819 #define EXT4_SIM_BBITMAP_EIO 1 1820 #define EXT4_SIM_BBITMAP_CRC 2 1821 #define EXT4_SIM_IBITMAP_EIO 3 1822 #define EXT4_SIM_IBITMAP_CRC 4 1823 #define EXT4_SIM_INODE_EIO 5 1824 #define EXT4_SIM_INODE_CRC 6 1825 #define EXT4_SIM_DIRBLOCK_EIO 7 1826 #define EXT4_SIM_DIRBLOCK_CRC 8 1827 1828 static inline bool ext4_simulate_fail(struct super_block *sb, 1829 unsigned long code) 1830 { 1831 #ifdef CONFIG_EXT4_DEBUG 1832 struct ext4_sb_info *sbi = EXT4_SB(sb); 1833 1834 if (unlikely(sbi->s_simulate_fail == code)) { 1835 sbi->s_simulate_fail = 0; 1836 return true; 1837 } 1838 #endif 1839 return false; 1840 } 1841 1842 static inline void ext4_simulate_fail_bh(struct super_block *sb, 1843 struct buffer_head *bh, 1844 unsigned long code) 1845 { 1846 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code)) 1847 clear_buffer_uptodate(bh); 1848 } 1849 1850 /* 1851 * Error number codes for s_{first,last}_error_errno 1852 * 1853 * Linux errno numbers are architecture specific, so we need to translate 1854 * them into something which is architecture independent. We don't define 1855 * codes for all errno's; just the ones which are most likely to be the cause 1856 * of an ext4_error() call. 1857 */ 1858 #define EXT4_ERR_UNKNOWN 1 1859 #define EXT4_ERR_EIO 2 1860 #define EXT4_ERR_ENOMEM 3 1861 #define EXT4_ERR_EFSBADCRC 4 1862 #define EXT4_ERR_EFSCORRUPTED 5 1863 #define EXT4_ERR_ENOSPC 6 1864 #define EXT4_ERR_ENOKEY 7 1865 #define EXT4_ERR_EROFS 8 1866 #define EXT4_ERR_EFBIG 9 1867 #define EXT4_ERR_EEXIST 10 1868 #define EXT4_ERR_ERANGE 11 1869 #define EXT4_ERR_EOVERFLOW 12 1870 #define EXT4_ERR_EBUSY 13 1871 #define EXT4_ERR_ENOTDIR 14 1872 #define EXT4_ERR_ENOTEMPTY 15 1873 #define EXT4_ERR_ESHUTDOWN 16 1874 #define EXT4_ERR_EFAULT 17 1875 1876 /* 1877 * Inode dynamic state flags 1878 */ 1879 enum { 1880 EXT4_STATE_NEW, /* inode is newly created */ 1881 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1882 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1883 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1884 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1885 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1886 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1887 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1888 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1889 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1890 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1891 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1892 }; 1893 1894 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1895 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1896 { \ 1897 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1898 } \ 1899 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1900 { \ 1901 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1902 } \ 1903 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1904 { \ 1905 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1906 } 1907 1908 /* Add these declarations here only so that these functions can be 1909 * found by name. Otherwise, they are very hard to locate. */ 1910 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1911 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1912 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1913 EXT4_INODE_BIT_FNS(flag, flags, 0) 1914 1915 /* Add these declarations here only so that these functions can be 1916 * found by name. Otherwise, they are very hard to locate. */ 1917 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1918 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1919 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1920 #if (BITS_PER_LONG < 64) 1921 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1922 1923 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1924 { 1925 (ei)->i_state_flags = 0; 1926 } 1927 #else 1928 EXT4_INODE_BIT_FNS(state, flags, 32) 1929 1930 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1931 { 1932 /* We depend on the fact that callers will set i_flags */ 1933 } 1934 #endif 1935 #else 1936 /* Assume that user mode programs are passing in an ext4fs superblock, not 1937 * a kernel struct super_block. This will allow us to call the feature-test 1938 * macros from user land. */ 1939 #define EXT4_SB(sb) (sb) 1940 #endif 1941 1942 static inline bool ext4_verity_in_progress(struct inode *inode) 1943 { 1944 return IS_ENABLED(CONFIG_FS_VERITY) && 1945 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 1946 } 1947 1948 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1949 1950 /* 1951 * Codes for operating systems 1952 */ 1953 #define EXT4_OS_LINUX 0 1954 #define EXT4_OS_HURD 1 1955 #define EXT4_OS_MASIX 2 1956 #define EXT4_OS_FREEBSD 3 1957 #define EXT4_OS_LITES 4 1958 1959 /* 1960 * Revision levels 1961 */ 1962 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1963 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1964 1965 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1966 1967 #define EXT4_GOOD_OLD_INODE_SIZE 128 1968 1969 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 1970 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 1971 #define EXT4_TIMESTAMP_MIN S32_MIN 1972 1973 /* 1974 * Feature set definitions 1975 */ 1976 1977 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1978 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1979 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1980 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1981 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1982 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1983 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1984 /* 1985 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 1986 * incompatible only if fast commit blocks are present in the FS. Since we 1987 * clear the journal (and thus the fast commit blocks), we don't mark FS as 1988 * incompatible. We also have a JBD2 incompat feature, which gets set when 1989 * there are fast commit blocks present in the journal. 1990 */ 1991 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 1992 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 1993 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 1994 1995 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1996 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1997 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1998 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1999 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2000 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2001 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2002 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2003 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2004 /* 2005 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2006 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2007 * all other data structures' checksums. However, the METADATA_CSUM and 2008 * GDT_CSUM bits are mutually exclusive. 2009 */ 2010 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2011 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2012 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2013 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2014 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2015 non-empty */ 2016 2017 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2018 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2019 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2020 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2021 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2022 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2023 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2024 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2025 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2026 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2027 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2028 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2029 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2030 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2031 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2032 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2033 2034 extern void ext4_update_dynamic_rev(struct super_block *sb); 2035 2036 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2037 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2038 { \ 2039 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2040 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2041 } \ 2042 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2043 { \ 2044 ext4_update_dynamic_rev(sb); \ 2045 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2046 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2047 } \ 2048 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2049 { \ 2050 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2051 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2052 } 2053 2054 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2055 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2056 { \ 2057 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2058 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2059 } \ 2060 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2061 { \ 2062 ext4_update_dynamic_rev(sb); \ 2063 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2064 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2065 } \ 2066 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2067 { \ 2068 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2069 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2070 } 2071 2072 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2073 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2074 { \ 2075 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2076 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2077 } \ 2078 static inline void ext4_set_feature_##name(struct super_block *sb) \ 2079 { \ 2080 ext4_update_dynamic_rev(sb); \ 2081 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2082 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2083 } \ 2084 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2085 { \ 2086 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2087 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2088 } 2089 2090 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2091 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2092 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2093 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2094 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2095 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2096 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2097 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2098 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2099 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2100 2101 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2102 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2103 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2104 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2105 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2106 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2107 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2108 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2109 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2110 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2111 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2112 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2113 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2114 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2115 2116 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2117 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2118 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2119 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2120 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2121 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2122 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2123 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2124 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2125 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2126 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2127 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2128 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2129 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2130 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2131 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2132 2133 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2134 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2135 EXT4_FEATURE_INCOMPAT_META_BG) 2136 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2137 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2138 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2139 2140 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2141 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2142 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2143 EXT4_FEATURE_INCOMPAT_META_BG) 2144 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2145 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2146 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2147 2148 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2149 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2150 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2151 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2152 EXT4_FEATURE_INCOMPAT_META_BG| \ 2153 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2154 EXT4_FEATURE_INCOMPAT_64BIT| \ 2155 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2156 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2157 EXT4_FEATURE_INCOMPAT_MMP | \ 2158 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2159 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2160 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2161 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2162 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2163 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2164 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2165 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2166 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2167 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2168 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2169 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2170 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2171 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2172 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2173 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2174 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2175 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2176 2177 #define EXTN_FEATURE_FUNCS(ver) \ 2178 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2179 { \ 2180 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2181 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2182 } \ 2183 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2184 { \ 2185 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2186 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2187 } \ 2188 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2189 { \ 2190 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2191 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2192 } 2193 2194 EXTN_FEATURE_FUNCS(2) 2195 EXTN_FEATURE_FUNCS(3) 2196 EXTN_FEATURE_FUNCS(4) 2197 2198 static inline bool ext4_has_compat_features(struct super_block *sb) 2199 { 2200 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2201 } 2202 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2203 { 2204 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2205 } 2206 static inline bool ext4_has_incompat_features(struct super_block *sb) 2207 { 2208 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2209 } 2210 2211 extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2212 2213 /* 2214 * Superblock flags 2215 */ 2216 #define EXT4_FLAGS_RESIZING 0 2217 #define EXT4_FLAGS_SHUTDOWN 1 2218 #define EXT4_FLAGS_BDEV_IS_DAX 2 2219 2220 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 2221 { 2222 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 2223 } 2224 2225 /* 2226 * Default values for user and/or group using reserved blocks 2227 */ 2228 #define EXT4_DEF_RESUID 0 2229 #define EXT4_DEF_RESGID 0 2230 2231 /* 2232 * Default project ID 2233 */ 2234 #define EXT4_DEF_PROJID 0 2235 2236 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 2237 2238 /* 2239 * Default mount options 2240 */ 2241 #define EXT4_DEFM_DEBUG 0x0001 2242 #define EXT4_DEFM_BSDGROUPS 0x0002 2243 #define EXT4_DEFM_XATTR_USER 0x0004 2244 #define EXT4_DEFM_ACL 0x0008 2245 #define EXT4_DEFM_UID16 0x0010 2246 #define EXT4_DEFM_JMODE 0x0060 2247 #define EXT4_DEFM_JMODE_DATA 0x0020 2248 #define EXT4_DEFM_JMODE_ORDERED 0x0040 2249 #define EXT4_DEFM_JMODE_WBACK 0x0060 2250 #define EXT4_DEFM_NOBARRIER 0x0100 2251 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2252 #define EXT4_DEFM_DISCARD 0x0400 2253 #define EXT4_DEFM_NODELALLOC 0x0800 2254 2255 /* 2256 * Default journal batch times 2257 */ 2258 #define EXT4_DEF_MIN_BATCH_TIME 0 2259 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2260 2261 /* 2262 * Minimum number of groups in a flexgroup before we separate out 2263 * directories into the first block group of a flexgroup 2264 */ 2265 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2266 2267 /* 2268 * Structure of a directory entry 2269 */ 2270 #define EXT4_NAME_LEN 255 2271 /* 2272 * Base length of the ext4 directory entry excluding the name length 2273 */ 2274 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2275 2276 struct ext4_dir_entry { 2277 __le32 inode; /* Inode number */ 2278 __le16 rec_len; /* Directory entry length */ 2279 __le16 name_len; /* Name length */ 2280 char name[EXT4_NAME_LEN]; /* File name */ 2281 }; 2282 2283 2284 /* 2285 * Encrypted Casefolded entries require saving the hash on disk. This structure 2286 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2287 * boundary. 2288 */ 2289 struct ext4_dir_entry_hash { 2290 __le32 hash; 2291 __le32 minor_hash; 2292 }; 2293 2294 /* 2295 * The new version of the directory entry. Since EXT4 structures are 2296 * stored in intel byte order, and the name_len field could never be 2297 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2298 * file_type field. 2299 */ 2300 struct ext4_dir_entry_2 { 2301 __le32 inode; /* Inode number */ 2302 __le16 rec_len; /* Directory entry length */ 2303 __u8 name_len; /* Name length */ 2304 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2305 char name[EXT4_NAME_LEN]; /* File name */ 2306 }; 2307 2308 /* 2309 * Access the hashes at the end of ext4_dir_entry_2 2310 */ 2311 #define EXT4_DIRENT_HASHES(entry) \ 2312 ((struct ext4_dir_entry_hash *) \ 2313 (((void *)(entry)) + \ 2314 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2315 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash) 2316 #define EXT4_DIRENT_MINOR_HASH(entry) \ 2317 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash) 2318 2319 static inline bool ext4_hash_in_dirent(const struct inode *inode) 2320 { 2321 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2322 } 2323 2324 /* 2325 * This is a bogus directory entry at the end of each leaf block that 2326 * records checksums. 2327 */ 2328 struct ext4_dir_entry_tail { 2329 __le32 det_reserved_zero1; /* Pretend to be unused */ 2330 __le16 det_rec_len; /* 12 */ 2331 __u8 det_reserved_zero2; /* Zero name length */ 2332 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2333 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2334 }; 2335 2336 #define EXT4_DIRENT_TAIL(block, blocksize) \ 2337 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2338 ((blocksize) - \ 2339 sizeof(struct ext4_dir_entry_tail)))) 2340 2341 /* 2342 * Ext4 directory file types. Only the low 3 bits are used. The 2343 * other bits are reserved for now. 2344 */ 2345 #define EXT4_FT_UNKNOWN 0 2346 #define EXT4_FT_REG_FILE 1 2347 #define EXT4_FT_DIR 2 2348 #define EXT4_FT_CHRDEV 3 2349 #define EXT4_FT_BLKDEV 4 2350 #define EXT4_FT_FIFO 5 2351 #define EXT4_FT_SOCK 6 2352 #define EXT4_FT_SYMLINK 7 2353 2354 #define EXT4_FT_MAX 8 2355 2356 #define EXT4_FT_DIR_CSUM 0xDE 2357 2358 /* 2359 * EXT4_DIR_PAD defines the directory entries boundaries 2360 * 2361 * NOTE: It must be a multiple of 4 2362 */ 2363 #define EXT4_DIR_PAD 4 2364 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2365 #define EXT4_MAX_REC_LEN ((1<<16)-1) 2366 2367 /* 2368 * The rec_len is dependent on the type of directory. Directories that are 2369 * casefolded and encrypted need to store the hash as well, so we add room for 2370 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2371 * pass NULL for dir, as those entries do not use the extra fields. 2372 */ 2373 static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2374 const struct inode *dir) 2375 { 2376 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2377 2378 if (dir && ext4_hash_in_dirent(dir)) 2379 rec_len += sizeof(struct ext4_dir_entry_hash); 2380 return (rec_len & ~EXT4_DIR_ROUND); 2381 } 2382 2383 /* 2384 * If we ever get support for fs block sizes > page_size, we'll need 2385 * to remove the #if statements in the next two functions... 2386 */ 2387 static inline unsigned int 2388 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2389 { 2390 unsigned len = le16_to_cpu(dlen); 2391 2392 #if (PAGE_SIZE >= 65536) 2393 if (len == EXT4_MAX_REC_LEN || len == 0) 2394 return blocksize; 2395 return (len & 65532) | ((len & 3) << 16); 2396 #else 2397 return len; 2398 #endif 2399 } 2400 2401 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2402 { 2403 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2404 #if (PAGE_SIZE >= 65536) 2405 if (len < 65536) 2406 return cpu_to_le16(len); 2407 if (len == blocksize) { 2408 if (blocksize == 65536) 2409 return cpu_to_le16(EXT4_MAX_REC_LEN); 2410 else 2411 return cpu_to_le16(0); 2412 } 2413 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2414 #else 2415 return cpu_to_le16(len); 2416 #endif 2417 } 2418 2419 /* 2420 * Hash Tree Directory indexing 2421 * (c) Daniel Phillips, 2001 2422 */ 2423 2424 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2425 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2426 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2427 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2428 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2429 2430 /* Legal values for the dx_root hash_version field: */ 2431 2432 #define DX_HASH_LEGACY 0 2433 #define DX_HASH_HALF_MD4 1 2434 #define DX_HASH_TEA 2 2435 #define DX_HASH_LEGACY_UNSIGNED 3 2436 #define DX_HASH_HALF_MD4_UNSIGNED 4 2437 #define DX_HASH_TEA_UNSIGNED 5 2438 #define DX_HASH_SIPHASH 6 2439 2440 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2441 const void *address, unsigned int length) 2442 { 2443 struct { 2444 struct shash_desc shash; 2445 char ctx[4]; 2446 } desc; 2447 2448 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2449 2450 desc.shash.tfm = sbi->s_chksum_driver; 2451 *(u32 *)desc.ctx = crc; 2452 2453 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2454 2455 return *(u32 *)desc.ctx; 2456 } 2457 2458 #ifdef __KERNEL__ 2459 2460 /* hash info structure used by the directory hash */ 2461 struct dx_hash_info 2462 { 2463 u32 hash; 2464 u32 minor_hash; 2465 int hash_version; 2466 u32 *seed; 2467 }; 2468 2469 2470 /* 32 and 64 bit signed EOF for dx directories */ 2471 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2472 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2473 2474 2475 /* 2476 * Control parameters used by ext4_htree_next_block 2477 */ 2478 #define HASH_NB_ALWAYS 1 2479 2480 struct ext4_filename { 2481 const struct qstr *usr_fname; 2482 struct fscrypt_str disk_name; 2483 struct dx_hash_info hinfo; 2484 #ifdef CONFIG_FS_ENCRYPTION 2485 struct fscrypt_str crypto_buf; 2486 #endif 2487 #if IS_ENABLED(CONFIG_UNICODE) 2488 struct fscrypt_str cf_name; 2489 #endif 2490 }; 2491 2492 #define fname_name(p) ((p)->disk_name.name) 2493 #define fname_usr_name(p) ((p)->usr_fname->name) 2494 #define fname_len(p) ((p)->disk_name.len) 2495 2496 /* 2497 * Describe an inode's exact location on disk and in memory 2498 */ 2499 struct ext4_iloc 2500 { 2501 struct buffer_head *bh; 2502 unsigned long offset; 2503 ext4_group_t block_group; 2504 }; 2505 2506 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2507 { 2508 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2509 } 2510 2511 static inline bool ext4_is_quota_file(struct inode *inode) 2512 { 2513 return IS_NOQUOTA(inode) && 2514 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2515 } 2516 2517 /* 2518 * This structure is stuffed into the struct file's private_data field 2519 * for directories. It is where we put information so that we can do 2520 * readdir operations in hash tree order. 2521 */ 2522 struct dir_private_info { 2523 struct rb_root root; 2524 struct rb_node *curr_node; 2525 struct fname *extra_fname; 2526 loff_t last_pos; 2527 __u32 curr_hash; 2528 __u32 curr_minor_hash; 2529 __u32 next_hash; 2530 }; 2531 2532 /* calculate the first block number of the group */ 2533 static inline ext4_fsblk_t 2534 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2535 { 2536 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2537 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2538 } 2539 2540 /* 2541 * Special error return code only used by dx_probe() and its callers. 2542 */ 2543 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2544 2545 /* htree levels for ext4 */ 2546 #define EXT4_HTREE_LEVEL_COMPAT 2 2547 #define EXT4_HTREE_LEVEL 3 2548 2549 static inline int ext4_dir_htree_level(struct super_block *sb) 2550 { 2551 return ext4_has_feature_largedir(sb) ? 2552 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2553 } 2554 2555 /* 2556 * Timeout and state flag for lazy initialization inode thread. 2557 */ 2558 #define EXT4_DEF_LI_WAIT_MULT 10 2559 #define EXT4_DEF_LI_MAX_START_DELAY 5 2560 #define EXT4_LAZYINIT_QUIT 0x0001 2561 #define EXT4_LAZYINIT_RUNNING 0x0002 2562 2563 /* 2564 * Lazy inode table initialization info 2565 */ 2566 struct ext4_lazy_init { 2567 unsigned long li_state; 2568 struct list_head li_request_list; 2569 struct mutex li_list_mtx; 2570 }; 2571 2572 enum ext4_li_mode { 2573 EXT4_LI_MODE_PREFETCH_BBITMAP, 2574 EXT4_LI_MODE_ITABLE, 2575 }; 2576 2577 struct ext4_li_request { 2578 struct super_block *lr_super; 2579 enum ext4_li_mode lr_mode; 2580 ext4_group_t lr_first_not_zeroed; 2581 ext4_group_t lr_next_group; 2582 struct list_head lr_request; 2583 unsigned long lr_next_sched; 2584 unsigned long lr_timeout; 2585 }; 2586 2587 struct ext4_features { 2588 struct kobject f_kobj; 2589 struct completion f_kobj_unregister; 2590 }; 2591 2592 /* 2593 * This structure will be used for multiple mount protection. It will be 2594 * written into the block number saved in the s_mmp_block field in the 2595 * superblock. Programs that check MMP should assume that if 2596 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2597 * to use the filesystem, regardless of how old the timestamp is. 2598 */ 2599 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2600 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2601 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2602 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2603 2604 struct mmp_struct { 2605 __le32 mmp_magic; /* Magic number for MMP */ 2606 __le32 mmp_seq; /* Sequence no. updated periodically */ 2607 2608 /* 2609 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2610 * purposes and do not affect the correctness of the algorithm 2611 */ 2612 __le64 mmp_time; /* Time last updated */ 2613 char mmp_nodename[64]; /* Node which last updated MMP block */ 2614 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2615 2616 /* 2617 * mmp_check_interval is used to verify if the MMP block has been 2618 * updated on the block device. The value is updated based on the 2619 * maximum time to write the MMP block during an update cycle. 2620 */ 2621 __le16 mmp_check_interval; 2622 2623 __le16 mmp_pad1; 2624 __le32 mmp_pad2[226]; 2625 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2626 }; 2627 2628 /* arguments passed to the mmp thread */ 2629 struct mmpd_data { 2630 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2631 struct super_block *sb; /* super block of the fs */ 2632 }; 2633 2634 /* 2635 * Check interval multiplier 2636 * The MMP block is written every update interval and initially checked every 2637 * update interval x the multiplier (the value is then adapted based on the 2638 * write latency). The reason is that writes can be delayed under load and we 2639 * don't want readers to incorrectly assume that the filesystem is no longer 2640 * in use. 2641 */ 2642 #define EXT4_MMP_CHECK_MULT 2UL 2643 2644 /* 2645 * Minimum interval for MMP checking in seconds. 2646 */ 2647 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2648 2649 /* 2650 * Maximum interval for MMP checking in seconds. 2651 */ 2652 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2653 2654 /* 2655 * Function prototypes 2656 */ 2657 2658 /* 2659 * Ok, these declarations are also in <linux/kernel.h> but none of the 2660 * ext4 source programs needs to include it so they are duplicated here. 2661 */ 2662 # define NORET_TYPE /**/ 2663 # define ATTRIB_NORET __attribute__((noreturn)) 2664 # define NORET_AND noreturn, 2665 2666 /* bitmap.c */ 2667 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2668 void ext4_inode_bitmap_csum_set(struct super_block *sb, 2669 struct ext4_group_desc *gdp, 2670 struct buffer_head *bh, int sz); 2671 int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2672 struct ext4_group_desc *gdp, 2673 struct buffer_head *bh, int sz); 2674 void ext4_block_bitmap_csum_set(struct super_block *sb, 2675 struct ext4_group_desc *gdp, 2676 struct buffer_head *bh); 2677 int ext4_block_bitmap_csum_verify(struct super_block *sb, 2678 struct ext4_group_desc *gdp, 2679 struct buffer_head *bh); 2680 2681 /* balloc.c */ 2682 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2683 ext4_fsblk_t blocknr, 2684 ext4_group_t *blockgrpp, 2685 ext4_grpblk_t *offsetp); 2686 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2687 ext4_fsblk_t block); 2688 2689 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2690 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2691 ext4_group_t group); 2692 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2693 ext4_fsblk_t goal, 2694 unsigned int flags, 2695 unsigned long *count, 2696 int *errp); 2697 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2698 s64 nclusters, unsigned int flags); 2699 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2700 extern void ext4_check_blocks_bitmap(struct super_block *); 2701 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2702 ext4_group_t block_group, 2703 struct buffer_head ** bh); 2704 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2705 ext4_group_t group); 2706 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2707 2708 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2709 ext4_group_t block_group, 2710 bool ignore_locked); 2711 extern int ext4_wait_block_bitmap(struct super_block *sb, 2712 ext4_group_t block_group, 2713 struct buffer_head *bh); 2714 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2715 ext4_group_t block_group); 2716 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2717 ext4_group_t block_group, 2718 struct ext4_group_desc *gdp); 2719 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2720 2721 #if IS_ENABLED(CONFIG_UNICODE) 2722 extern int ext4_fname_setup_ci_filename(struct inode *dir, 2723 const struct qstr *iname, 2724 struct ext4_filename *fname); 2725 #endif 2726 2727 /* ext4 encryption related stuff goes here crypto.c */ 2728 #ifdef CONFIG_FS_ENCRYPTION 2729 extern const struct fscrypt_operations ext4_cryptops; 2730 2731 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2732 int lookup, struct ext4_filename *fname); 2733 2734 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2735 struct ext4_filename *fname); 2736 2737 void ext4_fname_free_filename(struct ext4_filename *fname); 2738 2739 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2740 2741 #else /* !CONFIG_FS_ENCRYPTION */ 2742 static inline int ext4_fname_setup_filename(struct inode *dir, 2743 const struct qstr *iname, 2744 int lookup, 2745 struct ext4_filename *fname) 2746 { 2747 int err = 0; 2748 fname->usr_fname = iname; 2749 fname->disk_name.name = (unsigned char *) iname->name; 2750 fname->disk_name.len = iname->len; 2751 2752 #if IS_ENABLED(CONFIG_UNICODE) 2753 err = ext4_fname_setup_ci_filename(dir, iname, fname); 2754 #endif 2755 2756 return err; 2757 } 2758 2759 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2760 struct dentry *dentry, 2761 struct ext4_filename *fname) 2762 { 2763 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2764 } 2765 2766 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2767 { 2768 #if IS_ENABLED(CONFIG_UNICODE) 2769 kfree(fname->cf_name.name); 2770 fname->cf_name.name = NULL; 2771 #endif 2772 } 2773 2774 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2775 void __user *arg) 2776 { 2777 return -EOPNOTSUPP; 2778 } 2779 #endif /* !CONFIG_FS_ENCRYPTION */ 2780 2781 /* dir.c */ 2782 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2783 struct file *, 2784 struct ext4_dir_entry_2 *, 2785 struct buffer_head *, char *, int, 2786 unsigned int); 2787 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2788 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2789 (de), (bh), (buf), (size), (offset))) 2790 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2791 __u32 minor_hash, 2792 struct ext4_dir_entry_2 *dirent, 2793 struct fscrypt_str *ent_name); 2794 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2795 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2796 struct buffer_head *bh, 2797 void *buf, int buf_size, 2798 struct ext4_filename *fname, 2799 struct ext4_dir_entry_2 **dest_de); 2800 void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2801 struct ext4_dir_entry_2 *de, 2802 int buf_size, 2803 struct ext4_filename *fname); 2804 static inline void ext4_update_dx_flag(struct inode *inode) 2805 { 2806 if (!ext4_has_feature_dir_index(inode->i_sb) && 2807 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2808 /* ext4_iget() should have caught this... */ 2809 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2810 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2811 } 2812 } 2813 static const unsigned char ext4_filetype_table[] = { 2814 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2815 }; 2816 2817 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2818 { 2819 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2820 return DT_UNKNOWN; 2821 2822 return ext4_filetype_table[filetype]; 2823 } 2824 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2825 void *buf, int buf_size); 2826 2827 /* fsync.c */ 2828 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2829 2830 /* hash.c */ 2831 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2832 struct dx_hash_info *hinfo); 2833 2834 /* ialloc.c */ 2835 extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2836 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2837 struct inode *, umode_t, 2838 const struct qstr *qstr, __u32 goal, 2839 uid_t *owner, __u32 i_flags, 2840 int handle_type, unsigned int line_no, 2841 int nblocks); 2842 2843 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2844 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2845 (goal), (owner), i_flags, 0, 0, 0) 2846 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2847 type, nblocks) \ 2848 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2849 0, (type), __LINE__, (nblocks)) 2850 2851 2852 extern void ext4_free_inode(handle_t *, struct inode *); 2853 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2854 extern unsigned long ext4_count_free_inodes(struct super_block *); 2855 extern unsigned long ext4_count_dirs(struct super_block *); 2856 extern void ext4_check_inodes_bitmap(struct super_block *); 2857 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2858 extern int ext4_init_inode_table(struct super_block *sb, 2859 ext4_group_t group, int barrier); 2860 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2861 2862 /* fast_commit.c */ 2863 int ext4_fc_info_show(struct seq_file *seq, void *v); 2864 void ext4_fc_init(struct super_block *sb, journal_t *journal); 2865 void ext4_fc_init_inode(struct inode *inode); 2866 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2867 ext4_lblk_t end); 2868 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2869 struct dentry *dentry); 2870 void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2871 struct dentry *dentry); 2872 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2873 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2874 void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2875 struct dentry *dentry); 2876 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2877 void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2878 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2879 void ext4_fc_start_update(struct inode *inode); 2880 void ext4_fc_stop_update(struct inode *inode); 2881 void ext4_fc_del(struct inode *inode); 2882 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2883 void ext4_fc_replay_cleanup(struct super_block *sb); 2884 int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2885 int __init ext4_fc_init_dentry_cache(void); 2886 void ext4_fc_destroy_dentry_cache(void); 2887 int ext4_fc_record_regions(struct super_block *sb, int ino, 2888 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2889 int len, int replay); 2890 2891 /* mballoc.c */ 2892 extern const struct seq_operations ext4_mb_seq_groups_ops; 2893 extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2894 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2895 extern int ext4_mb_init(struct super_block *); 2896 extern int ext4_mb_release(struct super_block *); 2897 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2898 struct ext4_allocation_request *, int *); 2899 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2900 extern void ext4_discard_preallocations(struct inode *, unsigned int); 2901 extern int __init ext4_init_mballoc(void); 2902 extern void ext4_exit_mballoc(void); 2903 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2904 ext4_group_t group, 2905 unsigned int nr, int *cnt); 2906 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2907 unsigned int nr); 2908 2909 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2910 struct buffer_head *bh, ext4_fsblk_t block, 2911 unsigned long count, int flags); 2912 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2913 ext4_group_t ngroups); 2914 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2915 ext4_group_t i, struct ext4_group_desc *desc); 2916 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2917 ext4_fsblk_t block, unsigned long count); 2918 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2919 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2920 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 2921 int len, int state); 2922 2923 /* inode.c */ 2924 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 2925 struct ext4_inode_info *ei); 2926 int ext4_inode_is_fast_symlink(struct inode *inode); 2927 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2928 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2929 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2930 bool wait, struct buffer_head **bhs); 2931 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2932 struct buffer_head *bh_result, int create); 2933 int ext4_get_block(struct inode *inode, sector_t iblock, 2934 struct buffer_head *bh_result, int create); 2935 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2936 struct buffer_head *bh, int create); 2937 int ext4_walk_page_buffers(handle_t *handle, 2938 struct inode *inode, 2939 struct buffer_head *head, 2940 unsigned from, 2941 unsigned to, 2942 int *partial, 2943 int (*fn)(handle_t *handle, struct inode *inode, 2944 struct buffer_head *bh)); 2945 int do_journal_get_write_access(handle_t *handle, struct inode *inode, 2946 struct buffer_head *bh); 2947 #define FALL_BACK_TO_NONDELALLOC 1 2948 #define CONVERT_INLINE_DATA 2 2949 2950 typedef enum { 2951 EXT4_IGET_NORMAL = 0, 2952 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2953 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 2954 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 2955 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 2956 } ext4_iget_flags; 2957 2958 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2959 ext4_iget_flags flags, const char *function, 2960 unsigned int line); 2961 2962 #define ext4_iget(sb, ino, flags) \ 2963 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2964 2965 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2966 extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 2967 struct iattr *); 2968 extern u32 ext4_dio_alignment(struct inode *inode); 2969 extern int ext4_getattr(struct mnt_idmap *, const struct path *, 2970 struct kstat *, u32, unsigned int); 2971 extern void ext4_evict_inode(struct inode *); 2972 extern void ext4_clear_inode(struct inode *); 2973 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 2974 struct kstat *, u32, unsigned int); 2975 extern int ext4_sync_inode(handle_t *, struct inode *); 2976 extern void ext4_dirty_inode(struct inode *, int); 2977 extern int ext4_change_inode_journal_flag(struct inode *, int); 2978 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2979 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 2980 struct ext4_iloc *iloc); 2981 extern int ext4_inode_attach_jinode(struct inode *inode); 2982 extern int ext4_can_truncate(struct inode *inode); 2983 extern int ext4_truncate(struct inode *); 2984 extern int ext4_break_layouts(struct inode *); 2985 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 2986 extern void ext4_set_inode_flags(struct inode *, bool init); 2987 extern int ext4_alloc_da_blocks(struct inode *inode); 2988 extern void ext4_set_aops(struct inode *inode); 2989 extern int ext4_writepage_trans_blocks(struct inode *); 2990 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 2991 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2992 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2993 loff_t lstart, loff_t lend); 2994 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2995 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2996 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2997 extern void ext4_da_release_space(struct inode *inode, int to_free); 2998 extern void ext4_da_update_reserve_space(struct inode *inode, 2999 int used, int quota_claim); 3000 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3001 ext4_fsblk_t pblk, ext4_lblk_t len); 3002 3003 /* indirect.c */ 3004 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3005 struct ext4_map_blocks *map, int flags); 3006 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3007 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3008 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3009 ext4_lblk_t start, ext4_lblk_t end); 3010 3011 /* ioctl.c */ 3012 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3013 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3014 int ext4_fileattr_set(struct mnt_idmap *idmap, 3015 struct dentry *dentry, struct fileattr *fa); 3016 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3017 extern void ext4_reset_inode_seed(struct inode *inode); 3018 int ext4_update_overhead(struct super_block *sb, bool force); 3019 3020 /* migrate.c */ 3021 extern int ext4_ext_migrate(struct inode *); 3022 extern int ext4_ind_migrate(struct inode *inode); 3023 3024 /* namei.c */ 3025 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3026 struct inode *inode); 3027 extern int ext4_dirblock_csum_verify(struct inode *inode, 3028 struct buffer_head *bh); 3029 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3030 __u32 start_minor_hash, __u32 *next_hash); 3031 extern int ext4_search_dir(struct buffer_head *bh, 3032 char *search_buf, 3033 int buf_size, 3034 struct inode *dir, 3035 struct ext4_filename *fname, 3036 unsigned int offset, 3037 struct ext4_dir_entry_2 **res_dir); 3038 extern int ext4_generic_delete_entry(struct inode *dir, 3039 struct ext4_dir_entry_2 *de_del, 3040 struct buffer_head *bh, 3041 void *entry_buf, 3042 int buf_size, 3043 int csum_size); 3044 extern bool ext4_empty_dir(struct inode *inode); 3045 3046 /* resize.c */ 3047 extern void ext4_kvfree_array_rcu(void *to_free); 3048 extern int ext4_group_add(struct super_block *sb, 3049 struct ext4_new_group_data *input); 3050 extern int ext4_group_extend(struct super_block *sb, 3051 struct ext4_super_block *es, 3052 ext4_fsblk_t n_blocks_count); 3053 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3054 extern unsigned int ext4_list_backups(struct super_block *sb, 3055 unsigned int *three, unsigned int *five, 3056 unsigned int *seven); 3057 3058 /* super.c */ 3059 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3060 sector_t block, blk_opf_t op_flags); 3061 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3062 sector_t block); 3063 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3064 bh_end_io_t *end_io); 3065 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3066 bh_end_io_t *end_io); 3067 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3068 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3069 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3070 extern int ext4_calculate_overhead(struct super_block *sb); 3071 extern __le32 ext4_superblock_csum(struct super_block *sb, 3072 struct ext4_super_block *es); 3073 extern void ext4_superblock_csum_set(struct super_block *sb); 3074 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3075 ext4_group_t ngroup); 3076 extern const char *ext4_decode_error(struct super_block *sb, int errno, 3077 char nbuf[16]); 3078 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3079 ext4_group_t block_group, 3080 unsigned int flags); 3081 3082 extern __printf(7, 8) 3083 void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3084 int, __u64, const char *, ...); 3085 extern __printf(6, 7) 3086 void __ext4_error_inode(struct inode *, const char *, unsigned int, 3087 ext4_fsblk_t, int, const char *, ...); 3088 extern __printf(5, 6) 3089 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3090 const char *, ...); 3091 extern void __ext4_std_error(struct super_block *, const char *, 3092 unsigned int, int); 3093 extern __printf(4, 5) 3094 void __ext4_warning(struct super_block *, const char *, unsigned int, 3095 const char *, ...); 3096 extern __printf(4, 5) 3097 void __ext4_warning_inode(const struct inode *inode, const char *function, 3098 unsigned int line, const char *fmt, ...); 3099 extern __printf(3, 4) 3100 void __ext4_msg(struct super_block *, const char *, const char *, ...); 3101 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3102 const char *, unsigned int, const char *); 3103 extern __printf(7, 8) 3104 void __ext4_grp_locked_error(const char *, unsigned int, 3105 struct super_block *, ext4_group_t, 3106 unsigned long, ext4_fsblk_t, 3107 const char *, ...); 3108 3109 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 3110 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3111 3112 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3113 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3114 3115 #define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3116 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3117 (fmt), ## a) 3118 3119 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3120 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3121 3122 #define ext4_abort(sb, err, fmt, a...) \ 3123 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3124 3125 #ifdef CONFIG_PRINTK 3126 3127 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3128 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3129 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3130 __ext4_error_inode((inode), (func), (line), (block), \ 3131 (err), (fmt), ##__VA_ARGS__) 3132 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3133 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3134 #define ext4_error(sb, fmt, ...) \ 3135 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3136 ##__VA_ARGS__) 3137 #define ext4_error_err(sb, err, fmt, ...) \ 3138 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3139 ##__VA_ARGS__) 3140 #define ext4_warning(sb, fmt, ...) \ 3141 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3142 #define ext4_warning_inode(inode, fmt, ...) \ 3143 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3144 #define ext4_msg(sb, level, fmt, ...) \ 3145 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3146 #define dump_mmp_msg(sb, mmp, msg) \ 3147 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3148 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3149 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3150 fmt, ##__VA_ARGS__) 3151 3152 #else 3153 3154 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3155 do { \ 3156 no_printk(fmt, ##__VA_ARGS__); \ 3157 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3158 } while (0) 3159 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3160 do { \ 3161 no_printk(fmt, ##__VA_ARGS__); \ 3162 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3163 } while (0) 3164 #define ext4_error_file(file, func, line, block, fmt, ...) \ 3165 do { \ 3166 no_printk(fmt, ##__VA_ARGS__); \ 3167 __ext4_error_file(file, "", 0, block, " "); \ 3168 } while (0) 3169 #define ext4_error(sb, fmt, ...) \ 3170 do { \ 3171 no_printk(fmt, ##__VA_ARGS__); \ 3172 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3173 } while (0) 3174 #define ext4_error_err(sb, err, fmt, ...) \ 3175 do { \ 3176 no_printk(fmt, ##__VA_ARGS__); \ 3177 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3178 } while (0) 3179 #define ext4_warning(sb, fmt, ...) \ 3180 do { \ 3181 no_printk(fmt, ##__VA_ARGS__); \ 3182 __ext4_warning(sb, "", 0, " "); \ 3183 } while (0) 3184 #define ext4_warning_inode(inode, fmt, ...) \ 3185 do { \ 3186 no_printk(fmt, ##__VA_ARGS__); \ 3187 __ext4_warning_inode(inode, "", 0, " "); \ 3188 } while (0) 3189 #define ext4_msg(sb, level, fmt, ...) \ 3190 do { \ 3191 no_printk(fmt, ##__VA_ARGS__); \ 3192 __ext4_msg(sb, "", " "); \ 3193 } while (0) 3194 #define dump_mmp_msg(sb, mmp, msg) \ 3195 __dump_mmp_msg(sb, mmp, "", 0, "") 3196 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3197 do { \ 3198 no_printk(fmt, ##__VA_ARGS__); \ 3199 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3200 } while (0) 3201 3202 #endif 3203 3204 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3205 struct ext4_group_desc *bg); 3206 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3207 struct ext4_group_desc *bg); 3208 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3209 struct ext4_group_desc *bg); 3210 extern __u32 ext4_free_group_clusters(struct super_block *sb, 3211 struct ext4_group_desc *bg); 3212 extern __u32 ext4_free_inodes_count(struct super_block *sb, 3213 struct ext4_group_desc *bg); 3214 extern __u32 ext4_used_dirs_count(struct super_block *sb, 3215 struct ext4_group_desc *bg); 3216 extern __u32 ext4_itable_unused_count(struct super_block *sb, 3217 struct ext4_group_desc *bg); 3218 extern void ext4_block_bitmap_set(struct super_block *sb, 3219 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3220 extern void ext4_inode_bitmap_set(struct super_block *sb, 3221 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3222 extern void ext4_inode_table_set(struct super_block *sb, 3223 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3224 extern void ext4_free_group_clusters_set(struct super_block *sb, 3225 struct ext4_group_desc *bg, 3226 __u32 count); 3227 extern void ext4_free_inodes_set(struct super_block *sb, 3228 struct ext4_group_desc *bg, __u32 count); 3229 extern void ext4_used_dirs_set(struct super_block *sb, 3230 struct ext4_group_desc *bg, __u32 count); 3231 extern void ext4_itable_unused_set(struct super_block *sb, 3232 struct ext4_group_desc *bg, __u32 count); 3233 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3234 struct ext4_group_desc *gdp); 3235 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3236 struct ext4_group_desc *gdp); 3237 extern int ext4_register_li_request(struct super_block *sb, 3238 ext4_group_t first_not_zeroed); 3239 3240 static inline int ext4_has_metadata_csum(struct super_block *sb) 3241 { 3242 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 3243 !EXT4_SB(sb)->s_chksum_driver); 3244 3245 return ext4_has_feature_metadata_csum(sb) && 3246 (EXT4_SB(sb)->s_chksum_driver != NULL); 3247 } 3248 3249 static inline int ext4_has_group_desc_csum(struct super_block *sb) 3250 { 3251 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 3252 } 3253 3254 #define ext4_read_incompat_64bit_val(es, name) \ 3255 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3256 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3257 le32_to_cpu(es->name##_lo)) 3258 3259 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3260 { 3261 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3262 } 3263 3264 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3265 { 3266 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3267 } 3268 3269 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3270 { 3271 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3272 } 3273 3274 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3275 ext4_fsblk_t blk) 3276 { 3277 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3278 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3279 } 3280 3281 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3282 ext4_fsblk_t blk) 3283 { 3284 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3285 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3286 } 3287 3288 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3289 ext4_fsblk_t blk) 3290 { 3291 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3292 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3293 } 3294 3295 static inline loff_t ext4_isize(struct super_block *sb, 3296 struct ext4_inode *raw_inode) 3297 { 3298 if (ext4_has_feature_largedir(sb) || 3299 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3300 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3301 le32_to_cpu(raw_inode->i_size_lo); 3302 3303 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3304 } 3305 3306 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3307 { 3308 raw_inode->i_size_lo = cpu_to_le32(i_size); 3309 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3310 } 3311 3312 /* 3313 * Reading s_groups_count requires using smp_rmb() afterwards. See 3314 * the locking protocol documented in the comments of ext4_group_add() 3315 * in resize.c 3316 */ 3317 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3318 { 3319 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3320 3321 smp_rmb(); 3322 return ngroups; 3323 } 3324 3325 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3326 ext4_group_t block_group) 3327 { 3328 return block_group >> sbi->s_log_groups_per_flex; 3329 } 3330 3331 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3332 { 3333 return 1 << sbi->s_log_groups_per_flex; 3334 } 3335 3336 #define ext4_std_error(sb, errno) \ 3337 do { \ 3338 if ((errno)) \ 3339 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3340 } while (0) 3341 3342 #ifdef CONFIG_SMP 3343 /* Each CPU can accumulate percpu_counter_batch clusters in their local 3344 * counters. So we need to make sure we have free clusters more 3345 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3346 */ 3347 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3348 #else 3349 #define EXT4_FREECLUSTERS_WATERMARK 0 3350 #endif 3351 3352 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3353 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3354 { 3355 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3356 !inode_is_locked(inode)); 3357 down_write(&EXT4_I(inode)->i_data_sem); 3358 if (newsize > EXT4_I(inode)->i_disksize) 3359 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3360 up_write(&EXT4_I(inode)->i_data_sem); 3361 } 3362 3363 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3364 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3365 { 3366 int changed = 0; 3367 3368 if (newsize > inode->i_size) { 3369 i_size_write(inode, newsize); 3370 changed = 1; 3371 } 3372 if (newsize > EXT4_I(inode)->i_disksize) { 3373 ext4_update_i_disksize(inode, newsize); 3374 changed |= 2; 3375 } 3376 return changed; 3377 } 3378 3379 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3380 loff_t len); 3381 3382 struct ext4_group_info { 3383 unsigned long bb_state; 3384 #ifdef AGGRESSIVE_CHECK 3385 unsigned long bb_check_counter; 3386 #endif 3387 struct rb_root bb_free_root; 3388 ext4_grpblk_t bb_first_free; /* first free block */ 3389 ext4_grpblk_t bb_free; /* total free blocks */ 3390 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3391 int bb_avg_fragment_size_order; /* order of average 3392 fragment in BG */ 3393 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3394 ext4_group_t bb_group; /* Group number */ 3395 struct list_head bb_prealloc_list; 3396 #ifdef DOUBLE_CHECK 3397 void *bb_bitmap; 3398 #endif 3399 struct rw_semaphore alloc_sem; 3400 struct list_head bb_avg_fragment_size_node; 3401 struct list_head bb_largest_free_order_node; 3402 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3403 * regions, index is order. 3404 * bb_counters[3] = 5 means 3405 * 5 free 8-block regions. */ 3406 }; 3407 3408 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3409 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3410 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3411 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3412 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3413 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3414 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3415 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3416 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3417 3418 #define EXT4_MB_GRP_NEED_INIT(grp) \ 3419 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3420 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3421 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3422 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3423 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3424 3425 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3426 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3427 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3428 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3429 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3430 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3431 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3432 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3433 3434 #define EXT4_MAX_CONTENTION 8 3435 #define EXT4_CONTENTION_THRESHOLD 2 3436 3437 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3438 ext4_group_t group) 3439 { 3440 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3441 } 3442 3443 /* 3444 * Returns true if the filesystem is busy enough that attempts to 3445 * access the block group locks has run into contention. 3446 */ 3447 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3448 { 3449 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3450 } 3451 3452 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3453 { 3454 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3455 if (spin_trylock(lock)) 3456 /* 3457 * We're able to grab the lock right away, so drop the 3458 * lock contention counter. 3459 */ 3460 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3461 else { 3462 /* 3463 * The lock is busy, so bump the contention counter, 3464 * and then wait on the spin lock. 3465 */ 3466 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3467 EXT4_MAX_CONTENTION); 3468 spin_lock(lock); 3469 } 3470 } 3471 3472 static inline void ext4_unlock_group(struct super_block *sb, 3473 ext4_group_t group) 3474 { 3475 spin_unlock(ext4_group_lock_ptr(sb, group)); 3476 } 3477 3478 #ifdef CONFIG_QUOTA 3479 static inline bool ext4_quota_capable(struct super_block *sb) 3480 { 3481 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3482 } 3483 3484 static inline bool ext4_is_quota_journalled(struct super_block *sb) 3485 { 3486 struct ext4_sb_info *sbi = EXT4_SB(sb); 3487 3488 return (ext4_has_feature_quota(sb) || 3489 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3490 } 3491 int ext4_enable_quotas(struct super_block *sb); 3492 #endif 3493 3494 /* 3495 * Block validity checking 3496 */ 3497 #define ext4_check_indirect_blockref(inode, bh) \ 3498 ext4_check_blockref(__func__, __LINE__, inode, \ 3499 (__le32 *)(bh)->b_data, \ 3500 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3501 3502 #define ext4_ind_check_inode(inode) \ 3503 ext4_check_blockref(__func__, __LINE__, inode, \ 3504 EXT4_I(inode)->i_data, \ 3505 EXT4_NDIR_BLOCKS) 3506 3507 /* 3508 * Inodes and files operations 3509 */ 3510 3511 /* dir.c */ 3512 extern const struct file_operations ext4_dir_operations; 3513 3514 /* file.c */ 3515 extern const struct inode_operations ext4_file_inode_operations; 3516 extern const struct file_operations ext4_file_operations; 3517 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3518 3519 /* inline.c */ 3520 extern int ext4_get_max_inline_size(struct inode *inode); 3521 extern int ext4_find_inline_data_nolock(struct inode *inode); 3522 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3523 unsigned int len); 3524 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3525 3526 int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3527 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3528 struct inode *inode, 3529 loff_t pos, unsigned len, 3530 struct page **pagep); 3531 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3532 unsigned copied, struct folio *folio); 3533 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3534 struct inode *inode, 3535 loff_t pos, unsigned len, 3536 struct page **pagep, 3537 void **fsdata); 3538 extern int ext4_try_add_inline_entry(handle_t *handle, 3539 struct ext4_filename *fname, 3540 struct inode *dir, struct inode *inode); 3541 extern int ext4_try_create_inline_dir(handle_t *handle, 3542 struct inode *parent, 3543 struct inode *inode); 3544 extern int ext4_read_inline_dir(struct file *filp, 3545 struct dir_context *ctx, 3546 int *has_inline_data); 3547 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3548 struct inode *dir, ext4_lblk_t block, 3549 struct dx_hash_info *hinfo, 3550 __u32 start_hash, __u32 start_minor_hash, 3551 int *has_inline_data); 3552 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3553 struct ext4_filename *fname, 3554 struct ext4_dir_entry_2 **res_dir, 3555 int *has_inline_data); 3556 extern int ext4_delete_inline_entry(handle_t *handle, 3557 struct inode *dir, 3558 struct ext4_dir_entry_2 *de_del, 3559 struct buffer_head *bh, 3560 int *has_inline_data); 3561 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3562 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3563 struct ext4_dir_entry_2 **parent_de, 3564 int *retval); 3565 extern void *ext4_read_inline_link(struct inode *inode); 3566 3567 struct iomap; 3568 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3569 3570 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3571 3572 extern int ext4_convert_inline_data(struct inode *inode); 3573 3574 static inline int ext4_has_inline_data(struct inode *inode) 3575 { 3576 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3577 EXT4_I(inode)->i_inline_off; 3578 } 3579 3580 /* namei.c */ 3581 extern const struct inode_operations ext4_dir_inode_operations; 3582 extern const struct inode_operations ext4_special_inode_operations; 3583 extern struct dentry *ext4_get_parent(struct dentry *child); 3584 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3585 struct ext4_dir_entry_2 *de, 3586 int blocksize, int csum_size, 3587 unsigned int parent_ino, int dotdot_real_len); 3588 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3589 unsigned int blocksize); 3590 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3591 struct buffer_head *bh); 3592 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3593 struct inode *inode, struct dentry *dentry); 3594 extern int __ext4_link(struct inode *dir, struct inode *inode, 3595 struct dentry *dentry); 3596 3597 #define S_SHIFT 12 3598 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3599 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3600 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3601 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3602 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3603 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3604 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3605 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3606 }; 3607 3608 static inline void ext4_set_de_type(struct super_block *sb, 3609 struct ext4_dir_entry_2 *de, 3610 umode_t mode) { 3611 if (ext4_has_feature_filetype(sb)) 3612 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3613 } 3614 3615 /* readpages.c */ 3616 extern int ext4_mpage_readpages(struct inode *inode, 3617 struct readahead_control *rac, struct folio *folio); 3618 extern int __init ext4_init_post_read_processing(void); 3619 extern void ext4_exit_post_read_processing(void); 3620 3621 /* symlink.c */ 3622 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3623 extern const struct inode_operations ext4_symlink_inode_operations; 3624 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3625 3626 /* sysfs.c */ 3627 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3628 extern int ext4_register_sysfs(struct super_block *sb); 3629 extern void ext4_unregister_sysfs(struct super_block *sb); 3630 extern int __init ext4_init_sysfs(void); 3631 extern void ext4_exit_sysfs(void); 3632 3633 /* block_validity */ 3634 extern void ext4_release_system_zone(struct super_block *sb); 3635 extern int ext4_setup_system_zone(struct super_block *sb); 3636 extern int __init ext4_init_system_zone(void); 3637 extern void ext4_exit_system_zone(void); 3638 extern int ext4_inode_block_valid(struct inode *inode, 3639 ext4_fsblk_t start_blk, 3640 unsigned int count); 3641 extern int ext4_check_blockref(const char *, unsigned int, 3642 struct inode *, __le32 *, unsigned int); 3643 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3644 ext4_fsblk_t start_blk, unsigned int count); 3645 3646 3647 /* extents.c */ 3648 struct ext4_ext_path; 3649 struct ext4_extent; 3650 3651 /* 3652 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3653 * __le32. 3654 */ 3655 #define EXT_MAX_BLOCKS 0xffffffff 3656 3657 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3658 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3659 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3660 struct ext4_map_blocks *map, int flags); 3661 extern int ext4_ext_truncate(handle_t *, struct inode *); 3662 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3663 ext4_lblk_t end); 3664 extern void ext4_ext_init(struct super_block *); 3665 extern void ext4_ext_release(struct super_block *); 3666 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3667 loff_t len); 3668 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3669 loff_t offset, ssize_t len); 3670 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3671 ext4_io_end_t *io_end); 3672 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3673 struct ext4_map_blocks *map, int flags); 3674 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3675 int num, 3676 struct ext4_ext_path *path); 3677 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3678 struct ext4_ext_path **, 3679 struct ext4_extent *, int); 3680 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3681 struct ext4_ext_path **, 3682 int flags); 3683 extern void ext4_free_ext_path(struct ext4_ext_path *); 3684 extern int ext4_ext_check_inode(struct inode *inode); 3685 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3686 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3687 __u64 start, __u64 len); 3688 extern int ext4_get_es_cache(struct inode *inode, 3689 struct fiemap_extent_info *fieinfo, 3690 __u64 start, __u64 len); 3691 extern int ext4_ext_precache(struct inode *inode); 3692 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3693 struct inode *inode2, ext4_lblk_t lblk1, 3694 ext4_lblk_t lblk2, ext4_lblk_t count, 3695 int mark_unwritten,int *err); 3696 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3697 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3698 int check_cred, int restart_cred, 3699 int revoke_cred); 3700 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3701 extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3702 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3703 int len, int unwritten, ext4_fsblk_t pblk); 3704 extern int ext4_ext_clear_bb(struct inode *inode); 3705 3706 3707 /* move_extent.c */ 3708 extern void ext4_double_down_write_data_sem(struct inode *first, 3709 struct inode *second); 3710 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3711 struct inode *donor_inode); 3712 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3713 __u64 start_orig, __u64 start_donor, 3714 __u64 len, __u64 *moved_len); 3715 3716 /* page-io.c */ 3717 extern int __init ext4_init_pageio(void); 3718 extern void ext4_exit_pageio(void); 3719 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3720 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3721 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3722 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3723 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3724 struct writeback_control *wbc); 3725 extern void ext4_end_io_rsv_work(struct work_struct *work); 3726 extern void ext4_io_submit(struct ext4_io_submit *io); 3727 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3728 size_t len); 3729 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3730 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3731 3732 /* mmp.c */ 3733 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3734 3735 /* mmp.c */ 3736 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3737 3738 /* verity.c */ 3739 extern const struct fsverity_operations ext4_verityops; 3740 3741 /* orphan.c */ 3742 extern int ext4_orphan_add(handle_t *, struct inode *); 3743 extern int ext4_orphan_del(handle_t *, struct inode *); 3744 extern void ext4_orphan_cleanup(struct super_block *sb, 3745 struct ext4_super_block *es); 3746 extern void ext4_release_orphan_info(struct super_block *sb); 3747 extern int ext4_init_orphan_info(struct super_block *sb); 3748 extern int ext4_orphan_file_empty(struct super_block *sb); 3749 extern void ext4_orphan_file_block_trigger( 3750 struct jbd2_buffer_trigger_type *triggers, 3751 struct buffer_head *bh, 3752 void *data, size_t size); 3753 3754 /* 3755 * Add new method to test whether block and inode bitmaps are properly 3756 * initialized. With uninit_bg reading the block from disk is not enough 3757 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3758 */ 3759 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3760 3761 static inline int bitmap_uptodate(struct buffer_head *bh) 3762 { 3763 return (buffer_uptodate(bh) && 3764 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3765 } 3766 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3767 { 3768 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3769 } 3770 3771 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3772 3773 /* For ioend & aio unwritten conversion wait queues */ 3774 #define EXT4_WQ_HASH_SZ 37 3775 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3776 EXT4_WQ_HASH_SZ]) 3777 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3778 3779 extern int ext4_resize_begin(struct super_block *sb); 3780 extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3781 3782 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3783 struct ext4_io_end *io_end) 3784 { 3785 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3786 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3787 atomic_inc(&EXT4_I(inode)->i_unwritten); 3788 } 3789 } 3790 3791 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3792 { 3793 struct inode *inode = io_end->inode; 3794 3795 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3796 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3797 /* Wake up anyone waiting on unwritten extent conversion */ 3798 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3799 wake_up_all(ext4_ioend_wq(inode)); 3800 } 3801 } 3802 3803 extern const struct iomap_ops ext4_iomap_ops; 3804 extern const struct iomap_ops ext4_iomap_overwrite_ops; 3805 extern const struct iomap_ops ext4_iomap_report_ops; 3806 3807 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3808 { 3809 /* 3810 * If the buffer has the write error flag, we have failed 3811 * to write out data in the block. In this case, we don't 3812 * have to read the block because we may read the old data 3813 * successfully. 3814 */ 3815 if (buffer_write_io_error(bh)) 3816 set_buffer_uptodate(bh); 3817 return buffer_uptodate(bh); 3818 } 3819 3820 #endif /* __KERNEL__ */ 3821 3822 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3823 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3824 3825 #endif /* _EXT4_H */ 3826