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/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.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 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION) 44 #include <linux/fscrypt.h> 45 46 #include <linux/compiler.h> 47 48 /* 49 * The fourth extended filesystem constants/structures 50 */ 51 52 /* 53 * with AGGRESSIVE_CHECK allocator runs consistency checks over 54 * structures. these checks slow things down a lot 55 */ 56 #define AGGRESSIVE_CHECK__ 57 58 /* 59 * with DOUBLE_CHECK defined mballoc creates persistent in-core 60 * bitmaps, maintains and uses them to check for double allocations 61 */ 62 #define DOUBLE_CHECK__ 63 64 /* 65 * Define EXT4FS_DEBUG to produce debug messages 66 */ 67 #undef EXT4FS_DEBUG 68 69 /* 70 * Debug code 71 */ 72 #ifdef EXT4FS_DEBUG 73 #define ext4_debug(f, a...) \ 74 do { \ 75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 76 __FILE__, __LINE__, __func__); \ 77 printk(KERN_DEBUG f, ## a); \ 78 } while (0) 79 #else 80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 81 #endif 82 83 /* 84 * Turn on EXT_DEBUG to get lots of info about extents operations. 85 */ 86 #define EXT_DEBUG__ 87 #ifdef EXT_DEBUG 88 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 89 #else 90 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 91 #endif 92 93 /* data type for block offset of block group */ 94 typedef int ext4_grpblk_t; 95 96 /* data type for filesystem-wide blocks number */ 97 typedef unsigned long long ext4_fsblk_t; 98 99 /* data type for file logical block number */ 100 typedef __u32 ext4_lblk_t; 101 102 /* data type for block group number */ 103 typedef unsigned int ext4_group_t; 104 105 enum SHIFT_DIRECTION { 106 SHIFT_LEFT = 0, 107 SHIFT_RIGHT, 108 }; 109 110 /* 111 * Flags used in mballoc's allocation_context flags field. 112 * 113 * Also used to show what's going on for debugging purposes when the 114 * flag field is exported via the traceport interface 115 */ 116 117 /* prefer goal again. length */ 118 #define EXT4_MB_HINT_MERGE 0x0001 119 /* blocks already reserved */ 120 #define EXT4_MB_HINT_RESERVED 0x0002 121 /* metadata is being allocated */ 122 #define EXT4_MB_HINT_METADATA 0x0004 123 /* first blocks in the file */ 124 #define EXT4_MB_HINT_FIRST 0x0008 125 /* search for the best chunk */ 126 #define EXT4_MB_HINT_BEST 0x0010 127 /* data is being allocated */ 128 #define EXT4_MB_HINT_DATA 0x0020 129 /* don't preallocate (for tails) */ 130 #define EXT4_MB_HINT_NOPREALLOC 0x0040 131 /* allocate for locality group */ 132 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 133 /* allocate goal blocks or none */ 134 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 135 /* goal is meaningful */ 136 #define EXT4_MB_HINT_TRY_GOAL 0x0200 137 /* blocks already pre-reserved by delayed allocation */ 138 #define EXT4_MB_DELALLOC_RESERVED 0x0400 139 /* We are doing stream allocation */ 140 #define EXT4_MB_STREAM_ALLOC 0x0800 141 /* Use reserved root blocks if needed */ 142 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 143 /* Use blocks from reserved pool */ 144 #define EXT4_MB_USE_RESERVED 0x2000 145 146 struct ext4_allocation_request { 147 /* target inode for block we're allocating */ 148 struct inode *inode; 149 /* how many blocks we want to allocate */ 150 unsigned int len; 151 /* logical block in target inode */ 152 ext4_lblk_t logical; 153 /* the closest logical allocated block to the left */ 154 ext4_lblk_t lleft; 155 /* the closest logical allocated block to the right */ 156 ext4_lblk_t lright; 157 /* phys. target (a hint) */ 158 ext4_fsblk_t goal; 159 /* phys. block for the closest logical allocated block to the left */ 160 ext4_fsblk_t pleft; 161 /* phys. block for the closest logical allocated block to the right */ 162 ext4_fsblk_t pright; 163 /* flags. see above EXT4_MB_HINT_* */ 164 unsigned int flags; 165 }; 166 167 /* 168 * Logical to physical block mapping, used by ext4_map_blocks() 169 * 170 * This structure is used to pass requests into ext4_map_blocks() as 171 * well as to store the information returned by ext4_map_blocks(). It 172 * takes less room on the stack than a struct buffer_head. 173 */ 174 #define EXT4_MAP_NEW (1 << BH_New) 175 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 176 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 177 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 178 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 180 181 struct ext4_map_blocks { 182 ext4_fsblk_t m_pblk; 183 ext4_lblk_t m_lblk; 184 unsigned int m_len; 185 unsigned int m_flags; 186 }; 187 188 /* 189 * Flags for ext4_io_end->flags 190 */ 191 #define EXT4_IO_END_UNWRITTEN 0x0001 192 193 /* 194 * For converting unwritten extents on a work queue. 'handle' is used for 195 * buffered writeback. 196 */ 197 typedef struct ext4_io_end { 198 struct list_head list; /* per-file finished IO list */ 199 handle_t *handle; /* handle reserved for extent 200 * conversion */ 201 struct inode *inode; /* file being written to */ 202 struct bio *bio; /* Linked list of completed 203 * bios covering the extent */ 204 unsigned int flag; /* unwritten or not */ 205 atomic_t count; /* reference counter */ 206 loff_t offset; /* offset in the file */ 207 ssize_t size; /* size of the extent */ 208 } ext4_io_end_t; 209 210 struct ext4_io_submit { 211 struct writeback_control *io_wbc; 212 struct bio *io_bio; 213 ext4_io_end_t *io_end; 214 sector_t io_next_block; 215 }; 216 217 /* 218 * Special inodes numbers 219 */ 220 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 221 #define EXT4_ROOT_INO 2 /* Root inode */ 222 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 223 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 224 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 225 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 226 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 227 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 228 229 /* First non-reserved inode for old ext4 filesystems */ 230 #define EXT4_GOOD_OLD_FIRST_INO 11 231 232 /* 233 * Maximal count of links to a file 234 */ 235 #define EXT4_LINK_MAX 65000 236 237 /* 238 * Macro-instructions used to manage several block sizes 239 */ 240 #define EXT4_MIN_BLOCK_SIZE 1024 241 #define EXT4_MAX_BLOCK_SIZE 65536 242 #define EXT4_MIN_BLOCK_LOG_SIZE 10 243 #define EXT4_MAX_BLOCK_LOG_SIZE 16 244 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 245 #ifdef __KERNEL__ 246 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 247 #else 248 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 249 #endif 250 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 251 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 252 EXT4_SB(s)->s_cluster_bits) 253 #ifdef __KERNEL__ 254 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 255 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 256 #else 257 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 258 #endif 259 #ifdef __KERNEL__ 260 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 261 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 262 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 263 #else 264 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 265 EXT4_GOOD_OLD_INODE_SIZE : \ 266 (s)->s_inode_size) 267 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 268 EXT4_GOOD_OLD_FIRST_INO : \ 269 (s)->s_first_ino) 270 #endif 271 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 272 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 273 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 274 blkbits)) 275 276 /* Translate a block number to a cluster number */ 277 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 278 /* Translate a cluster number to a block number */ 279 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 280 /* Translate # of blks to # of clusters */ 281 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 282 (sbi)->s_cluster_bits) 283 /* Mask out the low bits to get the starting block of the cluster */ 284 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 285 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 286 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 287 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 288 /* Get the cluster offset */ 289 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 290 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 291 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 292 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 293 294 /* 295 * Structure of a blocks group descriptor 296 */ 297 struct ext4_group_desc 298 { 299 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 300 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 301 __le32 bg_inode_table_lo; /* Inodes table block */ 302 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 303 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 304 __le16 bg_used_dirs_count_lo; /* Directories count */ 305 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 306 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 307 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 308 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 309 __le16 bg_itable_unused_lo; /* Unused inodes count */ 310 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 311 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 312 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 313 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 314 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 315 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 316 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 317 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 318 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 319 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 320 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 321 __u32 bg_reserved; 322 }; 323 324 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 325 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 326 sizeof(__le16)) 327 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 328 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 329 sizeof(__le16)) 330 331 /* 332 * Structure of a flex block group info 333 */ 334 335 struct flex_groups { 336 atomic64_t free_clusters; 337 atomic_t free_inodes; 338 atomic_t used_dirs; 339 }; 340 341 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 342 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 343 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 344 345 /* 346 * Macro-instructions used to manage group descriptors 347 */ 348 #define EXT4_MIN_DESC_SIZE 32 349 #define EXT4_MIN_DESC_SIZE_64BIT 64 350 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 351 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 352 #ifdef __KERNEL__ 353 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 354 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 356 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 357 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 358 #else 359 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 361 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 362 #endif 363 364 /* 365 * Constants relative to the data blocks 366 */ 367 #define EXT4_NDIR_BLOCKS 12 368 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 369 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 370 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 371 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 372 373 /* 374 * Inode flags 375 */ 376 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 377 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 378 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 379 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 380 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 381 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 382 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 383 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 384 /* Reserved for compression usage... */ 385 #define EXT4_DIRTY_FL 0x00000100 386 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 387 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 388 /* nb: was previously EXT2_ECOMPR_FL */ 389 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 390 /* End compression flags --- maybe not all used */ 391 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 392 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 393 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 394 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 395 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 396 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 397 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 398 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 399 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 400 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 401 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 402 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 404 405 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 406 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 407 408 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 409 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 410 EXT4_IMMUTABLE_FL | \ 411 EXT4_APPEND_FL | \ 412 EXT4_NODUMP_FL | \ 413 EXT4_NOATIME_FL | \ 414 EXT4_PROJINHERIT_FL) 415 416 /* Flags that should be inherited by new inodes from their parent. */ 417 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 418 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 419 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 420 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 421 EXT4_PROJINHERIT_FL) 422 423 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 424 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 425 426 /* Flags that are appropriate for non-directories/regular files. */ 427 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 428 429 /* The only flags that should be swapped */ 430 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 431 432 /* Mask out flags that are inappropriate for the given type of inode. */ 433 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 434 { 435 if (S_ISDIR(mode)) 436 return flags; 437 else if (S_ISREG(mode)) 438 return flags & EXT4_REG_FLMASK; 439 else 440 return flags & EXT4_OTHER_FLMASK; 441 } 442 443 /* 444 * Inode flags used for atomic set/get 445 */ 446 enum { 447 EXT4_INODE_SECRM = 0, /* Secure deletion */ 448 EXT4_INODE_UNRM = 1, /* Undelete */ 449 EXT4_INODE_COMPR = 2, /* Compress file */ 450 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 451 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 452 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 453 EXT4_INODE_NODUMP = 6, /* do not dump file */ 454 EXT4_INODE_NOATIME = 7, /* do not update atime */ 455 /* Reserved for compression usage... */ 456 EXT4_INODE_DIRTY = 8, 457 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 458 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 459 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 460 /* End compression flags --- maybe not all used */ 461 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 462 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 463 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 464 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 465 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 466 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 467 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 468 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 469 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 470 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 471 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 472 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 473 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 474 }; 475 476 /* 477 * Since it's pretty easy to mix up bit numbers and hex values, we use a 478 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 479 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 480 * any extra space in the compiled kernel image, otherwise, the build will fail. 481 * It's important that these values are the same, since we are using 482 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 483 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 484 * values found in ext2, ext3 and ext4 filesystems, and of course the values 485 * defined in e2fsprogs. 486 * 487 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 488 */ 489 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 490 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 491 492 static inline void ext4_check_flag_values(void) 493 { 494 CHECK_FLAG_VALUE(SECRM); 495 CHECK_FLAG_VALUE(UNRM); 496 CHECK_FLAG_VALUE(COMPR); 497 CHECK_FLAG_VALUE(SYNC); 498 CHECK_FLAG_VALUE(IMMUTABLE); 499 CHECK_FLAG_VALUE(APPEND); 500 CHECK_FLAG_VALUE(NODUMP); 501 CHECK_FLAG_VALUE(NOATIME); 502 CHECK_FLAG_VALUE(DIRTY); 503 CHECK_FLAG_VALUE(COMPRBLK); 504 CHECK_FLAG_VALUE(NOCOMPR); 505 CHECK_FLAG_VALUE(ENCRYPT); 506 CHECK_FLAG_VALUE(INDEX); 507 CHECK_FLAG_VALUE(IMAGIC); 508 CHECK_FLAG_VALUE(JOURNAL_DATA); 509 CHECK_FLAG_VALUE(NOTAIL); 510 CHECK_FLAG_VALUE(DIRSYNC); 511 CHECK_FLAG_VALUE(TOPDIR); 512 CHECK_FLAG_VALUE(HUGE_FILE); 513 CHECK_FLAG_VALUE(EXTENTS); 514 CHECK_FLAG_VALUE(EA_INODE); 515 CHECK_FLAG_VALUE(EOFBLOCKS); 516 CHECK_FLAG_VALUE(INLINE_DATA); 517 CHECK_FLAG_VALUE(PROJINHERIT); 518 CHECK_FLAG_VALUE(RESERVED); 519 } 520 521 /* Used to pass group descriptor data when online resize is done */ 522 struct ext4_new_group_input { 523 __u32 group; /* Group number for this data */ 524 __u64 block_bitmap; /* Absolute block number of block bitmap */ 525 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 526 __u64 inode_table; /* Absolute block number of inode table start */ 527 __u32 blocks_count; /* Total number of blocks in this group */ 528 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 529 __u16 unused; 530 }; 531 532 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 533 struct compat_ext4_new_group_input { 534 u32 group; 535 compat_u64 block_bitmap; 536 compat_u64 inode_bitmap; 537 compat_u64 inode_table; 538 u32 blocks_count; 539 u16 reserved_blocks; 540 u16 unused; 541 }; 542 #endif 543 544 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 545 struct ext4_new_group_data { 546 __u32 group; 547 __u64 block_bitmap; 548 __u64 inode_bitmap; 549 __u64 inode_table; 550 __u32 blocks_count; 551 __u16 reserved_blocks; 552 __u16 mdata_blocks; 553 __u32 free_clusters_count; 554 }; 555 556 /* Indexes used to index group tables in ext4_new_group_data */ 557 enum { 558 BLOCK_BITMAP = 0, /* block bitmap */ 559 INODE_BITMAP, /* inode bitmap */ 560 INODE_TABLE, /* inode tables */ 561 GROUP_TABLE_COUNT, 562 }; 563 564 /* 565 * Flags used by ext4_map_blocks() 566 */ 567 /* Allocate any needed blocks and/or convert an unwritten 568 extent to be an initialized ext4 */ 569 #define EXT4_GET_BLOCKS_CREATE 0x0001 570 /* Request the creation of an unwritten extent */ 571 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 572 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 573 EXT4_GET_BLOCKS_CREATE) 574 /* Caller is from the delayed allocation writeout path 575 * finally doing the actual allocation of delayed blocks */ 576 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 577 /* caller is from the direct IO path, request to creation of an 578 unwritten extents if not allocated, split the unwritten 579 extent if blocks has been preallocated already*/ 580 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 581 #define EXT4_GET_BLOCKS_CONVERT 0x0010 582 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 583 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 584 /* Convert extent to initialized after IO complete */ 585 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 586 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 587 /* Eventual metadata allocation (due to growing extent tree) 588 * should not fail, so try to use reserved blocks for that.*/ 589 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 590 /* Don't normalize allocation size (used for fallocate) */ 591 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 592 /* Request will not result in inode size update (user for fallocate) */ 593 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 594 /* Convert written extents to unwritten */ 595 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 596 /* Write zeros to newly created written extents */ 597 #define EXT4_GET_BLOCKS_ZERO 0x0200 598 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 599 EXT4_GET_BLOCKS_ZERO) 600 /* Caller will submit data before dropping transaction handle. This 601 * allows jbd2 to avoid submitting data before commit. */ 602 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 603 604 /* 605 * The bit position of these flags must not overlap with any of the 606 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 607 * read_extent_tree_block(), ext4_split_extent_at(), 608 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 609 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 610 * caching the extents when reading from the extent tree while a 611 * truncate or punch hole operation is in progress. 612 */ 613 #define EXT4_EX_NOCACHE 0x40000000 614 #define EXT4_EX_FORCE_CACHE 0x20000000 615 616 /* 617 * Flags used by ext4_free_blocks 618 */ 619 #define EXT4_FREE_BLOCKS_METADATA 0x0001 620 #define EXT4_FREE_BLOCKS_FORGET 0x0002 621 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 622 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 623 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 624 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 625 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 626 627 /* 628 * ioctl commands 629 */ 630 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 631 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 632 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 633 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 634 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 635 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 636 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 637 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 638 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 639 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 640 #define EXT4_IOC_MIGRATE _IO('f', 9) 641 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 642 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 643 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 644 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 645 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 646 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 647 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 648 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 649 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 650 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 651 652 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 653 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 654 655 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 656 657 /* 658 * Flags for going down operation 659 */ 660 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 661 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 662 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 663 664 665 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 666 /* 667 * ioctl commands in 32 bit emulation 668 */ 669 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 670 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 671 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 672 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 673 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 674 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 675 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 676 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 677 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 678 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 679 #endif 680 681 /* Max physical block we can address w/o extents */ 682 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 683 684 /* Max logical block we can support */ 685 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF 686 687 /* 688 * Structure of an inode on the disk 689 */ 690 struct ext4_inode { 691 __le16 i_mode; /* File mode */ 692 __le16 i_uid; /* Low 16 bits of Owner Uid */ 693 __le32 i_size_lo; /* Size in bytes */ 694 __le32 i_atime; /* Access time */ 695 __le32 i_ctime; /* Inode Change time */ 696 __le32 i_mtime; /* Modification time */ 697 __le32 i_dtime; /* Deletion Time */ 698 __le16 i_gid; /* Low 16 bits of Group Id */ 699 __le16 i_links_count; /* Links count */ 700 __le32 i_blocks_lo; /* Blocks count */ 701 __le32 i_flags; /* File flags */ 702 union { 703 struct { 704 __le32 l_i_version; 705 } linux1; 706 struct { 707 __u32 h_i_translator; 708 } hurd1; 709 struct { 710 __u32 m_i_reserved1; 711 } masix1; 712 } osd1; /* OS dependent 1 */ 713 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 714 __le32 i_generation; /* File version (for NFS) */ 715 __le32 i_file_acl_lo; /* File ACL */ 716 __le32 i_size_high; 717 __le32 i_obso_faddr; /* Obsoleted fragment address */ 718 union { 719 struct { 720 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 721 __le16 l_i_file_acl_high; 722 __le16 l_i_uid_high; /* these 2 fields */ 723 __le16 l_i_gid_high; /* were reserved2[0] */ 724 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 725 __le16 l_i_reserved; 726 } linux2; 727 struct { 728 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 729 __u16 h_i_mode_high; 730 __u16 h_i_uid_high; 731 __u16 h_i_gid_high; 732 __u32 h_i_author; 733 } hurd2; 734 struct { 735 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 736 __le16 m_i_file_acl_high; 737 __u32 m_i_reserved2[2]; 738 } masix2; 739 } osd2; /* OS dependent 2 */ 740 __le16 i_extra_isize; 741 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 742 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 743 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 744 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 745 __le32 i_crtime; /* File Creation time */ 746 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 747 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 748 __le32 i_projid; /* Project ID */ 749 }; 750 751 struct move_extent { 752 __u32 reserved; /* should be zero */ 753 __u32 donor_fd; /* donor file descriptor */ 754 __u64 orig_start; /* logical start offset in block for orig */ 755 __u64 donor_start; /* logical start offset in block for donor */ 756 __u64 len; /* block length to be moved */ 757 __u64 moved_len; /* moved block length */ 758 }; 759 760 #define EXT4_EPOCH_BITS 2 761 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 762 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 763 764 /* 765 * Extended fields will fit into an inode if the filesystem was formatted 766 * with large inodes (-I 256 or larger) and there are not currently any EAs 767 * consuming all of the available space. For new inodes we always reserve 768 * enough space for the kernel's known extended fields, but for inodes 769 * created with an old kernel this might not have been the case. None of 770 * the extended inode fields is critical for correct filesystem operation. 771 * This macro checks if a certain field fits in the inode. Note that 772 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 773 */ 774 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 775 ((offsetof(typeof(*ext4_inode), field) + \ 776 sizeof((ext4_inode)->field)) \ 777 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 778 (einode)->i_extra_isize)) \ 779 780 /* 781 * We use an encoding that preserves the times for extra epoch "00": 782 * 783 * extra msb of adjust for signed 784 * epoch 32-bit 32-bit tv_sec to 785 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 786 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 787 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 788 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 789 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 790 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 791 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 792 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 793 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 794 * 795 * Note that previous versions of the kernel on 64-bit systems would 796 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 797 * 1970. e2fsck will correct this, assuming that it is run on the 798 * affected filesystem before 2242. 799 */ 800 801 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 802 { 803 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 804 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 805 } 806 807 static inline void ext4_decode_extra_time(struct timespec64 *time, 808 __le32 extra) 809 { 810 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 811 812 #if 1 813 /* Handle legacy encoding of pre-1970 dates with epoch 814 * bits 1,1. (This backwards compatibility may be removed 815 * at the discretion of the ext4 developers.) 816 */ 817 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 818 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 819 extra_bits = 0; 820 time->tv_sec += extra_bits << 32; 821 #else 822 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 823 #endif 824 } 825 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 826 } 827 828 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 829 do { \ 830 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 831 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 832 (raw_inode)->xtime ## _extra = \ 833 ext4_encode_extra_time(&(inode)->xtime); \ 834 } \ 835 } while (0) 836 837 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 838 do { \ 839 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 840 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 841 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 842 (raw_inode)->xtime ## _extra = \ 843 ext4_encode_extra_time(&(einode)->xtime); \ 844 } while (0) 845 846 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 847 do { \ 848 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 849 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 850 ext4_decode_extra_time(&(inode)->xtime, \ 851 raw_inode->xtime ## _extra); \ 852 } \ 853 else \ 854 (inode)->xtime.tv_nsec = 0; \ 855 } while (0) 856 857 858 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 859 do { \ 860 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 861 (einode)->xtime.tv_sec = \ 862 (signed)le32_to_cpu((raw_inode)->xtime); \ 863 else \ 864 (einode)->xtime.tv_sec = 0; \ 865 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 866 ext4_decode_extra_time(&(einode)->xtime, \ 867 raw_inode->xtime ## _extra); \ 868 else \ 869 (einode)->xtime.tv_nsec = 0; \ 870 } while (0) 871 872 #define i_disk_version osd1.linux1.l_i_version 873 874 #if defined(__KERNEL__) || defined(__linux__) 875 #define i_reserved1 osd1.linux1.l_i_reserved1 876 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 877 #define i_blocks_high osd2.linux2.l_i_blocks_high 878 #define i_uid_low i_uid 879 #define i_gid_low i_gid 880 #define i_uid_high osd2.linux2.l_i_uid_high 881 #define i_gid_high osd2.linux2.l_i_gid_high 882 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 883 884 #elif defined(__GNU__) 885 886 #define i_translator osd1.hurd1.h_i_translator 887 #define i_uid_high osd2.hurd2.h_i_uid_high 888 #define i_gid_high osd2.hurd2.h_i_gid_high 889 #define i_author osd2.hurd2.h_i_author 890 891 #elif defined(__masix__) 892 893 #define i_reserved1 osd1.masix1.m_i_reserved1 894 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 895 #define i_reserved2 osd2.masix2.m_i_reserved2 896 897 #endif /* defined(__KERNEL__) || defined(__linux__) */ 898 899 #include "extents_status.h" 900 901 /* 902 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 903 * 904 * These are needed to avoid lockdep false positives when we need to 905 * allocate blocks to the quota inode during ext4_map_blocks(), while 906 * holding i_data_sem for a normal (non-quota) inode. Since we don't 907 * do quota tracking for the quota inode, this avoids deadlock (as 908 * well as infinite recursion, since it isn't turtles all the way 909 * down...) 910 * 911 * I_DATA_SEM_NORMAL - Used for most inodes 912 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 913 * where the second inode has larger inode number 914 * than the first 915 * I_DATA_SEM_QUOTA - Used for quota inodes only 916 */ 917 enum { 918 I_DATA_SEM_NORMAL = 0, 919 I_DATA_SEM_OTHER, 920 I_DATA_SEM_QUOTA, 921 }; 922 923 924 /* 925 * fourth extended file system inode data in memory 926 */ 927 struct ext4_inode_info { 928 __le32 i_data[15]; /* unconverted */ 929 __u32 i_dtime; 930 ext4_fsblk_t i_file_acl; 931 932 /* 933 * i_block_group is the number of the block group which contains 934 * this file's inode. Constant across the lifetime of the inode, 935 * it is used for making block allocation decisions - we try to 936 * place a file's data blocks near its inode block, and new inodes 937 * near to their parent directory's inode. 938 */ 939 ext4_group_t i_block_group; 940 ext4_lblk_t i_dir_start_lookup; 941 #if (BITS_PER_LONG < 64) 942 unsigned long i_state_flags; /* Dynamic state flags */ 943 #endif 944 unsigned long i_flags; 945 946 /* 947 * Extended attributes can be read independently of the main file 948 * data. Taking i_mutex even when reading would cause contention 949 * between readers of EAs and writers of regular file data, so 950 * instead we synchronize on xattr_sem when reading or changing 951 * EAs. 952 */ 953 struct rw_semaphore xattr_sem; 954 955 struct list_head i_orphan; /* unlinked but open inodes */ 956 957 /* 958 * i_disksize keeps track of what the inode size is ON DISK, not 959 * in memory. During truncate, i_size is set to the new size by 960 * the VFS prior to calling ext4_truncate(), but the filesystem won't 961 * set i_disksize to 0 until the truncate is actually under way. 962 * 963 * The intent is that i_disksize always represents the blocks which 964 * are used by this file. This allows recovery to restart truncate 965 * on orphans if we crash during truncate. We actually write i_disksize 966 * into the on-disk inode when writing inodes out, instead of i_size. 967 * 968 * The only time when i_disksize and i_size may be different is when 969 * a truncate is in progress. The only things which change i_disksize 970 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 971 */ 972 loff_t i_disksize; 973 974 /* 975 * i_data_sem is for serialising ext4_truncate() against 976 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 977 * data tree are chopped off during truncate. We can't do that in 978 * ext4 because whenever we perform intermediate commits during 979 * truncate, the inode and all the metadata blocks *must* be in a 980 * consistent state which allows truncation of the orphans to restart 981 * during recovery. Hence we must fix the get_block-vs-truncate race 982 * by other means, so we have i_data_sem. 983 */ 984 struct rw_semaphore i_data_sem; 985 /* 986 * i_mmap_sem is for serializing page faults with truncate / punch hole 987 * operations. We have to make sure that new page cannot be faulted in 988 * a section of the inode that is being punched. We cannot easily use 989 * i_data_sem for this since we need protection for the whole punch 990 * operation and i_data_sem ranks below transaction start so we have 991 * to occasionally drop it. 992 */ 993 struct rw_semaphore i_mmap_sem; 994 struct inode vfs_inode; 995 struct jbd2_inode *jinode; 996 997 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 998 999 /* 1000 * File creation time. Its function is same as that of 1001 * struct timespec64 i_{a,c,m}time in the generic inode. 1002 */ 1003 struct timespec64 i_crtime; 1004 1005 /* mballoc */ 1006 struct list_head i_prealloc_list; 1007 spinlock_t i_prealloc_lock; 1008 1009 /* extents status tree */ 1010 struct ext4_es_tree i_es_tree; 1011 rwlock_t i_es_lock; 1012 struct list_head i_es_list; 1013 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1014 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1015 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1016 extents to shrink. Protected by 1017 i_es_lock */ 1018 1019 /* ialloc */ 1020 ext4_group_t i_last_alloc_group; 1021 1022 /* allocation reservation info for delalloc */ 1023 /* In case of bigalloc, this refer to clusters rather than blocks */ 1024 unsigned int i_reserved_data_blocks; 1025 ext4_lblk_t i_da_metadata_calc_last_lblock; 1026 int i_da_metadata_calc_len; 1027 1028 /* pending cluster reservations for bigalloc file systems */ 1029 struct ext4_pending_tree i_pending_tree; 1030 1031 /* on-disk additional length */ 1032 __u16 i_extra_isize; 1033 1034 /* Indicate the inline data space. */ 1035 u16 i_inline_off; 1036 u16 i_inline_size; 1037 1038 #ifdef CONFIG_QUOTA 1039 /* quota space reservation, managed internally by quota code */ 1040 qsize_t i_reserved_quota; 1041 #endif 1042 1043 /* Lock protecting lists below */ 1044 spinlock_t i_completed_io_lock; 1045 /* 1046 * Completed IOs that need unwritten extents handling and have 1047 * transaction reserved 1048 */ 1049 struct list_head i_rsv_conversion_list; 1050 struct work_struct i_rsv_conversion_work; 1051 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1052 1053 spinlock_t i_block_reservation_lock; 1054 1055 /* 1056 * Transactions that contain inode's metadata needed to complete 1057 * fsync and fdatasync, respectively. 1058 */ 1059 tid_t i_sync_tid; 1060 tid_t i_datasync_tid; 1061 1062 #ifdef CONFIG_QUOTA 1063 struct dquot *i_dquot[MAXQUOTAS]; 1064 #endif 1065 1066 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1067 __u32 i_csum_seed; 1068 1069 kprojid_t i_projid; 1070 }; 1071 1072 /* 1073 * File system states 1074 */ 1075 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1076 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1077 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1078 1079 /* 1080 * Misc. filesystem flags 1081 */ 1082 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1083 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1084 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1085 1086 /* 1087 * Mount flags set via mount options or defaults 1088 */ 1089 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1090 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1091 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1092 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1093 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1094 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1095 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1096 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1097 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1098 #ifdef CONFIG_FS_DAX 1099 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1100 #else 1101 #define EXT4_MOUNT_DAX 0 1102 #endif 1103 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1104 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1105 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1106 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1107 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1108 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1109 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1110 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1111 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1112 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1113 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1114 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1115 * enable enforcement for hidden 1116 * quota files */ 1117 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1118 * enforcement for hidden quota 1119 * files */ 1120 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1121 * enforcement */ 1122 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1123 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1124 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1125 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1126 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1127 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1128 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1129 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1130 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1131 1132 /* 1133 * Mount flags set either automatically (could not be set by mount option) 1134 * based on per file system feature or property or in special cases such as 1135 * distinguishing between explicit mount option definition and default. 1136 */ 1137 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1138 specified delalloc */ 1139 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1140 size of blocksize * 8 1141 blocks */ 1142 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1143 file systems */ 1144 1145 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1146 specified journal checksum */ 1147 1148 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1149 ~EXT4_MOUNT_##opt 1150 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1151 EXT4_MOUNT_##opt 1152 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1153 EXT4_MOUNT_##opt) 1154 1155 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1156 ~EXT4_MOUNT2_##opt 1157 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1158 EXT4_MOUNT2_##opt 1159 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1160 EXT4_MOUNT2_##opt) 1161 1162 #define ext4_test_and_set_bit __test_and_set_bit_le 1163 #define ext4_set_bit __set_bit_le 1164 #define ext4_set_bit_atomic ext2_set_bit_atomic 1165 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1166 #define ext4_clear_bit __clear_bit_le 1167 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1168 #define ext4_test_bit test_bit_le 1169 #define ext4_find_next_zero_bit find_next_zero_bit_le 1170 #define ext4_find_next_bit find_next_bit_le 1171 1172 extern void ext4_set_bits(void *bm, int cur, int len); 1173 1174 /* 1175 * Maximal mount counts between two filesystem checks 1176 */ 1177 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1178 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1179 1180 /* 1181 * Behaviour when detecting errors 1182 */ 1183 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1184 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1185 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1186 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1187 1188 /* Metadata checksum algorithm codes */ 1189 #define EXT4_CRC32C_CHKSUM 1 1190 1191 /* 1192 * Structure of the super block 1193 */ 1194 struct ext4_super_block { 1195 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1196 __le32 s_blocks_count_lo; /* Blocks count */ 1197 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1198 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1199 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1200 __le32 s_first_data_block; /* First Data Block */ 1201 __le32 s_log_block_size; /* Block size */ 1202 __le32 s_log_cluster_size; /* Allocation cluster size */ 1203 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1204 __le32 s_clusters_per_group; /* # Clusters per group */ 1205 __le32 s_inodes_per_group; /* # Inodes per group */ 1206 __le32 s_mtime; /* Mount time */ 1207 /*30*/ __le32 s_wtime; /* Write time */ 1208 __le16 s_mnt_count; /* Mount count */ 1209 __le16 s_max_mnt_count; /* Maximal mount count */ 1210 __le16 s_magic; /* Magic signature */ 1211 __le16 s_state; /* File system state */ 1212 __le16 s_errors; /* Behaviour when detecting errors */ 1213 __le16 s_minor_rev_level; /* minor revision level */ 1214 /*40*/ __le32 s_lastcheck; /* time of last check */ 1215 __le32 s_checkinterval; /* max. time between checks */ 1216 __le32 s_creator_os; /* OS */ 1217 __le32 s_rev_level; /* Revision level */ 1218 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1219 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1220 /* 1221 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1222 * 1223 * Note: the difference between the compatible feature set and 1224 * the incompatible feature set is that if there is a bit set 1225 * in the incompatible feature set that the kernel doesn't 1226 * know about, it should refuse to mount the filesystem. 1227 * 1228 * e2fsck's requirements are more strict; if it doesn't know 1229 * about a feature in either the compatible or incompatible 1230 * feature set, it must abort and not try to meddle with 1231 * things it doesn't understand... 1232 */ 1233 __le32 s_first_ino; /* First non-reserved inode */ 1234 __le16 s_inode_size; /* size of inode structure */ 1235 __le16 s_block_group_nr; /* block group # of this superblock */ 1236 __le32 s_feature_compat; /* compatible feature set */ 1237 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1238 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1239 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1240 /*78*/ char s_volume_name[16]; /* volume name */ 1241 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1242 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1243 /* 1244 * Performance hints. Directory preallocation should only 1245 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1246 */ 1247 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1248 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1249 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1250 /* 1251 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1252 */ 1253 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1254 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1255 __le32 s_journal_dev; /* device number of journal file */ 1256 __le32 s_last_orphan; /* start of list of inodes to delete */ 1257 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1258 __u8 s_def_hash_version; /* Default hash version to use */ 1259 __u8 s_jnl_backup_type; 1260 __le16 s_desc_size; /* size of group descriptor */ 1261 /*100*/ __le32 s_default_mount_opts; 1262 __le32 s_first_meta_bg; /* First metablock block group */ 1263 __le32 s_mkfs_time; /* When the filesystem was created */ 1264 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1265 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1266 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1267 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1268 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1269 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1270 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1271 __le32 s_flags; /* Miscellaneous flags */ 1272 __le16 s_raid_stride; /* RAID stride */ 1273 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1274 __le64 s_mmp_block; /* Block for multi-mount protection */ 1275 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1276 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1277 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1278 __u8 s_encryption_level; /* versioning level for encryption */ 1279 __u8 s_reserved_pad; /* Padding to next 32bits */ 1280 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1281 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1282 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1283 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1284 snapshot's future use */ 1285 __le32 s_snapshot_list; /* inode number of the head of the 1286 on-disk snapshot list */ 1287 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1288 __le32 s_error_count; /* number of fs errors */ 1289 __le32 s_first_error_time; /* first time an error happened */ 1290 __le32 s_first_error_ino; /* inode involved in first error */ 1291 __le64 s_first_error_block; /* block involved of first error */ 1292 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1293 __le32 s_first_error_line; /* line number where error happened */ 1294 __le32 s_last_error_time; /* most recent time of an error */ 1295 __le32 s_last_error_ino; /* inode involved in last error */ 1296 __le32 s_last_error_line; /* line number where error happened */ 1297 __le64 s_last_error_block; /* block involved of last error */ 1298 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1299 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1300 __u8 s_mount_opts[64]; 1301 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1302 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1303 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1304 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1305 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1306 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1307 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1308 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1309 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1310 __u8 s_wtime_hi; 1311 __u8 s_mtime_hi; 1312 __u8 s_mkfs_time_hi; 1313 __u8 s_lastcheck_hi; 1314 __u8 s_first_error_time_hi; 1315 __u8 s_last_error_time_hi; 1316 __u8 s_pad[2]; 1317 __le32 s_reserved[96]; /* Padding to the end of the block */ 1318 __le32 s_checksum; /* crc32c(superblock) */ 1319 }; 1320 1321 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1322 1323 #ifdef __KERNEL__ 1324 1325 /* 1326 * run-time mount flags 1327 */ 1328 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1329 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1330 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1331 1332 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1333 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1334 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1335 #else 1336 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1337 #endif 1338 1339 /* Number of quota types we support */ 1340 #define EXT4_MAXQUOTAS 3 1341 1342 /* 1343 * fourth extended-fs super-block data in memory 1344 */ 1345 struct ext4_sb_info { 1346 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1347 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1348 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1349 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1350 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1351 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1352 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1353 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1354 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1355 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1356 unsigned long s_overhead; /* # of fs overhead clusters */ 1357 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1358 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1359 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1360 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1361 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1362 struct buffer_head **s_group_desc; 1363 unsigned int s_mount_opt; 1364 unsigned int s_mount_opt2; 1365 unsigned int s_mount_flags; 1366 unsigned int s_def_mount_opt; 1367 ext4_fsblk_t s_sb_block; 1368 atomic64_t s_resv_clusters; 1369 kuid_t s_resuid; 1370 kgid_t s_resgid; 1371 unsigned short s_mount_state; 1372 unsigned short s_pad; 1373 int s_addr_per_block_bits; 1374 int s_desc_per_block_bits; 1375 int s_inode_size; 1376 int s_first_ino; 1377 unsigned int s_inode_readahead_blks; 1378 unsigned int s_inode_goal; 1379 u32 s_hash_seed[4]; 1380 int s_def_hash_version; 1381 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1382 struct percpu_counter s_freeclusters_counter; 1383 struct percpu_counter s_freeinodes_counter; 1384 struct percpu_counter s_dirs_counter; 1385 struct percpu_counter s_dirtyclusters_counter; 1386 struct blockgroup_lock *s_blockgroup_lock; 1387 struct proc_dir_entry *s_proc; 1388 struct kobject s_kobj; 1389 struct completion s_kobj_unregister; 1390 struct super_block *s_sb; 1391 1392 /* Journaling */ 1393 struct journal_s *s_journal; 1394 struct list_head s_orphan; 1395 struct mutex s_orphan_lock; 1396 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1397 unsigned long s_commit_interval; 1398 u32 s_max_batch_time; 1399 u32 s_min_batch_time; 1400 struct block_device *journal_bdev; 1401 #ifdef CONFIG_QUOTA 1402 /* Names of quota files with journalled quota */ 1403 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1404 int s_jquota_fmt; /* Format of quota to use */ 1405 #endif 1406 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1407 struct rb_root system_blks; 1408 1409 #ifdef EXTENTS_STATS 1410 /* ext4 extents stats */ 1411 unsigned long s_ext_min; 1412 unsigned long s_ext_max; 1413 unsigned long s_depth_max; 1414 spinlock_t s_ext_stats_lock; 1415 unsigned long s_ext_blocks; 1416 unsigned long s_ext_extents; 1417 #endif 1418 1419 /* for buddy allocator */ 1420 struct ext4_group_info ***s_group_info; 1421 struct inode *s_buddy_cache; 1422 spinlock_t s_md_lock; 1423 unsigned short *s_mb_offsets; 1424 unsigned int *s_mb_maxs; 1425 unsigned int s_group_info_size; 1426 unsigned int s_mb_free_pending; 1427 struct list_head s_freed_data_list; /* List of blocks to be freed 1428 after commit completed */ 1429 1430 /* tunables */ 1431 unsigned long s_stripe; 1432 unsigned int s_mb_stream_request; 1433 unsigned int s_mb_max_to_scan; 1434 unsigned int s_mb_min_to_scan; 1435 unsigned int s_mb_stats; 1436 unsigned int s_mb_order2_reqs; 1437 unsigned int s_mb_group_prealloc; 1438 unsigned int s_max_dir_size_kb; 1439 /* where last allocation was done - for stream allocation */ 1440 unsigned long s_mb_last_group; 1441 unsigned long s_mb_last_start; 1442 1443 /* stats for buddy allocator */ 1444 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1445 atomic_t s_bal_success; /* we found long enough chunks */ 1446 atomic_t s_bal_allocated; /* in blocks */ 1447 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1448 atomic_t s_bal_goals; /* goal hits */ 1449 atomic_t s_bal_breaks; /* too long searches */ 1450 atomic_t s_bal_2orders; /* 2^order hits */ 1451 spinlock_t s_bal_lock; 1452 unsigned long s_mb_buddies_generated; 1453 unsigned long long s_mb_generation_time; 1454 atomic_t s_mb_lost_chunks; 1455 atomic_t s_mb_preallocated; 1456 atomic_t s_mb_discarded; 1457 atomic_t s_lock_busy; 1458 1459 /* locality groups */ 1460 struct ext4_locality_group __percpu *s_locality_groups; 1461 1462 /* for write statistics */ 1463 unsigned long s_sectors_written_start; 1464 u64 s_kbytes_written; 1465 1466 /* the size of zero-out chunk */ 1467 unsigned int s_extent_max_zeroout_kb; 1468 1469 unsigned int s_log_groups_per_flex; 1470 struct flex_groups *s_flex_groups; 1471 ext4_group_t s_flex_groups_allocated; 1472 1473 /* workqueue for reserved extent conversions (buffered io) */ 1474 struct workqueue_struct *rsv_conversion_wq; 1475 1476 /* timer for periodic error stats printing */ 1477 struct timer_list s_err_report; 1478 1479 /* Lazy inode table initialization info */ 1480 struct ext4_li_request *s_li_request; 1481 /* Wait multiplier for lazy initialization thread */ 1482 unsigned int s_li_wait_mult; 1483 1484 /* Kernel thread for multiple mount protection */ 1485 struct task_struct *s_mmp_tsk; 1486 1487 /* record the last minlen when FITRIM is called. */ 1488 atomic_t s_last_trim_minblks; 1489 1490 /* Reference to checksum algorithm driver via cryptoapi */ 1491 struct crypto_shash *s_chksum_driver; 1492 1493 /* Precomputed FS UUID checksum for seeding other checksums */ 1494 __u32 s_csum_seed; 1495 1496 /* Reclaim extents from extent status tree */ 1497 struct shrinker s_es_shrinker; 1498 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1499 long s_es_nr_inode; 1500 struct ext4_es_stats s_es_stats; 1501 struct mb_cache *s_ea_block_cache; 1502 struct mb_cache *s_ea_inode_cache; 1503 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1504 1505 /* Ratelimit ext4 messages. */ 1506 struct ratelimit_state s_err_ratelimit_state; 1507 struct ratelimit_state s_warning_ratelimit_state; 1508 struct ratelimit_state s_msg_ratelimit_state; 1509 1510 /* Barrier between changing inodes' journal flags and writepages ops. */ 1511 struct percpu_rw_semaphore s_journal_flag_rwsem; 1512 struct dax_device *s_daxdev; 1513 }; 1514 1515 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1516 { 1517 return sb->s_fs_info; 1518 } 1519 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1520 { 1521 return container_of(inode, struct ext4_inode_info, vfs_inode); 1522 } 1523 1524 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1525 { 1526 return ino == EXT4_ROOT_INO || 1527 (ino >= EXT4_FIRST_INO(sb) && 1528 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1529 } 1530 1531 /* 1532 * Inode dynamic state flags 1533 */ 1534 enum { 1535 EXT4_STATE_JDATA, /* journaled data exists */ 1536 EXT4_STATE_NEW, /* inode is newly created */ 1537 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1538 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1539 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1540 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1541 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1542 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1543 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1544 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1545 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1546 }; 1547 1548 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1549 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1550 { \ 1551 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1552 } \ 1553 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1554 { \ 1555 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1556 } \ 1557 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1558 { \ 1559 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1560 } 1561 1562 /* Add these declarations here only so that these functions can be 1563 * found by name. Otherwise, they are very hard to locate. */ 1564 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1565 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1566 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1567 EXT4_INODE_BIT_FNS(flag, flags, 0) 1568 1569 /* Add these declarations here only so that these functions can be 1570 * found by name. Otherwise, they are very hard to locate. */ 1571 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1572 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1573 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1574 #if (BITS_PER_LONG < 64) 1575 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1576 1577 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1578 { 1579 (ei)->i_state_flags = 0; 1580 } 1581 #else 1582 EXT4_INODE_BIT_FNS(state, flags, 32) 1583 1584 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1585 { 1586 /* We depend on the fact that callers will set i_flags */ 1587 } 1588 #endif 1589 #else 1590 /* Assume that user mode programs are passing in an ext4fs superblock, not 1591 * a kernel struct super_block. This will allow us to call the feature-test 1592 * macros from user land. */ 1593 #define EXT4_SB(sb) (sb) 1594 #endif 1595 1596 /* 1597 * Returns true if the inode is inode is encrypted 1598 */ 1599 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1600 1601 /* 1602 * Codes for operating systems 1603 */ 1604 #define EXT4_OS_LINUX 0 1605 #define EXT4_OS_HURD 1 1606 #define EXT4_OS_MASIX 2 1607 #define EXT4_OS_FREEBSD 3 1608 #define EXT4_OS_LITES 4 1609 1610 /* 1611 * Revision levels 1612 */ 1613 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1614 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1615 1616 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1617 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1618 1619 #define EXT4_GOOD_OLD_INODE_SIZE 128 1620 1621 /* 1622 * Feature set definitions 1623 */ 1624 1625 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1626 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1627 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1628 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1629 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1630 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1631 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1632 1633 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1634 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1635 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1636 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1637 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1638 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1639 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1640 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1641 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1642 /* 1643 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1644 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1645 * all other data structures' checksums. However, the METADATA_CSUM and 1646 * GDT_CSUM bits are mutually exclusive. 1647 */ 1648 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1649 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1650 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1651 1652 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1653 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1654 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1655 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1656 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1657 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1658 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1659 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1660 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1661 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1662 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1663 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1664 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1665 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1666 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1667 1668 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1669 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1670 { \ 1671 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1672 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1673 } \ 1674 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1675 { \ 1676 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1677 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1678 } \ 1679 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1680 { \ 1681 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1682 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1683 } 1684 1685 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1686 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1687 { \ 1688 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1689 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1690 } \ 1691 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1692 { \ 1693 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1694 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1695 } \ 1696 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1697 { \ 1698 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1699 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1700 } 1701 1702 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1703 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1704 { \ 1705 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1706 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1707 } \ 1708 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1709 { \ 1710 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1711 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1712 } \ 1713 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1714 { \ 1715 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1716 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1717 } 1718 1719 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1720 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1721 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1722 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1723 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1724 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1725 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1726 1727 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1728 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1729 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1730 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1731 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1732 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1733 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1734 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1735 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1736 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1737 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1738 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1739 1740 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1741 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1742 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1743 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1744 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1745 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1746 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1747 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1748 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1749 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1750 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1751 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1752 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1753 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1754 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1755 1756 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1757 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1758 EXT4_FEATURE_INCOMPAT_META_BG) 1759 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1760 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1761 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1762 1763 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1764 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1765 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1766 EXT4_FEATURE_INCOMPAT_META_BG) 1767 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1768 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1769 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1770 1771 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1772 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1773 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1774 EXT4_FEATURE_INCOMPAT_META_BG| \ 1775 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1776 EXT4_FEATURE_INCOMPAT_64BIT| \ 1777 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1778 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1779 EXT4_FEATURE_INCOMPAT_MMP | \ 1780 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1781 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1782 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1783 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1784 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1785 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1786 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1787 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1788 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1789 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1790 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1791 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1792 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1793 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1794 EXT4_FEATURE_RO_COMPAT_PROJECT) 1795 1796 #define EXTN_FEATURE_FUNCS(ver) \ 1797 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1798 { \ 1799 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1800 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1801 } \ 1802 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1803 { \ 1804 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1805 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1806 } \ 1807 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1808 { \ 1809 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1810 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1811 } 1812 1813 EXTN_FEATURE_FUNCS(2) 1814 EXTN_FEATURE_FUNCS(3) 1815 EXTN_FEATURE_FUNCS(4) 1816 1817 static inline bool ext4_has_compat_features(struct super_block *sb) 1818 { 1819 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1820 } 1821 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1822 { 1823 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1824 } 1825 static inline bool ext4_has_incompat_features(struct super_block *sb) 1826 { 1827 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1828 } 1829 1830 /* 1831 * Superblock flags 1832 */ 1833 #define EXT4_FLAGS_RESIZING 0 1834 #define EXT4_FLAGS_SHUTDOWN 1 1835 1836 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1837 { 1838 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1839 } 1840 1841 1842 /* 1843 * Default values for user and/or group using reserved blocks 1844 */ 1845 #define EXT4_DEF_RESUID 0 1846 #define EXT4_DEF_RESGID 0 1847 1848 /* 1849 * Default project ID 1850 */ 1851 #define EXT4_DEF_PROJID 0 1852 1853 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1854 1855 /* 1856 * Default mount options 1857 */ 1858 #define EXT4_DEFM_DEBUG 0x0001 1859 #define EXT4_DEFM_BSDGROUPS 0x0002 1860 #define EXT4_DEFM_XATTR_USER 0x0004 1861 #define EXT4_DEFM_ACL 0x0008 1862 #define EXT4_DEFM_UID16 0x0010 1863 #define EXT4_DEFM_JMODE 0x0060 1864 #define EXT4_DEFM_JMODE_DATA 0x0020 1865 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1866 #define EXT4_DEFM_JMODE_WBACK 0x0060 1867 #define EXT4_DEFM_NOBARRIER 0x0100 1868 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1869 #define EXT4_DEFM_DISCARD 0x0400 1870 #define EXT4_DEFM_NODELALLOC 0x0800 1871 1872 /* 1873 * Default journal batch times 1874 */ 1875 #define EXT4_DEF_MIN_BATCH_TIME 0 1876 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1877 1878 /* 1879 * Minimum number of groups in a flexgroup before we separate out 1880 * directories into the first block group of a flexgroup 1881 */ 1882 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1883 1884 /* 1885 * Structure of a directory entry 1886 */ 1887 #define EXT4_NAME_LEN 255 1888 1889 struct ext4_dir_entry { 1890 __le32 inode; /* Inode number */ 1891 __le16 rec_len; /* Directory entry length */ 1892 __le16 name_len; /* Name length */ 1893 char name[EXT4_NAME_LEN]; /* File name */ 1894 }; 1895 1896 /* 1897 * The new version of the directory entry. Since EXT4 structures are 1898 * stored in intel byte order, and the name_len field could never be 1899 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1900 * file_type field. 1901 */ 1902 struct ext4_dir_entry_2 { 1903 __le32 inode; /* Inode number */ 1904 __le16 rec_len; /* Directory entry length */ 1905 __u8 name_len; /* Name length */ 1906 __u8 file_type; 1907 char name[EXT4_NAME_LEN]; /* File name */ 1908 }; 1909 1910 /* 1911 * This is a bogus directory entry at the end of each leaf block that 1912 * records checksums. 1913 */ 1914 struct ext4_dir_entry_tail { 1915 __le32 det_reserved_zero1; /* Pretend to be unused */ 1916 __le16 det_rec_len; /* 12 */ 1917 __u8 det_reserved_zero2; /* Zero name length */ 1918 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1919 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1920 }; 1921 1922 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1923 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1924 ((blocksize) - \ 1925 sizeof(struct ext4_dir_entry_tail)))) 1926 1927 /* 1928 * Ext4 directory file types. Only the low 3 bits are used. The 1929 * other bits are reserved for now. 1930 */ 1931 #define EXT4_FT_UNKNOWN 0 1932 #define EXT4_FT_REG_FILE 1 1933 #define EXT4_FT_DIR 2 1934 #define EXT4_FT_CHRDEV 3 1935 #define EXT4_FT_BLKDEV 4 1936 #define EXT4_FT_FIFO 5 1937 #define EXT4_FT_SOCK 6 1938 #define EXT4_FT_SYMLINK 7 1939 1940 #define EXT4_FT_MAX 8 1941 1942 #define EXT4_FT_DIR_CSUM 0xDE 1943 1944 /* 1945 * EXT4_DIR_PAD defines the directory entries boundaries 1946 * 1947 * NOTE: It must be a multiple of 4 1948 */ 1949 #define EXT4_DIR_PAD 4 1950 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1951 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1952 ~EXT4_DIR_ROUND) 1953 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1954 1955 /* 1956 * If we ever get support for fs block sizes > page_size, we'll need 1957 * to remove the #if statements in the next two functions... 1958 */ 1959 static inline unsigned int 1960 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1961 { 1962 unsigned len = le16_to_cpu(dlen); 1963 1964 #if (PAGE_SIZE >= 65536) 1965 if (len == EXT4_MAX_REC_LEN || len == 0) 1966 return blocksize; 1967 return (len & 65532) | ((len & 3) << 16); 1968 #else 1969 return len; 1970 #endif 1971 } 1972 1973 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1974 { 1975 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1976 BUG(); 1977 #if (PAGE_SIZE >= 65536) 1978 if (len < 65536) 1979 return cpu_to_le16(len); 1980 if (len == blocksize) { 1981 if (blocksize == 65536) 1982 return cpu_to_le16(EXT4_MAX_REC_LEN); 1983 else 1984 return cpu_to_le16(0); 1985 } 1986 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1987 #else 1988 return cpu_to_le16(len); 1989 #endif 1990 } 1991 1992 /* 1993 * Hash Tree Directory indexing 1994 * (c) Daniel Phillips, 2001 1995 */ 1996 1997 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1998 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1999 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2000 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2001 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2002 2003 /* Legal values for the dx_root hash_version field: */ 2004 2005 #define DX_HASH_LEGACY 0 2006 #define DX_HASH_HALF_MD4 1 2007 #define DX_HASH_TEA 2 2008 #define DX_HASH_LEGACY_UNSIGNED 3 2009 #define DX_HASH_HALF_MD4_UNSIGNED 4 2010 #define DX_HASH_TEA_UNSIGNED 5 2011 2012 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2013 const void *address, unsigned int length) 2014 { 2015 struct { 2016 struct shash_desc shash; 2017 char ctx[4]; 2018 } desc; 2019 2020 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2021 2022 desc.shash.tfm = sbi->s_chksum_driver; 2023 desc.shash.flags = 0; 2024 *(u32 *)desc.ctx = crc; 2025 2026 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2027 2028 return *(u32 *)desc.ctx; 2029 } 2030 2031 #ifdef __KERNEL__ 2032 2033 /* hash info structure used by the directory hash */ 2034 struct dx_hash_info 2035 { 2036 u32 hash; 2037 u32 minor_hash; 2038 int hash_version; 2039 u32 *seed; 2040 }; 2041 2042 2043 /* 32 and 64 bit signed EOF for dx directories */ 2044 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2045 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2046 2047 2048 /* 2049 * Control parameters used by ext4_htree_next_block 2050 */ 2051 #define HASH_NB_ALWAYS 1 2052 2053 struct ext4_filename { 2054 const struct qstr *usr_fname; 2055 struct fscrypt_str disk_name; 2056 struct dx_hash_info hinfo; 2057 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2058 struct fscrypt_str crypto_buf; 2059 #endif 2060 }; 2061 2062 #define fname_name(p) ((p)->disk_name.name) 2063 #define fname_len(p) ((p)->disk_name.len) 2064 2065 /* 2066 * Describe an inode's exact location on disk and in memory 2067 */ 2068 struct ext4_iloc 2069 { 2070 struct buffer_head *bh; 2071 unsigned long offset; 2072 ext4_group_t block_group; 2073 }; 2074 2075 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2076 { 2077 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2078 } 2079 2080 static inline bool ext4_is_quota_file(struct inode *inode) 2081 { 2082 return IS_NOQUOTA(inode) && 2083 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2084 } 2085 2086 /* 2087 * This structure is stuffed into the struct file's private_data field 2088 * for directories. It is where we put information so that we can do 2089 * readdir operations in hash tree order. 2090 */ 2091 struct dir_private_info { 2092 struct rb_root root; 2093 struct rb_node *curr_node; 2094 struct fname *extra_fname; 2095 loff_t last_pos; 2096 __u32 curr_hash; 2097 __u32 curr_minor_hash; 2098 __u32 next_hash; 2099 }; 2100 2101 /* calculate the first block number of the group */ 2102 static inline ext4_fsblk_t 2103 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2104 { 2105 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2106 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2107 } 2108 2109 /* 2110 * Special error return code only used by dx_probe() and its callers. 2111 */ 2112 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2113 2114 /* htree levels for ext4 */ 2115 #define EXT4_HTREE_LEVEL_COMPAT 2 2116 #define EXT4_HTREE_LEVEL 3 2117 2118 static inline int ext4_dir_htree_level(struct super_block *sb) 2119 { 2120 return ext4_has_feature_largedir(sb) ? 2121 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2122 } 2123 2124 /* 2125 * Timeout and state flag for lazy initialization inode thread. 2126 */ 2127 #define EXT4_DEF_LI_WAIT_MULT 10 2128 #define EXT4_DEF_LI_MAX_START_DELAY 5 2129 #define EXT4_LAZYINIT_QUIT 0x0001 2130 #define EXT4_LAZYINIT_RUNNING 0x0002 2131 2132 /* 2133 * Lazy inode table initialization info 2134 */ 2135 struct ext4_lazy_init { 2136 unsigned long li_state; 2137 struct list_head li_request_list; 2138 struct mutex li_list_mtx; 2139 }; 2140 2141 struct ext4_li_request { 2142 struct super_block *lr_super; 2143 struct ext4_sb_info *lr_sbi; 2144 ext4_group_t lr_next_group; 2145 struct list_head lr_request; 2146 unsigned long lr_next_sched; 2147 unsigned long lr_timeout; 2148 }; 2149 2150 struct ext4_features { 2151 struct kobject f_kobj; 2152 struct completion f_kobj_unregister; 2153 }; 2154 2155 /* 2156 * This structure will be used for multiple mount protection. It will be 2157 * written into the block number saved in the s_mmp_block field in the 2158 * superblock. Programs that check MMP should assume that if 2159 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2160 * to use the filesystem, regardless of how old the timestamp is. 2161 */ 2162 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2163 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2164 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2165 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2166 2167 struct mmp_struct { 2168 __le32 mmp_magic; /* Magic number for MMP */ 2169 __le32 mmp_seq; /* Sequence no. updated periodically */ 2170 2171 /* 2172 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2173 * purposes and do not affect the correctness of the algorithm 2174 */ 2175 __le64 mmp_time; /* Time last updated */ 2176 char mmp_nodename[64]; /* Node which last updated MMP block */ 2177 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2178 2179 /* 2180 * mmp_check_interval is used to verify if the MMP block has been 2181 * updated on the block device. The value is updated based on the 2182 * maximum time to write the MMP block during an update cycle. 2183 */ 2184 __le16 mmp_check_interval; 2185 2186 __le16 mmp_pad1; 2187 __le32 mmp_pad2[226]; 2188 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2189 }; 2190 2191 /* arguments passed to the mmp thread */ 2192 struct mmpd_data { 2193 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2194 struct super_block *sb; /* super block of the fs */ 2195 }; 2196 2197 /* 2198 * Check interval multiplier 2199 * The MMP block is written every update interval and initially checked every 2200 * update interval x the multiplier (the value is then adapted based on the 2201 * write latency). The reason is that writes can be delayed under load and we 2202 * don't want readers to incorrectly assume that the filesystem is no longer 2203 * in use. 2204 */ 2205 #define EXT4_MMP_CHECK_MULT 2UL 2206 2207 /* 2208 * Minimum interval for MMP checking in seconds. 2209 */ 2210 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2211 2212 /* 2213 * Maximum interval for MMP checking in seconds. 2214 */ 2215 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2216 2217 /* 2218 * Function prototypes 2219 */ 2220 2221 /* 2222 * Ok, these declarations are also in <linux/kernel.h> but none of the 2223 * ext4 source programs needs to include it so they are duplicated here. 2224 */ 2225 # define NORET_TYPE /**/ 2226 # define ATTRIB_NORET __attribute__((noreturn)) 2227 # define NORET_AND noreturn, 2228 2229 /* bitmap.c */ 2230 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2231 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2232 struct ext4_group_desc *gdp, 2233 struct buffer_head *bh, int sz); 2234 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2235 struct ext4_group_desc *gdp, 2236 struct buffer_head *bh, int sz); 2237 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2238 struct ext4_group_desc *gdp, 2239 struct buffer_head *bh); 2240 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2241 struct ext4_group_desc *gdp, 2242 struct buffer_head *bh); 2243 2244 /* balloc.c */ 2245 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2246 ext4_fsblk_t blocknr, 2247 ext4_group_t *blockgrpp, 2248 ext4_grpblk_t *offsetp); 2249 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2250 ext4_fsblk_t block); 2251 2252 extern unsigned int ext4_block_group(struct super_block *sb, 2253 ext4_fsblk_t blocknr); 2254 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2255 ext4_fsblk_t blocknr); 2256 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2257 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2258 ext4_group_t group); 2259 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2260 ext4_fsblk_t goal, 2261 unsigned int flags, 2262 unsigned long *count, 2263 int *errp); 2264 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2265 s64 nclusters, unsigned int flags); 2266 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2267 extern void ext4_check_blocks_bitmap(struct super_block *); 2268 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2269 ext4_group_t block_group, 2270 struct buffer_head ** bh); 2271 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2272 2273 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2274 ext4_group_t block_group); 2275 extern int ext4_wait_block_bitmap(struct super_block *sb, 2276 ext4_group_t block_group, 2277 struct buffer_head *bh); 2278 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2279 ext4_group_t block_group); 2280 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2281 ext4_group_t block_group, 2282 struct ext4_group_desc *gdp); 2283 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2284 2285 static inline bool ext4_encrypted_inode(struct inode *inode) 2286 { 2287 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2288 } 2289 2290 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2291 static inline int ext4_fname_setup_filename(struct inode *dir, 2292 const struct qstr *iname, 2293 int lookup, struct ext4_filename *fname) 2294 { 2295 struct fscrypt_name name; 2296 int err; 2297 2298 memset(fname, 0, sizeof(struct ext4_filename)); 2299 2300 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2301 2302 fname->usr_fname = name.usr_fname; 2303 fname->disk_name = name.disk_name; 2304 fname->hinfo.hash = name.hash; 2305 fname->hinfo.minor_hash = name.minor_hash; 2306 fname->crypto_buf = name.crypto_buf; 2307 return err; 2308 } 2309 2310 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2311 { 2312 struct fscrypt_name name; 2313 2314 name.crypto_buf = fname->crypto_buf; 2315 fscrypt_free_filename(&name); 2316 2317 fname->crypto_buf.name = NULL; 2318 fname->usr_fname = NULL; 2319 fname->disk_name.name = NULL; 2320 } 2321 #else 2322 static inline int ext4_fname_setup_filename(struct inode *dir, 2323 const struct qstr *iname, 2324 int lookup, struct ext4_filename *fname) 2325 { 2326 fname->usr_fname = iname; 2327 fname->disk_name.name = (unsigned char *) iname->name; 2328 fname->disk_name.len = iname->len; 2329 return 0; 2330 } 2331 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2332 2333 #endif 2334 2335 /* dir.c */ 2336 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2337 struct file *, 2338 struct ext4_dir_entry_2 *, 2339 struct buffer_head *, char *, int, 2340 unsigned int); 2341 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2342 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2343 (de), (bh), (buf), (size), (offset))) 2344 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2345 __u32 minor_hash, 2346 struct ext4_dir_entry_2 *dirent, 2347 struct fscrypt_str *ent_name); 2348 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2349 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2350 struct buffer_head *bh, 2351 void *buf, int buf_size, 2352 struct ext4_filename *fname, 2353 struct ext4_dir_entry_2 **dest_de); 2354 void ext4_insert_dentry(struct inode *inode, 2355 struct ext4_dir_entry_2 *de, 2356 int buf_size, 2357 struct ext4_filename *fname); 2358 static inline void ext4_update_dx_flag(struct inode *inode) 2359 { 2360 if (!ext4_has_feature_dir_index(inode->i_sb)) 2361 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2362 } 2363 static const unsigned char ext4_filetype_table[] = { 2364 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2365 }; 2366 2367 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2368 { 2369 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2370 return DT_UNKNOWN; 2371 2372 return ext4_filetype_table[filetype]; 2373 } 2374 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2375 void *buf, int buf_size); 2376 2377 /* fsync.c */ 2378 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2379 2380 /* hash.c */ 2381 extern int ext4fs_dirhash(const char *name, int len, struct 2382 dx_hash_info *hinfo); 2383 2384 /* ialloc.c */ 2385 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2386 const struct qstr *qstr, __u32 goal, 2387 uid_t *owner, __u32 i_flags, 2388 int handle_type, unsigned int line_no, 2389 int nblocks); 2390 2391 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2392 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2393 i_flags, 0, 0, 0) 2394 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2395 type, nblocks) \ 2396 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2397 0, (type), __LINE__, (nblocks)) 2398 2399 2400 extern void ext4_free_inode(handle_t *, struct inode *); 2401 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2402 extern unsigned long ext4_count_free_inodes(struct super_block *); 2403 extern unsigned long ext4_count_dirs(struct super_block *); 2404 extern void ext4_check_inodes_bitmap(struct super_block *); 2405 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2406 extern int ext4_init_inode_table(struct super_block *sb, 2407 ext4_group_t group, int barrier); 2408 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2409 2410 /* mballoc.c */ 2411 extern const struct seq_operations ext4_mb_seq_groups_ops; 2412 extern long ext4_mb_stats; 2413 extern long ext4_mb_max_to_scan; 2414 extern int ext4_mb_init(struct super_block *); 2415 extern int ext4_mb_release(struct super_block *); 2416 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2417 struct ext4_allocation_request *, int *); 2418 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2419 extern void ext4_discard_preallocations(struct inode *); 2420 extern int __init ext4_init_mballoc(void); 2421 extern void ext4_exit_mballoc(void); 2422 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2423 struct buffer_head *bh, ext4_fsblk_t block, 2424 unsigned long count, int flags); 2425 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2426 ext4_group_t ngroups); 2427 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2428 ext4_group_t i, struct ext4_group_desc *desc); 2429 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2430 ext4_fsblk_t block, unsigned long count); 2431 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2432 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2433 2434 /* inode.c */ 2435 int ext4_inode_is_fast_symlink(struct inode *inode); 2436 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2437 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2438 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2439 bool wait, struct buffer_head **bhs); 2440 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2441 struct buffer_head *bh_result, int create); 2442 int ext4_get_block(struct inode *inode, sector_t iblock, 2443 struct buffer_head *bh_result, int create); 2444 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2445 struct buffer_head *bh_result, int create); 2446 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2447 struct buffer_head *bh, int create); 2448 int ext4_walk_page_buffers(handle_t *handle, 2449 struct buffer_head *head, 2450 unsigned from, 2451 unsigned to, 2452 int *partial, 2453 int (*fn)(handle_t *handle, 2454 struct buffer_head *bh)); 2455 int do_journal_get_write_access(handle_t *handle, 2456 struct buffer_head *bh); 2457 #define FALL_BACK_TO_NONDELALLOC 1 2458 #define CONVERT_INLINE_DATA 2 2459 2460 typedef enum { 2461 EXT4_IGET_NORMAL = 0, 2462 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2463 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2464 } ext4_iget_flags; 2465 2466 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2467 ext4_iget_flags flags, const char *function, 2468 unsigned int line); 2469 2470 #define ext4_iget(sb, ino, flags) \ 2471 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2472 2473 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2474 extern int ext4_setattr(struct dentry *, struct iattr *); 2475 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2476 extern void ext4_evict_inode(struct inode *); 2477 extern void ext4_clear_inode(struct inode *); 2478 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2479 extern int ext4_sync_inode(handle_t *, struct inode *); 2480 extern void ext4_dirty_inode(struct inode *, int); 2481 extern int ext4_change_inode_journal_flag(struct inode *, int); 2482 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2483 extern int ext4_inode_attach_jinode(struct inode *inode); 2484 extern int ext4_can_truncate(struct inode *inode); 2485 extern int ext4_truncate(struct inode *); 2486 extern int ext4_break_layouts(struct inode *); 2487 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2488 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2489 extern void ext4_set_inode_flags(struct inode *); 2490 extern int ext4_alloc_da_blocks(struct inode *inode); 2491 extern void ext4_set_aops(struct inode *inode); 2492 extern int ext4_writepage_trans_blocks(struct inode *); 2493 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2494 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2495 loff_t lstart, loff_t lend); 2496 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2497 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2498 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2499 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2500 extern void ext4_da_release_space(struct inode *inode, int to_free); 2501 extern void ext4_da_update_reserve_space(struct inode *inode, 2502 int used, int quota_claim); 2503 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2504 ext4_fsblk_t pblk, ext4_lblk_t len); 2505 2506 /* indirect.c */ 2507 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2508 struct ext4_map_blocks *map, int flags); 2509 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2510 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2511 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2512 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2513 ext4_lblk_t start, ext4_lblk_t end); 2514 2515 /* ioctl.c */ 2516 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2517 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2518 2519 /* migrate.c */ 2520 extern int ext4_ext_migrate(struct inode *); 2521 extern int ext4_ind_migrate(struct inode *inode); 2522 2523 /* namei.c */ 2524 extern int ext4_dirent_csum_verify(struct inode *inode, 2525 struct ext4_dir_entry *dirent); 2526 extern int ext4_orphan_add(handle_t *, struct inode *); 2527 extern int ext4_orphan_del(handle_t *, struct inode *); 2528 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2529 __u32 start_minor_hash, __u32 *next_hash); 2530 extern int ext4_search_dir(struct buffer_head *bh, 2531 char *search_buf, 2532 int buf_size, 2533 struct inode *dir, 2534 struct ext4_filename *fname, 2535 unsigned int offset, 2536 struct ext4_dir_entry_2 **res_dir); 2537 extern int ext4_generic_delete_entry(handle_t *handle, 2538 struct inode *dir, 2539 struct ext4_dir_entry_2 *de_del, 2540 struct buffer_head *bh, 2541 void *entry_buf, 2542 int buf_size, 2543 int csum_size); 2544 extern bool ext4_empty_dir(struct inode *inode); 2545 2546 /* resize.c */ 2547 extern int ext4_group_add(struct super_block *sb, 2548 struct ext4_new_group_data *input); 2549 extern int ext4_group_extend(struct super_block *sb, 2550 struct ext4_super_block *es, 2551 ext4_fsblk_t n_blocks_count); 2552 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2553 2554 /* super.c */ 2555 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2556 sector_t block, int op_flags); 2557 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2558 extern int ext4_calculate_overhead(struct super_block *sb); 2559 extern void ext4_superblock_csum_set(struct super_block *sb); 2560 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2561 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2562 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2563 ext4_group_t ngroup); 2564 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2565 char nbuf[16]); 2566 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2567 ext4_group_t block_group, 2568 unsigned int flags); 2569 2570 extern __printf(4, 5) 2571 void __ext4_error(struct super_block *, const char *, unsigned int, 2572 const char *, ...); 2573 extern __printf(5, 6) 2574 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2575 const char *, ...); 2576 extern __printf(5, 6) 2577 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2578 const char *, ...); 2579 extern void __ext4_std_error(struct super_block *, const char *, 2580 unsigned int, int); 2581 extern __printf(4, 5) 2582 void __ext4_abort(struct super_block *, const char *, unsigned int, 2583 const char *, ...); 2584 extern __printf(4, 5) 2585 void __ext4_warning(struct super_block *, const char *, unsigned int, 2586 const char *, ...); 2587 extern __printf(4, 5) 2588 void __ext4_warning_inode(const struct inode *inode, const char *function, 2589 unsigned int line, const char *fmt, ...); 2590 extern __printf(3, 4) 2591 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2592 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2593 const char *, unsigned int, const char *); 2594 extern __printf(7, 8) 2595 void __ext4_grp_locked_error(const char *, unsigned int, 2596 struct super_block *, ext4_group_t, 2597 unsigned long, ext4_fsblk_t, 2598 const char *, ...); 2599 2600 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2601 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2602 2603 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2604 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2605 2606 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2607 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2608 2609 #ifdef CONFIG_PRINTK 2610 2611 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2612 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2613 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2614 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2615 #define ext4_error(sb, fmt, ...) \ 2616 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2617 #define ext4_abort(sb, fmt, ...) \ 2618 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2619 #define ext4_warning(sb, fmt, ...) \ 2620 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2621 #define ext4_warning_inode(inode, fmt, ...) \ 2622 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2623 #define ext4_msg(sb, level, fmt, ...) \ 2624 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2625 #define dump_mmp_msg(sb, mmp, msg) \ 2626 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2627 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2628 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2629 fmt, ##__VA_ARGS__) 2630 2631 #else 2632 2633 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2634 do { \ 2635 no_printk(fmt, ##__VA_ARGS__); \ 2636 __ext4_error_inode(inode, "", 0, block, " "); \ 2637 } while (0) 2638 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2639 do { \ 2640 no_printk(fmt, ##__VA_ARGS__); \ 2641 __ext4_error_file(file, "", 0, block, " "); \ 2642 } while (0) 2643 #define ext4_error(sb, fmt, ...) \ 2644 do { \ 2645 no_printk(fmt, ##__VA_ARGS__); \ 2646 __ext4_error(sb, "", 0, " "); \ 2647 } while (0) 2648 #define ext4_abort(sb, fmt, ...) \ 2649 do { \ 2650 no_printk(fmt, ##__VA_ARGS__); \ 2651 __ext4_abort(sb, "", 0, " "); \ 2652 } while (0) 2653 #define ext4_warning(sb, fmt, ...) \ 2654 do { \ 2655 no_printk(fmt, ##__VA_ARGS__); \ 2656 __ext4_warning(sb, "", 0, " "); \ 2657 } while (0) 2658 #define ext4_warning_inode(inode, fmt, ...) \ 2659 do { \ 2660 no_printk(fmt, ##__VA_ARGS__); \ 2661 __ext4_warning_inode(inode, "", 0, " "); \ 2662 } while (0) 2663 #define ext4_msg(sb, level, fmt, ...) \ 2664 do { \ 2665 no_printk(fmt, ##__VA_ARGS__); \ 2666 __ext4_msg(sb, "", " "); \ 2667 } while (0) 2668 #define dump_mmp_msg(sb, mmp, msg) \ 2669 __dump_mmp_msg(sb, mmp, "", 0, "") 2670 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2671 do { \ 2672 no_printk(fmt, ##__VA_ARGS__); \ 2673 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2674 } while (0) 2675 2676 #endif 2677 2678 extern void ext4_update_dynamic_rev(struct super_block *sb); 2679 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2680 __u32 compat); 2681 extern int ext4_update_rocompat_feature(handle_t *handle, 2682 struct super_block *sb, __u32 rocompat); 2683 extern int ext4_update_incompat_feature(handle_t *handle, 2684 struct super_block *sb, __u32 incompat); 2685 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2686 struct ext4_group_desc *bg); 2687 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2688 struct ext4_group_desc *bg); 2689 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2690 struct ext4_group_desc *bg); 2691 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2692 struct ext4_group_desc *bg); 2693 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2694 struct ext4_group_desc *bg); 2695 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2696 struct ext4_group_desc *bg); 2697 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2698 struct ext4_group_desc *bg); 2699 extern void ext4_block_bitmap_set(struct super_block *sb, 2700 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2701 extern void ext4_inode_bitmap_set(struct super_block *sb, 2702 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2703 extern void ext4_inode_table_set(struct super_block *sb, 2704 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2705 extern void ext4_free_group_clusters_set(struct super_block *sb, 2706 struct ext4_group_desc *bg, 2707 __u32 count); 2708 extern void ext4_free_inodes_set(struct super_block *sb, 2709 struct ext4_group_desc *bg, __u32 count); 2710 extern void ext4_used_dirs_set(struct super_block *sb, 2711 struct ext4_group_desc *bg, __u32 count); 2712 extern void ext4_itable_unused_set(struct super_block *sb, 2713 struct ext4_group_desc *bg, __u32 count); 2714 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2715 struct ext4_group_desc *gdp); 2716 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2717 struct ext4_group_desc *gdp); 2718 extern int ext4_register_li_request(struct super_block *sb, 2719 ext4_group_t first_not_zeroed); 2720 2721 static inline int ext4_has_metadata_csum(struct super_block *sb) 2722 { 2723 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2724 !EXT4_SB(sb)->s_chksum_driver); 2725 2726 return ext4_has_feature_metadata_csum(sb) && 2727 (EXT4_SB(sb)->s_chksum_driver != NULL); 2728 } 2729 2730 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2731 { 2732 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2733 } 2734 2735 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2736 { 2737 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2738 le32_to_cpu(es->s_blocks_count_lo); 2739 } 2740 2741 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2742 { 2743 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2744 le32_to_cpu(es->s_r_blocks_count_lo); 2745 } 2746 2747 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2748 { 2749 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2750 le32_to_cpu(es->s_free_blocks_count_lo); 2751 } 2752 2753 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2754 ext4_fsblk_t blk) 2755 { 2756 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2757 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2758 } 2759 2760 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2761 ext4_fsblk_t blk) 2762 { 2763 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2764 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2765 } 2766 2767 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2768 ext4_fsblk_t blk) 2769 { 2770 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2771 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2772 } 2773 2774 static inline loff_t ext4_isize(struct super_block *sb, 2775 struct ext4_inode *raw_inode) 2776 { 2777 if (ext4_has_feature_largedir(sb) || 2778 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2779 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2780 le32_to_cpu(raw_inode->i_size_lo); 2781 2782 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2783 } 2784 2785 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2786 { 2787 raw_inode->i_size_lo = cpu_to_le32(i_size); 2788 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2789 } 2790 2791 static inline 2792 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2793 ext4_group_t group) 2794 { 2795 struct ext4_group_info ***grp_info; 2796 long indexv, indexh; 2797 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2798 grp_info = EXT4_SB(sb)->s_group_info; 2799 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2800 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2801 return grp_info[indexv][indexh]; 2802 } 2803 2804 /* 2805 * Reading s_groups_count requires using smp_rmb() afterwards. See 2806 * the locking protocol documented in the comments of ext4_group_add() 2807 * in resize.c 2808 */ 2809 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2810 { 2811 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2812 2813 smp_rmb(); 2814 return ngroups; 2815 } 2816 2817 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2818 ext4_group_t block_group) 2819 { 2820 return block_group >> sbi->s_log_groups_per_flex; 2821 } 2822 2823 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2824 { 2825 return 1 << sbi->s_log_groups_per_flex; 2826 } 2827 2828 #define ext4_std_error(sb, errno) \ 2829 do { \ 2830 if ((errno)) \ 2831 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2832 } while (0) 2833 2834 #ifdef CONFIG_SMP 2835 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2836 * counters. So we need to make sure we have free clusters more 2837 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2838 */ 2839 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2840 #else 2841 #define EXT4_FREECLUSTERS_WATERMARK 0 2842 #endif 2843 2844 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2845 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2846 { 2847 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2848 !inode_is_locked(inode)); 2849 down_write(&EXT4_I(inode)->i_data_sem); 2850 if (newsize > EXT4_I(inode)->i_disksize) 2851 EXT4_I(inode)->i_disksize = newsize; 2852 up_write(&EXT4_I(inode)->i_data_sem); 2853 } 2854 2855 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2856 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2857 { 2858 int changed = 0; 2859 2860 if (newsize > inode->i_size) { 2861 i_size_write(inode, newsize); 2862 changed = 1; 2863 } 2864 if (newsize > EXT4_I(inode)->i_disksize) { 2865 ext4_update_i_disksize(inode, newsize); 2866 changed |= 2; 2867 } 2868 return changed; 2869 } 2870 2871 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2872 loff_t len); 2873 2874 struct ext4_group_info { 2875 unsigned long bb_state; 2876 struct rb_root bb_free_root; 2877 ext4_grpblk_t bb_first_free; /* first free block */ 2878 ext4_grpblk_t bb_free; /* total free blocks */ 2879 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2880 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2881 struct list_head bb_prealloc_list; 2882 #ifdef DOUBLE_CHECK 2883 void *bb_bitmap; 2884 #endif 2885 struct rw_semaphore alloc_sem; 2886 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2887 * regions, index is order. 2888 * bb_counters[3] = 5 means 2889 * 5 free 8-block regions. */ 2890 }; 2891 2892 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2893 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2894 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2895 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2896 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2897 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2898 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2899 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2900 2901 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2902 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2903 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2904 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2905 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2906 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2907 2908 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2909 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2910 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2911 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2912 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2913 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2914 2915 #define EXT4_MAX_CONTENTION 8 2916 #define EXT4_CONTENTION_THRESHOLD 2 2917 2918 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2919 ext4_group_t group) 2920 { 2921 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2922 } 2923 2924 /* 2925 * Returns true if the filesystem is busy enough that attempts to 2926 * access the block group locks has run into contention. 2927 */ 2928 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2929 { 2930 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2931 } 2932 2933 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2934 { 2935 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2936 if (spin_trylock(lock)) 2937 /* 2938 * We're able to grab the lock right away, so drop the 2939 * lock contention counter. 2940 */ 2941 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2942 else { 2943 /* 2944 * The lock is busy, so bump the contention counter, 2945 * and then wait on the spin lock. 2946 */ 2947 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2948 EXT4_MAX_CONTENTION); 2949 spin_lock(lock); 2950 } 2951 } 2952 2953 static inline void ext4_unlock_group(struct super_block *sb, 2954 ext4_group_t group) 2955 { 2956 spin_unlock(ext4_group_lock_ptr(sb, group)); 2957 } 2958 2959 /* 2960 * Block validity checking 2961 */ 2962 #define ext4_check_indirect_blockref(inode, bh) \ 2963 ext4_check_blockref(__func__, __LINE__, inode, \ 2964 (__le32 *)(bh)->b_data, \ 2965 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2966 2967 #define ext4_ind_check_inode(inode) \ 2968 ext4_check_blockref(__func__, __LINE__, inode, \ 2969 EXT4_I(inode)->i_data, \ 2970 EXT4_NDIR_BLOCKS) 2971 2972 /* 2973 * Inodes and files operations 2974 */ 2975 2976 /* dir.c */ 2977 extern const struct file_operations ext4_dir_operations; 2978 2979 /* file.c */ 2980 extern const struct inode_operations ext4_file_inode_operations; 2981 extern const struct file_operations ext4_file_operations; 2982 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2983 2984 /* inline.c */ 2985 extern int ext4_get_max_inline_size(struct inode *inode); 2986 extern int ext4_find_inline_data_nolock(struct inode *inode); 2987 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2988 unsigned int len); 2989 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2990 2991 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2992 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2993 struct inode *inode, 2994 loff_t pos, unsigned len, 2995 unsigned flags, 2996 struct page **pagep); 2997 extern int ext4_write_inline_data_end(struct inode *inode, 2998 loff_t pos, unsigned len, 2999 unsigned copied, 3000 struct page *page); 3001 extern struct buffer_head * 3002 ext4_journalled_write_inline_data(struct inode *inode, 3003 unsigned len, 3004 struct page *page); 3005 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3006 struct inode *inode, 3007 loff_t pos, unsigned len, 3008 unsigned flags, 3009 struct page **pagep, 3010 void **fsdata); 3011 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3012 unsigned len, unsigned copied, 3013 struct page *page); 3014 extern int ext4_try_add_inline_entry(handle_t *handle, 3015 struct ext4_filename *fname, 3016 struct inode *dir, struct inode *inode); 3017 extern int ext4_try_create_inline_dir(handle_t *handle, 3018 struct inode *parent, 3019 struct inode *inode); 3020 extern int ext4_read_inline_dir(struct file *filp, 3021 struct dir_context *ctx, 3022 int *has_inline_data); 3023 extern int htree_inlinedir_to_tree(struct file *dir_file, 3024 struct inode *dir, ext4_lblk_t block, 3025 struct dx_hash_info *hinfo, 3026 __u32 start_hash, __u32 start_minor_hash, 3027 int *has_inline_data); 3028 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3029 struct ext4_filename *fname, 3030 struct ext4_dir_entry_2 **res_dir, 3031 int *has_inline_data); 3032 extern int ext4_delete_inline_entry(handle_t *handle, 3033 struct inode *dir, 3034 struct ext4_dir_entry_2 *de_del, 3035 struct buffer_head *bh, 3036 int *has_inline_data); 3037 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3038 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3039 struct ext4_dir_entry_2 **parent_de, 3040 int *retval); 3041 extern int ext4_inline_data_fiemap(struct inode *inode, 3042 struct fiemap_extent_info *fieinfo, 3043 int *has_inline, __u64 start, __u64 len); 3044 3045 struct iomap; 3046 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3047 3048 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3049 3050 extern int ext4_convert_inline_data(struct inode *inode); 3051 3052 static inline int ext4_has_inline_data(struct inode *inode) 3053 { 3054 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3055 EXT4_I(inode)->i_inline_off; 3056 } 3057 3058 /* namei.c */ 3059 extern const struct inode_operations ext4_dir_inode_operations; 3060 extern const struct inode_operations ext4_special_inode_operations; 3061 extern struct dentry *ext4_get_parent(struct dentry *child); 3062 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3063 struct ext4_dir_entry_2 *de, 3064 int blocksize, int csum_size, 3065 unsigned int parent_ino, int dotdot_real_len); 3066 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3067 unsigned int blocksize); 3068 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3069 struct inode *inode, 3070 struct buffer_head *bh); 3071 #define S_SHIFT 12 3072 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3073 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3074 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3075 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3076 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3077 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3078 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3079 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3080 }; 3081 3082 static inline void ext4_set_de_type(struct super_block *sb, 3083 struct ext4_dir_entry_2 *de, 3084 umode_t mode) { 3085 if (ext4_has_feature_filetype(sb)) 3086 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3087 } 3088 3089 /* readpages.c */ 3090 extern int ext4_mpage_readpages(struct address_space *mapping, 3091 struct list_head *pages, struct page *page, 3092 unsigned nr_pages, bool is_readahead); 3093 3094 /* symlink.c */ 3095 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3096 extern const struct inode_operations ext4_symlink_inode_operations; 3097 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3098 3099 /* sysfs.c */ 3100 extern int ext4_register_sysfs(struct super_block *sb); 3101 extern void ext4_unregister_sysfs(struct super_block *sb); 3102 extern int __init ext4_init_sysfs(void); 3103 extern void ext4_exit_sysfs(void); 3104 3105 /* block_validity */ 3106 extern void ext4_release_system_zone(struct super_block *sb); 3107 extern int ext4_setup_system_zone(struct super_block *sb); 3108 extern int __init ext4_init_system_zone(void); 3109 extern void ext4_exit_system_zone(void); 3110 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3111 ext4_fsblk_t start_blk, 3112 unsigned int count); 3113 extern int ext4_check_blockref(const char *, unsigned int, 3114 struct inode *, __le32 *, unsigned int); 3115 3116 /* extents.c */ 3117 struct ext4_ext_path; 3118 struct ext4_extent; 3119 3120 /* 3121 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3122 * __le32. 3123 */ 3124 #define EXT_MAX_BLOCKS 0xffffffff 3125 3126 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3127 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3128 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3129 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3130 struct ext4_map_blocks *map, int flags); 3131 extern int ext4_ext_truncate(handle_t *, struct inode *); 3132 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3133 ext4_lblk_t end); 3134 extern void ext4_ext_init(struct super_block *); 3135 extern void ext4_ext_release(struct super_block *); 3136 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3137 loff_t len); 3138 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3139 loff_t offset, ssize_t len); 3140 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3141 struct ext4_map_blocks *map, int flags); 3142 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3143 ext4_lblk_t lblocks); 3144 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3145 int num, 3146 struct ext4_ext_path *path); 3147 extern int ext4_can_extents_be_merged(struct inode *inode, 3148 struct ext4_extent *ex1, 3149 struct ext4_extent *ex2); 3150 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3151 struct ext4_ext_path **, 3152 struct ext4_extent *, int); 3153 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3154 struct ext4_ext_path **, 3155 int flags); 3156 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3157 extern int ext4_ext_check_inode(struct inode *inode); 3158 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3159 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3160 __u64 start, __u64 len); 3161 extern int ext4_ext_precache(struct inode *inode); 3162 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3163 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3164 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3165 struct inode *inode2, ext4_lblk_t lblk1, 3166 ext4_lblk_t lblk2, ext4_lblk_t count, 3167 int mark_unwritten,int *err); 3168 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3169 3170 /* move_extent.c */ 3171 extern void ext4_double_down_write_data_sem(struct inode *first, 3172 struct inode *second); 3173 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3174 struct inode *donor_inode); 3175 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3176 __u64 start_orig, __u64 start_donor, 3177 __u64 len, __u64 *moved_len); 3178 3179 /* page-io.c */ 3180 extern int __init ext4_init_pageio(void); 3181 extern void ext4_exit_pageio(void); 3182 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3183 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3184 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3185 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3186 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3187 struct writeback_control *wbc); 3188 extern void ext4_end_io_rsv_work(struct work_struct *work); 3189 extern void ext4_io_submit(struct ext4_io_submit *io); 3190 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3191 struct page *page, 3192 int len, 3193 struct writeback_control *wbc, 3194 bool keep_towrite); 3195 3196 /* mmp.c */ 3197 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3198 3199 /* 3200 * Add new method to test whether block and inode bitmaps are properly 3201 * initialized. With uninit_bg reading the block from disk is not enough 3202 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3203 */ 3204 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3205 3206 static inline int bitmap_uptodate(struct buffer_head *bh) 3207 { 3208 return (buffer_uptodate(bh) && 3209 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3210 } 3211 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3212 { 3213 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3214 } 3215 3216 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3217 3218 /* For ioend & aio unwritten conversion wait queues */ 3219 #define EXT4_WQ_HASH_SZ 37 3220 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3221 EXT4_WQ_HASH_SZ]) 3222 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3223 3224 extern int ext4_resize_begin(struct super_block *sb); 3225 extern void ext4_resize_end(struct super_block *sb); 3226 3227 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3228 struct ext4_io_end *io_end) 3229 { 3230 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3231 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3232 atomic_inc(&EXT4_I(inode)->i_unwritten); 3233 } 3234 } 3235 3236 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3237 { 3238 struct inode *inode = io_end->inode; 3239 3240 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3241 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3242 /* Wake up anyone waiting on unwritten extent conversion */ 3243 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3244 wake_up_all(ext4_ioend_wq(inode)); 3245 } 3246 } 3247 3248 extern const struct iomap_ops ext4_iomap_ops; 3249 3250 #endif /* __KERNEL__ */ 3251 3252 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3253 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3254 3255 #endif /* _EXT4_H */ 3256