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 #include <linux/fscrypt.h> 44 45 #include <linux/compiler.h> 46 47 /* 48 * The fourth extended filesystem constants/structures 49 */ 50 51 /* 52 * with AGGRESSIVE_CHECK allocator runs consistency checks over 53 * structures. these checks slow things down a lot 54 */ 55 #define AGGRESSIVE_CHECK__ 56 57 /* 58 * with DOUBLE_CHECK defined mballoc creates persistent in-core 59 * bitmaps, maintains and uses them to check for double allocations 60 */ 61 #define DOUBLE_CHECK__ 62 63 /* 64 * Define EXT4FS_DEBUG to produce debug messages 65 */ 66 #undef EXT4FS_DEBUG 67 68 /* 69 * Debug code 70 */ 71 #ifdef EXT4FS_DEBUG 72 #define ext4_debug(f, a...) \ 73 do { \ 74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 75 __FILE__, __LINE__, __func__); \ 76 printk(KERN_DEBUG f, ## a); \ 77 } while (0) 78 #else 79 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 80 #endif 81 82 /* 83 * Turn on EXT_DEBUG to get lots of info about extents operations. 84 */ 85 #define EXT_DEBUG__ 86 #ifdef EXT_DEBUG 87 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 88 #else 89 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 90 #endif 91 92 /* data type for block offset of block group */ 93 typedef int ext4_grpblk_t; 94 95 /* data type for filesystem-wide blocks number */ 96 typedef unsigned long long ext4_fsblk_t; 97 98 /* data type for file logical block number */ 99 typedef __u32 ext4_lblk_t; 100 101 /* data type for block group number */ 102 typedef unsigned int ext4_group_t; 103 104 enum SHIFT_DIRECTION { 105 SHIFT_LEFT = 0, 106 SHIFT_RIGHT, 107 }; 108 109 /* 110 * Flags used in mballoc's allocation_context flags field. 111 * 112 * Also used to show what's going on for debugging purposes when the 113 * flag field is exported via the traceport interface 114 */ 115 116 /* prefer goal again. length */ 117 #define EXT4_MB_HINT_MERGE 0x0001 118 /* blocks already reserved */ 119 #define EXT4_MB_HINT_RESERVED 0x0002 120 /* metadata is being allocated */ 121 #define EXT4_MB_HINT_METADATA 0x0004 122 /* first blocks in the file */ 123 #define EXT4_MB_HINT_FIRST 0x0008 124 /* search for the best chunk */ 125 #define EXT4_MB_HINT_BEST 0x0010 126 /* data is being allocated */ 127 #define EXT4_MB_HINT_DATA 0x0020 128 /* don't preallocate (for tails) */ 129 #define EXT4_MB_HINT_NOPREALLOC 0x0040 130 /* allocate for locality group */ 131 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 132 /* allocate goal blocks or none */ 133 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 134 /* goal is meaningful */ 135 #define EXT4_MB_HINT_TRY_GOAL 0x0200 136 /* blocks already pre-reserved by delayed allocation */ 137 #define EXT4_MB_DELALLOC_RESERVED 0x0400 138 /* We are doing stream allocation */ 139 #define EXT4_MB_STREAM_ALLOC 0x0800 140 /* Use reserved root blocks if needed */ 141 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 142 /* Use blocks from reserved pool */ 143 #define EXT4_MB_USE_RESERVED 0x2000 144 145 struct ext4_allocation_request { 146 /* target inode for block we're allocating */ 147 struct inode *inode; 148 /* how many blocks we want to allocate */ 149 unsigned int len; 150 /* logical block in target inode */ 151 ext4_lblk_t logical; 152 /* the closest logical allocated block to the left */ 153 ext4_lblk_t lleft; 154 /* the closest logical allocated block to the right */ 155 ext4_lblk_t lright; 156 /* phys. target (a hint) */ 157 ext4_fsblk_t goal; 158 /* phys. block for the closest logical allocated block to the left */ 159 ext4_fsblk_t pleft; 160 /* phys. block for the closest logical allocated block to the right */ 161 ext4_fsblk_t pright; 162 /* flags. see above EXT4_MB_HINT_* */ 163 unsigned int flags; 164 }; 165 166 /* 167 * Logical to physical block mapping, used by ext4_map_blocks() 168 * 169 * This structure is used to pass requests into ext4_map_blocks() as 170 * well as to store the information returned by ext4_map_blocks(). It 171 * takes less room on the stack than a struct buffer_head. 172 */ 173 #define EXT4_MAP_NEW (1 << BH_New) 174 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 175 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 176 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 177 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 179 180 struct ext4_map_blocks { 181 ext4_fsblk_t m_pblk; 182 ext4_lblk_t m_lblk; 183 unsigned int m_len; 184 unsigned int m_flags; 185 }; 186 187 /* 188 * Flags for ext4_io_end->flags 189 */ 190 #define EXT4_IO_END_UNWRITTEN 0x0001 191 192 /* 193 * For converting unwritten extents on a work queue. 'handle' is used for 194 * buffered writeback. 195 */ 196 typedef struct ext4_io_end { 197 struct list_head list; /* per-file finished IO list */ 198 handle_t *handle; /* handle reserved for extent 199 * conversion */ 200 struct inode *inode; /* file being written to */ 201 struct bio *bio; /* Linked list of completed 202 * bios covering the extent */ 203 unsigned int flag; /* unwritten or not */ 204 atomic_t count; /* reference counter */ 205 loff_t offset; /* offset in the file */ 206 ssize_t size; /* size of the extent */ 207 } ext4_io_end_t; 208 209 struct ext4_io_submit { 210 struct writeback_control *io_wbc; 211 struct bio *io_bio; 212 ext4_io_end_t *io_end; 213 sector_t io_next_block; 214 }; 215 216 /* 217 * Special inodes numbers 218 */ 219 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 220 #define EXT4_ROOT_INO 2 /* Root inode */ 221 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 222 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 223 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 224 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 225 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 226 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 227 228 /* First non-reserved inode for old ext4 filesystems */ 229 #define EXT4_GOOD_OLD_FIRST_INO 11 230 231 /* 232 * Maximal count of links to a file 233 */ 234 #define EXT4_LINK_MAX 65000 235 236 /* 237 * Macro-instructions used to manage several block sizes 238 */ 239 #define EXT4_MIN_BLOCK_SIZE 1024 240 #define EXT4_MAX_BLOCK_SIZE 65536 241 #define EXT4_MIN_BLOCK_LOG_SIZE 10 242 #define EXT4_MAX_BLOCK_LOG_SIZE 16 243 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 244 #ifdef __KERNEL__ 245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 246 #else 247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 248 #endif 249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 251 EXT4_SB(s)->s_cluster_bits) 252 #ifdef __KERNEL__ 253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 255 #else 256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 257 #endif 258 #ifdef __KERNEL__ 259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 262 #else 263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 264 EXT4_GOOD_OLD_INODE_SIZE : \ 265 (s)->s_inode_size) 266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 267 EXT4_GOOD_OLD_FIRST_INO : \ 268 (s)->s_first_ino) 269 #endif 270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 271 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 273 blkbits)) 274 275 /* Translate a block number to a cluster number */ 276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 277 /* Translate a cluster number to a block number */ 278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 279 /* Translate # of blks to # of clusters */ 280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 281 (sbi)->s_cluster_bits) 282 /* Mask out the low bits to get the starting block of the cluster */ 283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 287 /* Get the cluster offset */ 288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 292 293 /* 294 * Structure of a blocks group descriptor 295 */ 296 struct ext4_group_desc 297 { 298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 300 __le32 bg_inode_table_lo; /* Inodes table block */ 301 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 302 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 303 __le16 bg_used_dirs_count_lo; /* Directories count */ 304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 308 __le16 bg_itable_unused_lo; /* Unused inodes count */ 309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 312 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 320 __u32 bg_reserved; 321 }; 322 323 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 325 sizeof(__le16)) 326 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 328 sizeof(__le16)) 329 330 /* 331 * Structure of a flex block group info 332 */ 333 334 struct flex_groups { 335 atomic64_t free_clusters; 336 atomic_t free_inodes; 337 atomic_t used_dirs; 338 }; 339 340 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 341 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 342 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 343 344 /* 345 * Macro-instructions used to manage group descriptors 346 */ 347 #define EXT4_MIN_DESC_SIZE 32 348 #define EXT4_MIN_DESC_SIZE_64BIT 64 349 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 350 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 351 #ifdef __KERNEL__ 352 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 353 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 355 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 356 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 357 #else 358 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 359 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 360 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 361 #endif 362 363 /* 364 * Constants relative to the data blocks 365 */ 366 #define EXT4_NDIR_BLOCKS 12 367 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 368 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 369 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 370 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 371 372 /* 373 * Inode flags 374 */ 375 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 376 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 377 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 378 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 379 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 380 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 381 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 382 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 383 /* Reserved for compression usage... */ 384 #define EXT4_DIRTY_FL 0x00000100 385 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 386 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 387 /* nb: was previously EXT2_ECOMPR_FL */ 388 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 389 /* End compression flags --- maybe not all used */ 390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 401 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 402 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */ 403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 404 405 #define EXT4_FL_USER_VISIBLE 0x704BDFFF /* User visible flags */ 406 #define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* 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 | EXT4_CASEFOLD_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 | EXT4_CASEFOLD_FL |\ 425 EXT4_PROJINHERIT_FL)) 426 427 /* Flags that are appropriate for non-directories/regular files. */ 428 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 429 430 /* The only flags that should be swapped */ 431 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 432 433 /* Mask out flags that are inappropriate for the given type of inode. */ 434 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 435 { 436 if (S_ISDIR(mode)) 437 return flags; 438 else if (S_ISREG(mode)) 439 return flags & EXT4_REG_FLMASK; 440 else 441 return flags & EXT4_OTHER_FLMASK; 442 } 443 444 /* 445 * Inode flags used for atomic set/get 446 */ 447 enum { 448 EXT4_INODE_SECRM = 0, /* Secure deletion */ 449 EXT4_INODE_UNRM = 1, /* Undelete */ 450 EXT4_INODE_COMPR = 2, /* Compress file */ 451 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 452 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 453 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 454 EXT4_INODE_NODUMP = 6, /* do not dump file */ 455 EXT4_INODE_NOATIME = 7, /* do not update atime */ 456 /* Reserved for compression usage... */ 457 EXT4_INODE_DIRTY = 8, 458 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 459 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 460 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 461 /* End compression flags --- maybe not all used */ 462 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 463 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 464 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 465 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 466 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 467 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 468 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 469 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 470 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 471 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 472 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 473 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 474 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 475 }; 476 477 /* 478 * Since it's pretty easy to mix up bit numbers and hex values, we use a 479 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 480 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 481 * any extra space in the compiled kernel image, otherwise, the build will fail. 482 * It's important that these values are the same, since we are using 483 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 484 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 485 * values found in ext2, ext3 and ext4 filesystems, and of course the values 486 * defined in e2fsprogs. 487 * 488 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 489 */ 490 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 491 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 492 493 static inline void ext4_check_flag_values(void) 494 { 495 CHECK_FLAG_VALUE(SECRM); 496 CHECK_FLAG_VALUE(UNRM); 497 CHECK_FLAG_VALUE(COMPR); 498 CHECK_FLAG_VALUE(SYNC); 499 CHECK_FLAG_VALUE(IMMUTABLE); 500 CHECK_FLAG_VALUE(APPEND); 501 CHECK_FLAG_VALUE(NODUMP); 502 CHECK_FLAG_VALUE(NOATIME); 503 CHECK_FLAG_VALUE(DIRTY); 504 CHECK_FLAG_VALUE(COMPRBLK); 505 CHECK_FLAG_VALUE(NOCOMPR); 506 CHECK_FLAG_VALUE(ENCRYPT); 507 CHECK_FLAG_VALUE(INDEX); 508 CHECK_FLAG_VALUE(IMAGIC); 509 CHECK_FLAG_VALUE(JOURNAL_DATA); 510 CHECK_FLAG_VALUE(NOTAIL); 511 CHECK_FLAG_VALUE(DIRSYNC); 512 CHECK_FLAG_VALUE(TOPDIR); 513 CHECK_FLAG_VALUE(HUGE_FILE); 514 CHECK_FLAG_VALUE(EXTENTS); 515 CHECK_FLAG_VALUE(EA_INODE); 516 CHECK_FLAG_VALUE(EOFBLOCKS); 517 CHECK_FLAG_VALUE(INLINE_DATA); 518 CHECK_FLAG_VALUE(PROJINHERIT); 519 CHECK_FLAG_VALUE(RESERVED); 520 } 521 522 /* Used to pass group descriptor data when online resize is done */ 523 struct ext4_new_group_input { 524 __u32 group; /* Group number for this data */ 525 __u64 block_bitmap; /* Absolute block number of block bitmap */ 526 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 527 __u64 inode_table; /* Absolute block number of inode table start */ 528 __u32 blocks_count; /* Total number of blocks in this group */ 529 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 530 __u16 unused; 531 }; 532 533 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 534 struct compat_ext4_new_group_input { 535 u32 group; 536 compat_u64 block_bitmap; 537 compat_u64 inode_bitmap; 538 compat_u64 inode_table; 539 u32 blocks_count; 540 u16 reserved_blocks; 541 u16 unused; 542 }; 543 #endif 544 545 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 546 struct ext4_new_group_data { 547 __u32 group; 548 __u64 block_bitmap; 549 __u64 inode_bitmap; 550 __u64 inode_table; 551 __u32 blocks_count; 552 __u16 reserved_blocks; 553 __u16 mdata_blocks; 554 __u32 free_clusters_count; 555 }; 556 557 /* Indexes used to index group tables in ext4_new_group_data */ 558 enum { 559 BLOCK_BITMAP = 0, /* block bitmap */ 560 INODE_BITMAP, /* inode bitmap */ 561 INODE_TABLE, /* inode tables */ 562 GROUP_TABLE_COUNT, 563 }; 564 565 /* 566 * Flags used by ext4_map_blocks() 567 */ 568 /* Allocate any needed blocks and/or convert an unwritten 569 extent to be an initialized ext4 */ 570 #define EXT4_GET_BLOCKS_CREATE 0x0001 571 /* Request the creation of an unwritten extent */ 572 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 573 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 574 EXT4_GET_BLOCKS_CREATE) 575 /* Caller is from the delayed allocation writeout path 576 * finally doing the actual allocation of delayed blocks */ 577 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 578 /* caller is from the direct IO path, request to creation of an 579 unwritten extents if not allocated, split the unwritten 580 extent if blocks has been preallocated already*/ 581 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 582 #define EXT4_GET_BLOCKS_CONVERT 0x0010 583 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 584 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 585 /* Convert extent to initialized after IO complete */ 586 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 587 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 588 /* Eventual metadata allocation (due to growing extent tree) 589 * should not fail, so try to use reserved blocks for that.*/ 590 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 591 /* Don't normalize allocation size (used for fallocate) */ 592 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 593 /* Request will not result in inode size update (user for fallocate) */ 594 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 595 /* Convert written extents to unwritten */ 596 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 597 /* Write zeros to newly created written extents */ 598 #define EXT4_GET_BLOCKS_ZERO 0x0200 599 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 600 EXT4_GET_BLOCKS_ZERO) 601 /* Caller will submit data before dropping transaction handle. This 602 * allows jbd2 to avoid submitting data before commit. */ 603 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 604 605 /* 606 * The bit position of these flags must not overlap with any of the 607 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 608 * read_extent_tree_block(), ext4_split_extent_at(), 609 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 610 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 611 * caching the extents when reading from the extent tree while a 612 * truncate or punch hole operation is in progress. 613 */ 614 #define EXT4_EX_NOCACHE 0x40000000 615 #define EXT4_EX_FORCE_CACHE 0x20000000 616 617 /* 618 * Flags used by ext4_free_blocks 619 */ 620 #define EXT4_FREE_BLOCKS_METADATA 0x0001 621 #define EXT4_FREE_BLOCKS_FORGET 0x0002 622 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 623 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 624 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 625 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 626 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 627 628 /* 629 * ioctl commands 630 */ 631 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 632 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 633 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 634 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 635 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 636 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 637 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 638 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 639 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 640 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 641 #define EXT4_IOC_MIGRATE _IO('f', 9) 642 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 643 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 644 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 645 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 646 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 647 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 648 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 649 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 650 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 651 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 652 /* ioctl codes 19--39 are reserved for fscrypt */ 653 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40) 654 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32) 655 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap) 656 657 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 658 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 659 660 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 661 662 /* 663 * Flags for going down operation 664 */ 665 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 666 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 667 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 668 669 /* 670 * Flags returned by EXT4_IOC_GETSTATE 671 * 672 * We only expose to userspace a subset of the state flags in 673 * i_state_flags 674 */ 675 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001 676 #define EXT4_STATE_FLAG_NEW 0x00000002 677 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004 678 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008 679 680 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 681 /* 682 * ioctl commands in 32 bit emulation 683 */ 684 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 685 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 686 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 687 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 688 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 689 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 690 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 691 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 692 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 693 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 694 #endif 695 696 /* 697 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag. 698 * It indicates that the entry in extent status cache is for a hole. 699 */ 700 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000 701 702 /* Max physical block we can address w/o extents */ 703 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 704 705 /* Max logical block we can support */ 706 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF 707 708 /* 709 * Structure of an inode on the disk 710 */ 711 struct ext4_inode { 712 __le16 i_mode; /* File mode */ 713 __le16 i_uid; /* Low 16 bits of Owner Uid */ 714 __le32 i_size_lo; /* Size in bytes */ 715 __le32 i_atime; /* Access time */ 716 __le32 i_ctime; /* Inode Change time */ 717 __le32 i_mtime; /* Modification time */ 718 __le32 i_dtime; /* Deletion Time */ 719 __le16 i_gid; /* Low 16 bits of Group Id */ 720 __le16 i_links_count; /* Links count */ 721 __le32 i_blocks_lo; /* Blocks count */ 722 __le32 i_flags; /* File flags */ 723 union { 724 struct { 725 __le32 l_i_version; 726 } linux1; 727 struct { 728 __u32 h_i_translator; 729 } hurd1; 730 struct { 731 __u32 m_i_reserved1; 732 } masix1; 733 } osd1; /* OS dependent 1 */ 734 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 735 __le32 i_generation; /* File version (for NFS) */ 736 __le32 i_file_acl_lo; /* File ACL */ 737 __le32 i_size_high; 738 __le32 i_obso_faddr; /* Obsoleted fragment address */ 739 union { 740 struct { 741 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 742 __le16 l_i_file_acl_high; 743 __le16 l_i_uid_high; /* these 2 fields */ 744 __le16 l_i_gid_high; /* were reserved2[0] */ 745 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 746 __le16 l_i_reserved; 747 } linux2; 748 struct { 749 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 750 __u16 h_i_mode_high; 751 __u16 h_i_uid_high; 752 __u16 h_i_gid_high; 753 __u32 h_i_author; 754 } hurd2; 755 struct { 756 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 757 __le16 m_i_file_acl_high; 758 __u32 m_i_reserved2[2]; 759 } masix2; 760 } osd2; /* OS dependent 2 */ 761 __le16 i_extra_isize; 762 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 763 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 764 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 765 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 766 __le32 i_crtime; /* File Creation time */ 767 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 768 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 769 __le32 i_projid; /* Project ID */ 770 }; 771 772 struct move_extent { 773 __u32 reserved; /* should be zero */ 774 __u32 donor_fd; /* donor file descriptor */ 775 __u64 orig_start; /* logical start offset in block for orig */ 776 __u64 donor_start; /* logical start offset in block for donor */ 777 __u64 len; /* block length to be moved */ 778 __u64 moved_len; /* moved block length */ 779 }; 780 781 #define EXT4_EPOCH_BITS 2 782 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 783 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 784 785 /* 786 * Extended fields will fit into an inode if the filesystem was formatted 787 * with large inodes (-I 256 or larger) and there are not currently any EAs 788 * consuming all of the available space. For new inodes we always reserve 789 * enough space for the kernel's known extended fields, but for inodes 790 * created with an old kernel this might not have been the case. None of 791 * the extended inode fields is critical for correct filesystem operation. 792 * This macro checks if a certain field fits in the inode. Note that 793 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 794 */ 795 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 796 ((offsetof(typeof(*ext4_inode), field) + \ 797 sizeof((ext4_inode)->field)) \ 798 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 799 (einode)->i_extra_isize)) \ 800 801 /* 802 * We use an encoding that preserves the times for extra epoch "00": 803 * 804 * extra msb of adjust for signed 805 * epoch 32-bit 32-bit tv_sec to 806 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 807 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 808 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 809 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 810 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 811 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 812 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 813 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 814 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 815 * 816 * Note that previous versions of the kernel on 64-bit systems would 817 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 818 * 1970. e2fsck will correct this, assuming that it is run on the 819 * affected filesystem before 2242. 820 */ 821 822 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 823 { 824 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 825 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 826 } 827 828 static inline void ext4_decode_extra_time(struct timespec64 *time, 829 __le32 extra) 830 { 831 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 832 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 833 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 834 } 835 836 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 837 do { \ 838 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 839 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 840 (raw_inode)->xtime ## _extra = \ 841 ext4_encode_extra_time(&(inode)->xtime); \ 842 } \ 843 } while (0) 844 845 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 846 do { \ 847 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 848 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 849 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 850 (raw_inode)->xtime ## _extra = \ 851 ext4_encode_extra_time(&(einode)->xtime); \ 852 } while (0) 853 854 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 855 do { \ 856 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 857 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 858 ext4_decode_extra_time(&(inode)->xtime, \ 859 raw_inode->xtime ## _extra); \ 860 } \ 861 else \ 862 (inode)->xtime.tv_nsec = 0; \ 863 } while (0) 864 865 866 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 867 do { \ 868 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 869 (einode)->xtime.tv_sec = \ 870 (signed)le32_to_cpu((raw_inode)->xtime); \ 871 else \ 872 (einode)->xtime.tv_sec = 0; \ 873 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 874 ext4_decode_extra_time(&(einode)->xtime, \ 875 raw_inode->xtime ## _extra); \ 876 else \ 877 (einode)->xtime.tv_nsec = 0; \ 878 } while (0) 879 880 #define i_disk_version osd1.linux1.l_i_version 881 882 #if defined(__KERNEL__) || defined(__linux__) 883 #define i_reserved1 osd1.linux1.l_i_reserved1 884 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 885 #define i_blocks_high osd2.linux2.l_i_blocks_high 886 #define i_uid_low i_uid 887 #define i_gid_low i_gid 888 #define i_uid_high osd2.linux2.l_i_uid_high 889 #define i_gid_high osd2.linux2.l_i_gid_high 890 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 891 892 #elif defined(__GNU__) 893 894 #define i_translator osd1.hurd1.h_i_translator 895 #define i_uid_high osd2.hurd2.h_i_uid_high 896 #define i_gid_high osd2.hurd2.h_i_gid_high 897 #define i_author osd2.hurd2.h_i_author 898 899 #elif defined(__masix__) 900 901 #define i_reserved1 osd1.masix1.m_i_reserved1 902 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 903 #define i_reserved2 osd2.masix2.m_i_reserved2 904 905 #endif /* defined(__KERNEL__) || defined(__linux__) */ 906 907 #include "extents_status.h" 908 909 /* 910 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 911 * 912 * These are needed to avoid lockdep false positives when we need to 913 * allocate blocks to the quota inode during ext4_map_blocks(), while 914 * holding i_data_sem for a normal (non-quota) inode. Since we don't 915 * do quota tracking for the quota inode, this avoids deadlock (as 916 * well as infinite recursion, since it isn't turtles all the way 917 * down...) 918 * 919 * I_DATA_SEM_NORMAL - Used for most inodes 920 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 921 * where the second inode has larger inode number 922 * than the first 923 * I_DATA_SEM_QUOTA - Used for quota inodes only 924 */ 925 enum { 926 I_DATA_SEM_NORMAL = 0, 927 I_DATA_SEM_OTHER, 928 I_DATA_SEM_QUOTA, 929 }; 930 931 932 /* 933 * fourth extended file system inode data in memory 934 */ 935 struct ext4_inode_info { 936 __le32 i_data[15]; /* unconverted */ 937 __u32 i_dtime; 938 ext4_fsblk_t i_file_acl; 939 940 /* 941 * i_block_group is the number of the block group which contains 942 * this file's inode. Constant across the lifetime of the inode, 943 * it is used for making block allocation decisions - we try to 944 * place a file's data blocks near its inode block, and new inodes 945 * near to their parent directory's inode. 946 */ 947 ext4_group_t i_block_group; 948 ext4_lblk_t i_dir_start_lookup; 949 #if (BITS_PER_LONG < 64) 950 unsigned long i_state_flags; /* Dynamic state flags */ 951 #endif 952 unsigned long i_flags; 953 954 /* 955 * Extended attributes can be read independently of the main file 956 * data. Taking i_mutex even when reading would cause contention 957 * between readers of EAs and writers of regular file data, so 958 * instead we synchronize on xattr_sem when reading or changing 959 * EAs. 960 */ 961 struct rw_semaphore xattr_sem; 962 963 struct list_head i_orphan; /* unlinked but open inodes */ 964 965 /* 966 * i_disksize keeps track of what the inode size is ON DISK, not 967 * in memory. During truncate, i_size is set to the new size by 968 * the VFS prior to calling ext4_truncate(), but the filesystem won't 969 * set i_disksize to 0 until the truncate is actually under way. 970 * 971 * The intent is that i_disksize always represents the blocks which 972 * are used by this file. This allows recovery to restart truncate 973 * on orphans if we crash during truncate. We actually write i_disksize 974 * into the on-disk inode when writing inodes out, instead of i_size. 975 * 976 * The only time when i_disksize and i_size may be different is when 977 * a truncate is in progress. The only things which change i_disksize 978 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 979 */ 980 loff_t i_disksize; 981 982 /* 983 * i_data_sem is for serialising ext4_truncate() against 984 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 985 * data tree are chopped off during truncate. We can't do that in 986 * ext4 because whenever we perform intermediate commits during 987 * truncate, the inode and all the metadata blocks *must* be in a 988 * consistent state which allows truncation of the orphans to restart 989 * during recovery. Hence we must fix the get_block-vs-truncate race 990 * by other means, so we have i_data_sem. 991 */ 992 struct rw_semaphore i_data_sem; 993 /* 994 * i_mmap_sem is for serializing page faults with truncate / punch hole 995 * operations. We have to make sure that new page cannot be faulted in 996 * a section of the inode that is being punched. We cannot easily use 997 * i_data_sem for this since we need protection for the whole punch 998 * operation and i_data_sem ranks below transaction start so we have 999 * to occasionally drop it. 1000 */ 1001 struct rw_semaphore i_mmap_sem; 1002 struct inode vfs_inode; 1003 struct jbd2_inode *jinode; 1004 1005 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1006 1007 /* 1008 * File creation time. Its function is same as that of 1009 * struct timespec64 i_{a,c,m}time in the generic inode. 1010 */ 1011 struct timespec64 i_crtime; 1012 1013 /* mballoc */ 1014 struct list_head i_prealloc_list; 1015 spinlock_t i_prealloc_lock; 1016 1017 /* extents status tree */ 1018 struct ext4_es_tree i_es_tree; 1019 rwlock_t i_es_lock; 1020 struct list_head i_es_list; 1021 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1022 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1023 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1024 extents to shrink. Protected by 1025 i_es_lock */ 1026 1027 /* ialloc */ 1028 ext4_group_t i_last_alloc_group; 1029 1030 /* allocation reservation info for delalloc */ 1031 /* In case of bigalloc, this refer to clusters rather than blocks */ 1032 unsigned int i_reserved_data_blocks; 1033 ext4_lblk_t i_da_metadata_calc_last_lblock; 1034 int i_da_metadata_calc_len; 1035 1036 /* pending cluster reservations for bigalloc file systems */ 1037 struct ext4_pending_tree i_pending_tree; 1038 1039 /* on-disk additional length */ 1040 __u16 i_extra_isize; 1041 1042 /* Indicate the inline data space. */ 1043 u16 i_inline_off; 1044 u16 i_inline_size; 1045 1046 #ifdef CONFIG_QUOTA 1047 /* quota space reservation, managed internally by quota code */ 1048 qsize_t i_reserved_quota; 1049 #endif 1050 1051 /* Lock protecting lists below */ 1052 spinlock_t i_completed_io_lock; 1053 /* 1054 * Completed IOs that need unwritten extents handling and have 1055 * transaction reserved 1056 */ 1057 struct list_head i_rsv_conversion_list; 1058 struct work_struct i_rsv_conversion_work; 1059 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1060 1061 spinlock_t i_block_reservation_lock; 1062 1063 /* 1064 * Transactions that contain inode's metadata needed to complete 1065 * fsync and fdatasync, respectively. 1066 */ 1067 tid_t i_sync_tid; 1068 tid_t i_datasync_tid; 1069 1070 #ifdef CONFIG_QUOTA 1071 struct dquot *i_dquot[MAXQUOTAS]; 1072 #endif 1073 1074 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1075 __u32 i_csum_seed; 1076 1077 kprojid_t i_projid; 1078 }; 1079 1080 /* 1081 * File system states 1082 */ 1083 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1084 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1085 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1086 1087 /* 1088 * Misc. filesystem flags 1089 */ 1090 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1091 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1092 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1093 1094 /* 1095 * Mount flags set via mount options or defaults 1096 */ 1097 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1098 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1099 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1100 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1101 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1102 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1103 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1104 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1105 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1106 #ifdef CONFIG_FS_DAX 1107 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1108 #else 1109 #define EXT4_MOUNT_DAX 0 1110 #endif 1111 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1112 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1113 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1114 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1115 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1116 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1117 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1118 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1119 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1120 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1121 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1122 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1123 * enable enforcement for hidden 1124 * quota files */ 1125 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1126 * enforcement for hidden quota 1127 * files */ 1128 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1129 * enforcement */ 1130 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1131 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1132 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1133 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1134 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1135 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1136 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1137 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1138 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1139 1140 /* 1141 * Mount flags set either automatically (could not be set by mount option) 1142 * based on per file system feature or property or in special cases such as 1143 * distinguishing between explicit mount option definition and default. 1144 */ 1145 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1146 specified delalloc */ 1147 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1148 size of blocksize * 8 1149 blocks */ 1150 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1151 file systems */ 1152 1153 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1154 specified journal checksum */ 1155 1156 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1157 ~EXT4_MOUNT_##opt 1158 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1159 EXT4_MOUNT_##opt 1160 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1161 EXT4_MOUNT_##opt) 1162 1163 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1164 ~EXT4_MOUNT2_##opt 1165 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1166 EXT4_MOUNT2_##opt 1167 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1168 EXT4_MOUNT2_##opt) 1169 1170 #define ext4_test_and_set_bit __test_and_set_bit_le 1171 #define ext4_set_bit __set_bit_le 1172 #define ext4_set_bit_atomic ext2_set_bit_atomic 1173 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1174 #define ext4_clear_bit __clear_bit_le 1175 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1176 #define ext4_test_bit test_bit_le 1177 #define ext4_find_next_zero_bit find_next_zero_bit_le 1178 #define ext4_find_next_bit find_next_bit_le 1179 1180 extern void ext4_set_bits(void *bm, int cur, int len); 1181 1182 /* 1183 * Maximal mount counts between two filesystem checks 1184 */ 1185 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1186 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1187 1188 /* 1189 * Behaviour when detecting errors 1190 */ 1191 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1192 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1193 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1194 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1195 1196 /* Metadata checksum algorithm codes */ 1197 #define EXT4_CRC32C_CHKSUM 1 1198 1199 /* 1200 * Structure of the super block 1201 */ 1202 struct ext4_super_block { 1203 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1204 __le32 s_blocks_count_lo; /* Blocks count */ 1205 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1206 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1207 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1208 __le32 s_first_data_block; /* First Data Block */ 1209 __le32 s_log_block_size; /* Block size */ 1210 __le32 s_log_cluster_size; /* Allocation cluster size */ 1211 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1212 __le32 s_clusters_per_group; /* # Clusters per group */ 1213 __le32 s_inodes_per_group; /* # Inodes per group */ 1214 __le32 s_mtime; /* Mount time */ 1215 /*30*/ __le32 s_wtime; /* Write time */ 1216 __le16 s_mnt_count; /* Mount count */ 1217 __le16 s_max_mnt_count; /* Maximal mount count */ 1218 __le16 s_magic; /* Magic signature */ 1219 __le16 s_state; /* File system state */ 1220 __le16 s_errors; /* Behaviour when detecting errors */ 1221 __le16 s_minor_rev_level; /* minor revision level */ 1222 /*40*/ __le32 s_lastcheck; /* time of last check */ 1223 __le32 s_checkinterval; /* max. time between checks */ 1224 __le32 s_creator_os; /* OS */ 1225 __le32 s_rev_level; /* Revision level */ 1226 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1227 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1228 /* 1229 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1230 * 1231 * Note: the difference between the compatible feature set and 1232 * the incompatible feature set is that if there is a bit set 1233 * in the incompatible feature set that the kernel doesn't 1234 * know about, it should refuse to mount the filesystem. 1235 * 1236 * e2fsck's requirements are more strict; if it doesn't know 1237 * about a feature in either the compatible or incompatible 1238 * feature set, it must abort and not try to meddle with 1239 * things it doesn't understand... 1240 */ 1241 __le32 s_first_ino; /* First non-reserved inode */ 1242 __le16 s_inode_size; /* size of inode structure */ 1243 __le16 s_block_group_nr; /* block group # of this superblock */ 1244 __le32 s_feature_compat; /* compatible feature set */ 1245 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1246 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1247 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1248 /*78*/ char s_volume_name[16]; /* volume name */ 1249 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1250 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1251 /* 1252 * Performance hints. Directory preallocation should only 1253 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1254 */ 1255 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1256 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1257 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1258 /* 1259 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1260 */ 1261 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1262 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1263 __le32 s_journal_dev; /* device number of journal file */ 1264 __le32 s_last_orphan; /* start of list of inodes to delete */ 1265 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1266 __u8 s_def_hash_version; /* Default hash version to use */ 1267 __u8 s_jnl_backup_type; 1268 __le16 s_desc_size; /* size of group descriptor */ 1269 /*100*/ __le32 s_default_mount_opts; 1270 __le32 s_first_meta_bg; /* First metablock block group */ 1271 __le32 s_mkfs_time; /* When the filesystem was created */ 1272 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1273 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1274 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1275 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1276 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1277 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1278 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1279 __le32 s_flags; /* Miscellaneous flags */ 1280 __le16 s_raid_stride; /* RAID stride */ 1281 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1282 __le64 s_mmp_block; /* Block for multi-mount protection */ 1283 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1284 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1285 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1286 __u8 s_encryption_level; /* versioning level for encryption */ 1287 __u8 s_reserved_pad; /* Padding to next 32bits */ 1288 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1289 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1290 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1291 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1292 snapshot's future use */ 1293 __le32 s_snapshot_list; /* inode number of the head of the 1294 on-disk snapshot list */ 1295 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1296 __le32 s_error_count; /* number of fs errors */ 1297 __le32 s_first_error_time; /* first time an error happened */ 1298 __le32 s_first_error_ino; /* inode involved in first error */ 1299 __le64 s_first_error_block; /* block involved of first error */ 1300 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1301 __le32 s_first_error_line; /* line number where error happened */ 1302 __le32 s_last_error_time; /* most recent time of an error */ 1303 __le32 s_last_error_ino; /* inode involved in last error */ 1304 __le32 s_last_error_line; /* line number where error happened */ 1305 __le64 s_last_error_block; /* block involved of last error */ 1306 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1307 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1308 __u8 s_mount_opts[64]; 1309 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1310 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1311 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1312 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1313 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1314 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1315 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1316 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1317 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1318 __u8 s_wtime_hi; 1319 __u8 s_mtime_hi; 1320 __u8 s_mkfs_time_hi; 1321 __u8 s_lastcheck_hi; 1322 __u8 s_first_error_time_hi; 1323 __u8 s_last_error_time_hi; 1324 __u8 s_pad[2]; 1325 __le16 s_encoding; /* Filename charset encoding */ 1326 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1327 __le32 s_reserved[95]; /* Padding to the end of the block */ 1328 __le32 s_checksum; /* crc32c(superblock) */ 1329 }; 1330 1331 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1332 1333 #ifdef __KERNEL__ 1334 1335 /* 1336 * run-time mount flags 1337 */ 1338 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1339 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1340 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1341 1342 #ifdef CONFIG_FS_ENCRYPTION 1343 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1344 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1345 #else 1346 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1347 #endif 1348 1349 /* Number of quota types we support */ 1350 #define EXT4_MAXQUOTAS 3 1351 1352 #define EXT4_ENC_UTF8_12_1 1 1353 1354 /* 1355 * Flags for ext4_sb_info.s_encoding_flags. 1356 */ 1357 #define EXT4_ENC_STRICT_MODE_FL (1 << 0) 1358 1359 #define ext4_has_strict_mode(sbi) \ 1360 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL) 1361 1362 /* 1363 * fourth extended-fs super-block data in memory 1364 */ 1365 struct ext4_sb_info { 1366 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1367 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1368 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1369 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1370 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1371 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1372 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1373 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1374 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1375 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1376 unsigned long s_overhead; /* # of fs overhead clusters */ 1377 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1378 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1379 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1380 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1381 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1382 struct buffer_head **s_group_desc; 1383 unsigned int s_mount_opt; 1384 unsigned int s_mount_opt2; 1385 unsigned int s_mount_flags; 1386 unsigned int s_def_mount_opt; 1387 ext4_fsblk_t s_sb_block; 1388 atomic64_t s_resv_clusters; 1389 kuid_t s_resuid; 1390 kgid_t s_resgid; 1391 unsigned short s_mount_state; 1392 unsigned short s_pad; 1393 int s_addr_per_block_bits; 1394 int s_desc_per_block_bits; 1395 int s_inode_size; 1396 int s_first_ino; 1397 unsigned int s_inode_readahead_blks; 1398 unsigned int s_inode_goal; 1399 u32 s_hash_seed[4]; 1400 int s_def_hash_version; 1401 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1402 struct percpu_counter s_freeclusters_counter; 1403 struct percpu_counter s_freeinodes_counter; 1404 struct percpu_counter s_dirs_counter; 1405 struct percpu_counter s_dirtyclusters_counter; 1406 struct blockgroup_lock *s_blockgroup_lock; 1407 struct proc_dir_entry *s_proc; 1408 struct kobject s_kobj; 1409 struct completion s_kobj_unregister; 1410 struct super_block *s_sb; 1411 #ifdef CONFIG_UNICODE 1412 struct unicode_map *s_encoding; 1413 __u16 s_encoding_flags; 1414 #endif 1415 1416 /* Journaling */ 1417 struct journal_s *s_journal; 1418 struct list_head s_orphan; 1419 struct mutex s_orphan_lock; 1420 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1421 unsigned long s_commit_interval; 1422 u32 s_max_batch_time; 1423 u32 s_min_batch_time; 1424 struct block_device *journal_bdev; 1425 #ifdef CONFIG_QUOTA 1426 /* Names of quota files with journalled quota */ 1427 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1428 int s_jquota_fmt; /* Format of quota to use */ 1429 #endif 1430 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1431 struct rb_root system_blks; 1432 1433 #ifdef EXTENTS_STATS 1434 /* ext4 extents stats */ 1435 unsigned long s_ext_min; 1436 unsigned long s_ext_max; 1437 unsigned long s_depth_max; 1438 spinlock_t s_ext_stats_lock; 1439 unsigned long s_ext_blocks; 1440 unsigned long s_ext_extents; 1441 #endif 1442 1443 /* for buddy allocator */ 1444 struct ext4_group_info ***s_group_info; 1445 struct inode *s_buddy_cache; 1446 spinlock_t s_md_lock; 1447 unsigned short *s_mb_offsets; 1448 unsigned int *s_mb_maxs; 1449 unsigned int s_group_info_size; 1450 unsigned int s_mb_free_pending; 1451 struct list_head s_freed_data_list; /* List of blocks to be freed 1452 after commit completed */ 1453 1454 /* tunables */ 1455 unsigned long s_stripe; 1456 unsigned int s_mb_stream_request; 1457 unsigned int s_mb_max_to_scan; 1458 unsigned int s_mb_min_to_scan; 1459 unsigned int s_mb_stats; 1460 unsigned int s_mb_order2_reqs; 1461 unsigned int s_mb_group_prealloc; 1462 unsigned int s_max_dir_size_kb; 1463 /* where last allocation was done - for stream allocation */ 1464 unsigned long s_mb_last_group; 1465 unsigned long s_mb_last_start; 1466 1467 /* stats for buddy allocator */ 1468 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1469 atomic_t s_bal_success; /* we found long enough chunks */ 1470 atomic_t s_bal_allocated; /* in blocks */ 1471 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1472 atomic_t s_bal_goals; /* goal hits */ 1473 atomic_t s_bal_breaks; /* too long searches */ 1474 atomic_t s_bal_2orders; /* 2^order hits */ 1475 spinlock_t s_bal_lock; 1476 unsigned long s_mb_buddies_generated; 1477 unsigned long long s_mb_generation_time; 1478 atomic_t s_mb_lost_chunks; 1479 atomic_t s_mb_preallocated; 1480 atomic_t s_mb_discarded; 1481 atomic_t s_lock_busy; 1482 1483 /* locality groups */ 1484 struct ext4_locality_group __percpu *s_locality_groups; 1485 1486 /* for write statistics */ 1487 unsigned long s_sectors_written_start; 1488 u64 s_kbytes_written; 1489 1490 /* the size of zero-out chunk */ 1491 unsigned int s_extent_max_zeroout_kb; 1492 1493 unsigned int s_log_groups_per_flex; 1494 struct flex_groups *s_flex_groups; 1495 ext4_group_t s_flex_groups_allocated; 1496 1497 /* workqueue for reserved extent conversions (buffered io) */ 1498 struct workqueue_struct *rsv_conversion_wq; 1499 1500 /* timer for periodic error stats printing */ 1501 struct timer_list s_err_report; 1502 1503 /* Lazy inode table initialization info */ 1504 struct ext4_li_request *s_li_request; 1505 /* Wait multiplier for lazy initialization thread */ 1506 unsigned int s_li_wait_mult; 1507 1508 /* Kernel thread for multiple mount protection */ 1509 struct task_struct *s_mmp_tsk; 1510 1511 /* record the last minlen when FITRIM is called. */ 1512 atomic_t s_last_trim_minblks; 1513 1514 /* Reference to checksum algorithm driver via cryptoapi */ 1515 struct crypto_shash *s_chksum_driver; 1516 1517 /* Precomputed FS UUID checksum for seeding other checksums */ 1518 __u32 s_csum_seed; 1519 1520 /* Reclaim extents from extent status tree */ 1521 struct shrinker s_es_shrinker; 1522 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1523 long s_es_nr_inode; 1524 struct ext4_es_stats s_es_stats; 1525 struct mb_cache *s_ea_block_cache; 1526 struct mb_cache *s_ea_inode_cache; 1527 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1528 1529 /* Ratelimit ext4 messages. */ 1530 struct ratelimit_state s_err_ratelimit_state; 1531 struct ratelimit_state s_warning_ratelimit_state; 1532 struct ratelimit_state s_msg_ratelimit_state; 1533 1534 /* Barrier between changing inodes' journal flags and writepages ops. */ 1535 struct percpu_rw_semaphore s_journal_flag_rwsem; 1536 struct dax_device *s_daxdev; 1537 }; 1538 1539 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1540 { 1541 return sb->s_fs_info; 1542 } 1543 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1544 { 1545 return container_of(inode, struct ext4_inode_info, vfs_inode); 1546 } 1547 1548 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1549 { 1550 return ino == EXT4_ROOT_INO || 1551 (ino >= EXT4_FIRST_INO(sb) && 1552 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1553 } 1554 1555 /* 1556 * Inode dynamic state flags 1557 */ 1558 enum { 1559 EXT4_STATE_JDATA, /* journaled data exists */ 1560 EXT4_STATE_NEW, /* inode is newly created */ 1561 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1562 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1563 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1564 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1565 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1566 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1567 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1568 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1569 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1570 }; 1571 1572 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1573 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1574 { \ 1575 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1576 } \ 1577 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1578 { \ 1579 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1580 } \ 1581 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1582 { \ 1583 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1584 } 1585 1586 /* Add these declarations here only so that these functions can be 1587 * found by name. Otherwise, they are very hard to locate. */ 1588 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1589 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1590 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1591 EXT4_INODE_BIT_FNS(flag, flags, 0) 1592 1593 /* Add these declarations here only so that these functions can be 1594 * found by name. Otherwise, they are very hard to locate. */ 1595 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1596 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1597 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1598 #if (BITS_PER_LONG < 64) 1599 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1600 1601 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1602 { 1603 (ei)->i_state_flags = 0; 1604 } 1605 #else 1606 EXT4_INODE_BIT_FNS(state, flags, 32) 1607 1608 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1609 { 1610 /* We depend on the fact that callers will set i_flags */ 1611 } 1612 #endif 1613 #else 1614 /* Assume that user mode programs are passing in an ext4fs superblock, not 1615 * a kernel struct super_block. This will allow us to call the feature-test 1616 * macros from user land. */ 1617 #define EXT4_SB(sb) (sb) 1618 #endif 1619 1620 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1621 1622 /* 1623 * Codes for operating systems 1624 */ 1625 #define EXT4_OS_LINUX 0 1626 #define EXT4_OS_HURD 1 1627 #define EXT4_OS_MASIX 2 1628 #define EXT4_OS_FREEBSD 3 1629 #define EXT4_OS_LITES 4 1630 1631 /* 1632 * Revision levels 1633 */ 1634 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1635 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1636 1637 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1638 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1639 1640 #define EXT4_GOOD_OLD_INODE_SIZE 128 1641 1642 /* 1643 * Feature set definitions 1644 */ 1645 1646 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1647 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1648 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1649 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1650 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1651 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1652 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1653 1654 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1655 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1656 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1657 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1658 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1659 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1660 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1661 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1662 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1663 /* 1664 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1665 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1666 * all other data structures' checksums. However, the METADATA_CSUM and 1667 * GDT_CSUM bits are mutually exclusive. 1668 */ 1669 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1670 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1671 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1672 1673 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1674 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1675 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1676 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1677 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1678 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1679 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1680 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1681 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1682 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1683 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1684 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1685 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1686 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1687 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1688 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 1689 1690 extern void ext4_update_dynamic_rev(struct super_block *sb); 1691 1692 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1693 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1694 { \ 1695 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1696 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1697 } \ 1698 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1699 { \ 1700 ext4_update_dynamic_rev(sb); \ 1701 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1702 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1703 } \ 1704 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1705 { \ 1706 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1707 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1708 } 1709 1710 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1711 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1712 { \ 1713 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1714 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1715 } \ 1716 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1717 { \ 1718 ext4_update_dynamic_rev(sb); \ 1719 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1720 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1721 } \ 1722 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1723 { \ 1724 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1725 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1726 } 1727 1728 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1729 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1730 { \ 1731 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1732 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1733 } \ 1734 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1735 { \ 1736 ext4_update_dynamic_rev(sb); \ 1737 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1738 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1739 } \ 1740 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1741 { \ 1742 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1743 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1744 } 1745 1746 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1747 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1748 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1749 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1750 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1751 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1752 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1753 1754 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1755 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1756 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1757 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1758 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1759 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1760 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1761 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1762 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1763 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1764 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1765 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1766 1767 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1768 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1769 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1770 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1771 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1772 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1773 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1774 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1775 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1776 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1777 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1778 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1779 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1780 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1781 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1782 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 1783 1784 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1785 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1786 EXT4_FEATURE_INCOMPAT_META_BG) 1787 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1788 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1789 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1790 1791 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1792 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1793 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1794 EXT4_FEATURE_INCOMPAT_META_BG) 1795 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1796 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1797 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1798 1799 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1800 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1801 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1802 EXT4_FEATURE_INCOMPAT_META_BG| \ 1803 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1804 EXT4_FEATURE_INCOMPAT_64BIT| \ 1805 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1806 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1807 EXT4_FEATURE_INCOMPAT_MMP | \ 1808 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1809 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1810 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 1811 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1812 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1813 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1814 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1815 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1816 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1817 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1818 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1819 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1820 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1821 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1822 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1823 EXT4_FEATURE_RO_COMPAT_PROJECT) 1824 1825 #define EXTN_FEATURE_FUNCS(ver) \ 1826 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1827 { \ 1828 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1829 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1830 } \ 1831 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1832 { \ 1833 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1834 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1835 } \ 1836 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1837 { \ 1838 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1839 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1840 } 1841 1842 EXTN_FEATURE_FUNCS(2) 1843 EXTN_FEATURE_FUNCS(3) 1844 EXTN_FEATURE_FUNCS(4) 1845 1846 static inline bool ext4_has_compat_features(struct super_block *sb) 1847 { 1848 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1849 } 1850 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1851 { 1852 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1853 } 1854 static inline bool ext4_has_incompat_features(struct super_block *sb) 1855 { 1856 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1857 } 1858 1859 /* 1860 * Superblock flags 1861 */ 1862 #define EXT4_FLAGS_RESIZING 0 1863 #define EXT4_FLAGS_SHUTDOWN 1 1864 1865 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1866 { 1867 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1868 } 1869 1870 1871 /* 1872 * Default values for user and/or group using reserved blocks 1873 */ 1874 #define EXT4_DEF_RESUID 0 1875 #define EXT4_DEF_RESGID 0 1876 1877 /* 1878 * Default project ID 1879 */ 1880 #define EXT4_DEF_PROJID 0 1881 1882 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1883 1884 /* 1885 * Default mount options 1886 */ 1887 #define EXT4_DEFM_DEBUG 0x0001 1888 #define EXT4_DEFM_BSDGROUPS 0x0002 1889 #define EXT4_DEFM_XATTR_USER 0x0004 1890 #define EXT4_DEFM_ACL 0x0008 1891 #define EXT4_DEFM_UID16 0x0010 1892 #define EXT4_DEFM_JMODE 0x0060 1893 #define EXT4_DEFM_JMODE_DATA 0x0020 1894 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1895 #define EXT4_DEFM_JMODE_WBACK 0x0060 1896 #define EXT4_DEFM_NOBARRIER 0x0100 1897 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1898 #define EXT4_DEFM_DISCARD 0x0400 1899 #define EXT4_DEFM_NODELALLOC 0x0800 1900 1901 /* 1902 * Default journal batch times 1903 */ 1904 #define EXT4_DEF_MIN_BATCH_TIME 0 1905 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1906 1907 /* 1908 * Minimum number of groups in a flexgroup before we separate out 1909 * directories into the first block group of a flexgroup 1910 */ 1911 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1912 1913 /* 1914 * Structure of a directory entry 1915 */ 1916 #define EXT4_NAME_LEN 255 1917 1918 struct ext4_dir_entry { 1919 __le32 inode; /* Inode number */ 1920 __le16 rec_len; /* Directory entry length */ 1921 __le16 name_len; /* Name length */ 1922 char name[EXT4_NAME_LEN]; /* File name */ 1923 }; 1924 1925 /* 1926 * The new version of the directory entry. Since EXT4 structures are 1927 * stored in intel byte order, and the name_len field could never be 1928 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1929 * file_type field. 1930 */ 1931 struct ext4_dir_entry_2 { 1932 __le32 inode; /* Inode number */ 1933 __le16 rec_len; /* Directory entry length */ 1934 __u8 name_len; /* Name length */ 1935 __u8 file_type; 1936 char name[EXT4_NAME_LEN]; /* File name */ 1937 }; 1938 1939 /* 1940 * This is a bogus directory entry at the end of each leaf block that 1941 * records checksums. 1942 */ 1943 struct ext4_dir_entry_tail { 1944 __le32 det_reserved_zero1; /* Pretend to be unused */ 1945 __le16 det_rec_len; /* 12 */ 1946 __u8 det_reserved_zero2; /* Zero name length */ 1947 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1948 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1949 }; 1950 1951 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1952 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1953 ((blocksize) - \ 1954 sizeof(struct ext4_dir_entry_tail)))) 1955 1956 /* 1957 * Ext4 directory file types. Only the low 3 bits are used. The 1958 * other bits are reserved for now. 1959 */ 1960 #define EXT4_FT_UNKNOWN 0 1961 #define EXT4_FT_REG_FILE 1 1962 #define EXT4_FT_DIR 2 1963 #define EXT4_FT_CHRDEV 3 1964 #define EXT4_FT_BLKDEV 4 1965 #define EXT4_FT_FIFO 5 1966 #define EXT4_FT_SOCK 6 1967 #define EXT4_FT_SYMLINK 7 1968 1969 #define EXT4_FT_MAX 8 1970 1971 #define EXT4_FT_DIR_CSUM 0xDE 1972 1973 /* 1974 * EXT4_DIR_PAD defines the directory entries boundaries 1975 * 1976 * NOTE: It must be a multiple of 4 1977 */ 1978 #define EXT4_DIR_PAD 4 1979 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1980 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1981 ~EXT4_DIR_ROUND) 1982 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1983 1984 /* 1985 * If we ever get support for fs block sizes > page_size, we'll need 1986 * to remove the #if statements in the next two functions... 1987 */ 1988 static inline unsigned int 1989 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1990 { 1991 unsigned len = le16_to_cpu(dlen); 1992 1993 #if (PAGE_SIZE >= 65536) 1994 if (len == EXT4_MAX_REC_LEN || len == 0) 1995 return blocksize; 1996 return (len & 65532) | ((len & 3) << 16); 1997 #else 1998 return len; 1999 #endif 2000 } 2001 2002 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2003 { 2004 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2005 BUG(); 2006 #if (PAGE_SIZE >= 65536) 2007 if (len < 65536) 2008 return cpu_to_le16(len); 2009 if (len == blocksize) { 2010 if (blocksize == 65536) 2011 return cpu_to_le16(EXT4_MAX_REC_LEN); 2012 else 2013 return cpu_to_le16(0); 2014 } 2015 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2016 #else 2017 return cpu_to_le16(len); 2018 #endif 2019 } 2020 2021 /* 2022 * Hash Tree Directory indexing 2023 * (c) Daniel Phillips, 2001 2024 */ 2025 2026 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2027 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2028 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2029 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2030 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2031 2032 /* Legal values for the dx_root hash_version field: */ 2033 2034 #define DX_HASH_LEGACY 0 2035 #define DX_HASH_HALF_MD4 1 2036 #define DX_HASH_TEA 2 2037 #define DX_HASH_LEGACY_UNSIGNED 3 2038 #define DX_HASH_HALF_MD4_UNSIGNED 4 2039 #define DX_HASH_TEA_UNSIGNED 5 2040 2041 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2042 const void *address, unsigned int length) 2043 { 2044 struct { 2045 struct shash_desc shash; 2046 char ctx[4]; 2047 } desc; 2048 2049 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2050 2051 desc.shash.tfm = sbi->s_chksum_driver; 2052 *(u32 *)desc.ctx = crc; 2053 2054 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2055 2056 return *(u32 *)desc.ctx; 2057 } 2058 2059 #ifdef __KERNEL__ 2060 2061 /* hash info structure used by the directory hash */ 2062 struct dx_hash_info 2063 { 2064 u32 hash; 2065 u32 minor_hash; 2066 int hash_version; 2067 u32 *seed; 2068 }; 2069 2070 2071 /* 32 and 64 bit signed EOF for dx directories */ 2072 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2073 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2074 2075 2076 /* 2077 * Control parameters used by ext4_htree_next_block 2078 */ 2079 #define HASH_NB_ALWAYS 1 2080 2081 struct ext4_filename { 2082 const struct qstr *usr_fname; 2083 struct fscrypt_str disk_name; 2084 struct dx_hash_info hinfo; 2085 #ifdef CONFIG_FS_ENCRYPTION 2086 struct fscrypt_str crypto_buf; 2087 #endif 2088 #ifdef CONFIG_UNICODE 2089 struct fscrypt_str cf_name; 2090 #endif 2091 }; 2092 2093 #define fname_name(p) ((p)->disk_name.name) 2094 #define fname_len(p) ((p)->disk_name.len) 2095 2096 /* 2097 * Describe an inode's exact location on disk and in memory 2098 */ 2099 struct ext4_iloc 2100 { 2101 struct buffer_head *bh; 2102 unsigned long offset; 2103 ext4_group_t block_group; 2104 }; 2105 2106 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2107 { 2108 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2109 } 2110 2111 static inline bool ext4_is_quota_file(struct inode *inode) 2112 { 2113 return IS_NOQUOTA(inode) && 2114 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2115 } 2116 2117 /* 2118 * This structure is stuffed into the struct file's private_data field 2119 * for directories. It is where we put information so that we can do 2120 * readdir operations in hash tree order. 2121 */ 2122 struct dir_private_info { 2123 struct rb_root root; 2124 struct rb_node *curr_node; 2125 struct fname *extra_fname; 2126 loff_t last_pos; 2127 __u32 curr_hash; 2128 __u32 curr_minor_hash; 2129 __u32 next_hash; 2130 }; 2131 2132 /* calculate the first block number of the group */ 2133 static inline ext4_fsblk_t 2134 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2135 { 2136 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2137 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2138 } 2139 2140 /* 2141 * Special error return code only used by dx_probe() and its callers. 2142 */ 2143 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2144 2145 /* htree levels for ext4 */ 2146 #define EXT4_HTREE_LEVEL_COMPAT 2 2147 #define EXT4_HTREE_LEVEL 3 2148 2149 static inline int ext4_dir_htree_level(struct super_block *sb) 2150 { 2151 return ext4_has_feature_largedir(sb) ? 2152 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2153 } 2154 2155 /* 2156 * Timeout and state flag for lazy initialization inode thread. 2157 */ 2158 #define EXT4_DEF_LI_WAIT_MULT 10 2159 #define EXT4_DEF_LI_MAX_START_DELAY 5 2160 #define EXT4_LAZYINIT_QUIT 0x0001 2161 #define EXT4_LAZYINIT_RUNNING 0x0002 2162 2163 /* 2164 * Lazy inode table initialization info 2165 */ 2166 struct ext4_lazy_init { 2167 unsigned long li_state; 2168 struct list_head li_request_list; 2169 struct mutex li_list_mtx; 2170 }; 2171 2172 struct ext4_li_request { 2173 struct super_block *lr_super; 2174 struct ext4_sb_info *lr_sbi; 2175 ext4_group_t lr_next_group; 2176 struct list_head lr_request; 2177 unsigned long lr_next_sched; 2178 unsigned long lr_timeout; 2179 }; 2180 2181 struct ext4_features { 2182 struct kobject f_kobj; 2183 struct completion f_kobj_unregister; 2184 }; 2185 2186 /* 2187 * This structure will be used for multiple mount protection. It will be 2188 * written into the block number saved in the s_mmp_block field in the 2189 * superblock. Programs that check MMP should assume that if 2190 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2191 * to use the filesystem, regardless of how old the timestamp is. 2192 */ 2193 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2194 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2195 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2196 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2197 2198 struct mmp_struct { 2199 __le32 mmp_magic; /* Magic number for MMP */ 2200 __le32 mmp_seq; /* Sequence no. updated periodically */ 2201 2202 /* 2203 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2204 * purposes and do not affect the correctness of the algorithm 2205 */ 2206 __le64 mmp_time; /* Time last updated */ 2207 char mmp_nodename[64]; /* Node which last updated MMP block */ 2208 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2209 2210 /* 2211 * mmp_check_interval is used to verify if the MMP block has been 2212 * updated on the block device. The value is updated based on the 2213 * maximum time to write the MMP block during an update cycle. 2214 */ 2215 __le16 mmp_check_interval; 2216 2217 __le16 mmp_pad1; 2218 __le32 mmp_pad2[226]; 2219 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2220 }; 2221 2222 /* arguments passed to the mmp thread */ 2223 struct mmpd_data { 2224 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2225 struct super_block *sb; /* super block of the fs */ 2226 }; 2227 2228 /* 2229 * Check interval multiplier 2230 * The MMP block is written every update interval and initially checked every 2231 * update interval x the multiplier (the value is then adapted based on the 2232 * write latency). The reason is that writes can be delayed under load and we 2233 * don't want readers to incorrectly assume that the filesystem is no longer 2234 * in use. 2235 */ 2236 #define EXT4_MMP_CHECK_MULT 2UL 2237 2238 /* 2239 * Minimum interval for MMP checking in seconds. 2240 */ 2241 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2242 2243 /* 2244 * Maximum interval for MMP checking in seconds. 2245 */ 2246 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2247 2248 /* 2249 * Function prototypes 2250 */ 2251 2252 /* 2253 * Ok, these declarations are also in <linux/kernel.h> but none of the 2254 * ext4 source programs needs to include it so they are duplicated here. 2255 */ 2256 # define NORET_TYPE /**/ 2257 # define ATTRIB_NORET __attribute__((noreturn)) 2258 # define NORET_AND noreturn, 2259 2260 /* bitmap.c */ 2261 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2262 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2263 struct ext4_group_desc *gdp, 2264 struct buffer_head *bh, int sz); 2265 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2266 struct ext4_group_desc *gdp, 2267 struct buffer_head *bh, int sz); 2268 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2269 struct ext4_group_desc *gdp, 2270 struct buffer_head *bh); 2271 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2272 struct ext4_group_desc *gdp, 2273 struct buffer_head *bh); 2274 2275 /* balloc.c */ 2276 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2277 ext4_fsblk_t blocknr, 2278 ext4_group_t *blockgrpp, 2279 ext4_grpblk_t *offsetp); 2280 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2281 ext4_fsblk_t block); 2282 2283 extern unsigned int ext4_block_group(struct super_block *sb, 2284 ext4_fsblk_t blocknr); 2285 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2286 ext4_fsblk_t blocknr); 2287 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2288 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2289 ext4_group_t group); 2290 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2291 ext4_fsblk_t goal, 2292 unsigned int flags, 2293 unsigned long *count, 2294 int *errp); 2295 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2296 s64 nclusters, unsigned int flags); 2297 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2298 extern void ext4_check_blocks_bitmap(struct super_block *); 2299 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2300 ext4_group_t block_group, 2301 struct buffer_head ** bh); 2302 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2303 2304 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2305 ext4_group_t block_group); 2306 extern int ext4_wait_block_bitmap(struct super_block *sb, 2307 ext4_group_t block_group, 2308 struct buffer_head *bh); 2309 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2310 ext4_group_t block_group); 2311 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2312 ext4_group_t block_group, 2313 struct ext4_group_desc *gdp); 2314 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2315 2316 #ifdef CONFIG_UNICODE 2317 extern void ext4_fname_setup_ci_filename(struct inode *dir, 2318 const struct qstr *iname, 2319 struct fscrypt_str *fname); 2320 #endif 2321 2322 #ifdef CONFIG_FS_ENCRYPTION 2323 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst, 2324 const struct fscrypt_name *src) 2325 { 2326 memset(dst, 0, sizeof(*dst)); 2327 2328 dst->usr_fname = src->usr_fname; 2329 dst->disk_name = src->disk_name; 2330 dst->hinfo.hash = src->hash; 2331 dst->hinfo.minor_hash = src->minor_hash; 2332 dst->crypto_buf = src->crypto_buf; 2333 } 2334 2335 static inline int ext4_fname_setup_filename(struct inode *dir, 2336 const struct qstr *iname, 2337 int lookup, 2338 struct ext4_filename *fname) 2339 { 2340 struct fscrypt_name name; 2341 int err; 2342 2343 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2344 if (err) 2345 return err; 2346 2347 ext4_fname_from_fscrypt_name(fname, &name); 2348 2349 #ifdef CONFIG_UNICODE 2350 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2351 #endif 2352 return 0; 2353 } 2354 2355 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2356 struct dentry *dentry, 2357 struct ext4_filename *fname) 2358 { 2359 struct fscrypt_name name; 2360 int err; 2361 2362 err = fscrypt_prepare_lookup(dir, dentry, &name); 2363 if (err) 2364 return err; 2365 2366 ext4_fname_from_fscrypt_name(fname, &name); 2367 2368 #ifdef CONFIG_UNICODE 2369 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name); 2370 #endif 2371 return 0; 2372 } 2373 2374 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2375 { 2376 struct fscrypt_name name; 2377 2378 name.crypto_buf = fname->crypto_buf; 2379 fscrypt_free_filename(&name); 2380 2381 fname->crypto_buf.name = NULL; 2382 fname->usr_fname = NULL; 2383 fname->disk_name.name = NULL; 2384 2385 #ifdef CONFIG_UNICODE 2386 kfree(fname->cf_name.name); 2387 fname->cf_name.name = NULL; 2388 #endif 2389 } 2390 #else /* !CONFIG_FS_ENCRYPTION */ 2391 static inline int ext4_fname_setup_filename(struct inode *dir, 2392 const struct qstr *iname, 2393 int lookup, 2394 struct ext4_filename *fname) 2395 { 2396 fname->usr_fname = iname; 2397 fname->disk_name.name = (unsigned char *) iname->name; 2398 fname->disk_name.len = iname->len; 2399 2400 #ifdef CONFIG_UNICODE 2401 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name); 2402 #endif 2403 2404 return 0; 2405 } 2406 2407 static inline int ext4_fname_prepare_lookup(struct inode *dir, 2408 struct dentry *dentry, 2409 struct ext4_filename *fname) 2410 { 2411 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2412 } 2413 2414 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2415 { 2416 #ifdef CONFIG_UNICODE 2417 kfree(fname->cf_name.name); 2418 fname->cf_name.name = NULL; 2419 #endif 2420 } 2421 #endif /* !CONFIG_FS_ENCRYPTION */ 2422 2423 /* dir.c */ 2424 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2425 struct file *, 2426 struct ext4_dir_entry_2 *, 2427 struct buffer_head *, char *, int, 2428 unsigned int); 2429 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2430 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2431 (de), (bh), (buf), (size), (offset))) 2432 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2433 __u32 minor_hash, 2434 struct ext4_dir_entry_2 *dirent, 2435 struct fscrypt_str *ent_name); 2436 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2437 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2438 struct buffer_head *bh, 2439 void *buf, int buf_size, 2440 struct ext4_filename *fname, 2441 struct ext4_dir_entry_2 **dest_de); 2442 void ext4_insert_dentry(struct inode *inode, 2443 struct ext4_dir_entry_2 *de, 2444 int buf_size, 2445 struct ext4_filename *fname); 2446 static inline void ext4_update_dx_flag(struct inode *inode) 2447 { 2448 if (!ext4_has_feature_dir_index(inode->i_sb)) 2449 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2450 } 2451 static const unsigned char ext4_filetype_table[] = { 2452 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2453 }; 2454 2455 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2456 { 2457 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2458 return DT_UNKNOWN; 2459 2460 return ext4_filetype_table[filetype]; 2461 } 2462 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2463 void *buf, int buf_size); 2464 2465 /* fsync.c */ 2466 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2467 2468 /* hash.c */ 2469 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2470 struct dx_hash_info *hinfo); 2471 2472 /* ialloc.c */ 2473 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2474 const struct qstr *qstr, __u32 goal, 2475 uid_t *owner, __u32 i_flags, 2476 int handle_type, unsigned int line_no, 2477 int nblocks); 2478 2479 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2480 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2481 i_flags, 0, 0, 0) 2482 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2483 type, nblocks) \ 2484 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2485 0, (type), __LINE__, (nblocks)) 2486 2487 2488 extern void ext4_free_inode(handle_t *, struct inode *); 2489 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2490 extern unsigned long ext4_count_free_inodes(struct super_block *); 2491 extern unsigned long ext4_count_dirs(struct super_block *); 2492 extern void ext4_check_inodes_bitmap(struct super_block *); 2493 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2494 extern int ext4_init_inode_table(struct super_block *sb, 2495 ext4_group_t group, int barrier); 2496 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2497 2498 /* mballoc.c */ 2499 extern const struct seq_operations ext4_mb_seq_groups_ops; 2500 extern long ext4_mb_stats; 2501 extern long ext4_mb_max_to_scan; 2502 extern int ext4_mb_init(struct super_block *); 2503 extern int ext4_mb_release(struct super_block *); 2504 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2505 struct ext4_allocation_request *, int *); 2506 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2507 extern void ext4_discard_preallocations(struct inode *); 2508 extern int __init ext4_init_mballoc(void); 2509 extern void ext4_exit_mballoc(void); 2510 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2511 struct buffer_head *bh, ext4_fsblk_t block, 2512 unsigned long count, int flags); 2513 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2514 ext4_group_t ngroups); 2515 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2516 ext4_group_t i, struct ext4_group_desc *desc); 2517 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2518 ext4_fsblk_t block, unsigned long count); 2519 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2520 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2521 2522 /* inode.c */ 2523 int ext4_inode_is_fast_symlink(struct inode *inode); 2524 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2525 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2526 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2527 bool wait, struct buffer_head **bhs); 2528 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2529 struct buffer_head *bh_result, int create); 2530 int ext4_get_block(struct inode *inode, sector_t iblock, 2531 struct buffer_head *bh_result, int create); 2532 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2533 struct buffer_head *bh_result, int create); 2534 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2535 struct buffer_head *bh, int create); 2536 int ext4_walk_page_buffers(handle_t *handle, 2537 struct buffer_head *head, 2538 unsigned from, 2539 unsigned to, 2540 int *partial, 2541 int (*fn)(handle_t *handle, 2542 struct buffer_head *bh)); 2543 int do_journal_get_write_access(handle_t *handle, 2544 struct buffer_head *bh); 2545 #define FALL_BACK_TO_NONDELALLOC 1 2546 #define CONVERT_INLINE_DATA 2 2547 2548 typedef enum { 2549 EXT4_IGET_NORMAL = 0, 2550 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2551 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2552 } ext4_iget_flags; 2553 2554 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2555 ext4_iget_flags flags, const char *function, 2556 unsigned int line); 2557 2558 #define ext4_iget(sb, ino, flags) \ 2559 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2560 2561 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2562 extern int ext4_setattr(struct dentry *, struct iattr *); 2563 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2564 extern void ext4_evict_inode(struct inode *); 2565 extern void ext4_clear_inode(struct inode *); 2566 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2567 extern int ext4_sync_inode(handle_t *, struct inode *); 2568 extern void ext4_dirty_inode(struct inode *, int); 2569 extern int ext4_change_inode_journal_flag(struct inode *, int); 2570 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2571 extern int ext4_inode_attach_jinode(struct inode *inode); 2572 extern int ext4_can_truncate(struct inode *inode); 2573 extern int ext4_truncate(struct inode *); 2574 extern int ext4_break_layouts(struct inode *); 2575 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2576 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2577 extern void ext4_set_inode_flags(struct inode *); 2578 extern int ext4_alloc_da_blocks(struct inode *inode); 2579 extern void ext4_set_aops(struct inode *inode); 2580 extern int ext4_writepage_trans_blocks(struct inode *); 2581 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2582 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2583 loff_t lstart, loff_t lend); 2584 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2585 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2586 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2587 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2588 extern void ext4_da_release_space(struct inode *inode, int to_free); 2589 extern void ext4_da_update_reserve_space(struct inode *inode, 2590 int used, int quota_claim); 2591 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2592 ext4_fsblk_t pblk, ext4_lblk_t len); 2593 2594 /* indirect.c */ 2595 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2596 struct ext4_map_blocks *map, int flags); 2597 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2598 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2599 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2600 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2601 ext4_lblk_t start, ext4_lblk_t end); 2602 2603 /* ioctl.c */ 2604 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2605 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2606 2607 /* migrate.c */ 2608 extern int ext4_ext_migrate(struct inode *); 2609 extern int ext4_ind_migrate(struct inode *inode); 2610 2611 /* namei.c */ 2612 extern int ext4_dirblock_csum_verify(struct inode *inode, 2613 struct buffer_head *bh); 2614 extern int ext4_orphan_add(handle_t *, struct inode *); 2615 extern int ext4_orphan_del(handle_t *, struct inode *); 2616 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2617 __u32 start_minor_hash, __u32 *next_hash); 2618 extern int ext4_search_dir(struct buffer_head *bh, 2619 char *search_buf, 2620 int buf_size, 2621 struct inode *dir, 2622 struct ext4_filename *fname, 2623 unsigned int offset, 2624 struct ext4_dir_entry_2 **res_dir); 2625 extern int ext4_generic_delete_entry(handle_t *handle, 2626 struct inode *dir, 2627 struct ext4_dir_entry_2 *de_del, 2628 struct buffer_head *bh, 2629 void *entry_buf, 2630 int buf_size, 2631 int csum_size); 2632 extern bool ext4_empty_dir(struct inode *inode); 2633 2634 /* resize.c */ 2635 extern int ext4_group_add(struct super_block *sb, 2636 struct ext4_new_group_data *input); 2637 extern int ext4_group_extend(struct super_block *sb, 2638 struct ext4_super_block *es, 2639 ext4_fsblk_t n_blocks_count); 2640 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2641 2642 /* super.c */ 2643 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2644 sector_t block, int op_flags); 2645 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2646 extern int ext4_calculate_overhead(struct super_block *sb); 2647 extern void ext4_superblock_csum_set(struct super_block *sb); 2648 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2649 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2650 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2651 ext4_group_t ngroup); 2652 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2653 char nbuf[16]); 2654 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2655 ext4_group_t block_group, 2656 unsigned int flags); 2657 2658 extern __printf(4, 5) 2659 void __ext4_error(struct super_block *, const char *, unsigned int, 2660 const char *, ...); 2661 extern __printf(5, 6) 2662 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2663 const char *, ...); 2664 extern __printf(5, 6) 2665 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2666 const char *, ...); 2667 extern void __ext4_std_error(struct super_block *, const char *, 2668 unsigned int, int); 2669 extern __printf(4, 5) 2670 void __ext4_abort(struct super_block *, const char *, unsigned int, 2671 const char *, ...); 2672 extern __printf(4, 5) 2673 void __ext4_warning(struct super_block *, const char *, unsigned int, 2674 const char *, ...); 2675 extern __printf(4, 5) 2676 void __ext4_warning_inode(const struct inode *inode, const char *function, 2677 unsigned int line, const char *fmt, ...); 2678 extern __printf(3, 4) 2679 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2680 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2681 const char *, unsigned int, const char *); 2682 extern __printf(7, 8) 2683 void __ext4_grp_locked_error(const char *, unsigned int, 2684 struct super_block *, ext4_group_t, 2685 unsigned long, ext4_fsblk_t, 2686 const char *, ...); 2687 2688 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2689 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2690 2691 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2692 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2693 2694 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2695 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2696 2697 #ifdef CONFIG_PRINTK 2698 2699 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2700 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2701 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2702 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2703 #define ext4_error(sb, fmt, ...) \ 2704 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2705 #define ext4_abort(sb, fmt, ...) \ 2706 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2707 #define ext4_warning(sb, fmt, ...) \ 2708 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2709 #define ext4_warning_inode(inode, fmt, ...) \ 2710 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2711 #define ext4_msg(sb, level, fmt, ...) \ 2712 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2713 #define dump_mmp_msg(sb, mmp, msg) \ 2714 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2715 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2716 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2717 fmt, ##__VA_ARGS__) 2718 2719 #else 2720 2721 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2722 do { \ 2723 no_printk(fmt, ##__VA_ARGS__); \ 2724 __ext4_error_inode(inode, "", 0, block, " "); \ 2725 } while (0) 2726 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2727 do { \ 2728 no_printk(fmt, ##__VA_ARGS__); \ 2729 __ext4_error_file(file, "", 0, block, " "); \ 2730 } while (0) 2731 #define ext4_error(sb, fmt, ...) \ 2732 do { \ 2733 no_printk(fmt, ##__VA_ARGS__); \ 2734 __ext4_error(sb, "", 0, " "); \ 2735 } while (0) 2736 #define ext4_abort(sb, fmt, ...) \ 2737 do { \ 2738 no_printk(fmt, ##__VA_ARGS__); \ 2739 __ext4_abort(sb, "", 0, " "); \ 2740 } while (0) 2741 #define ext4_warning(sb, fmt, ...) \ 2742 do { \ 2743 no_printk(fmt, ##__VA_ARGS__); \ 2744 __ext4_warning(sb, "", 0, " "); \ 2745 } while (0) 2746 #define ext4_warning_inode(inode, fmt, ...) \ 2747 do { \ 2748 no_printk(fmt, ##__VA_ARGS__); \ 2749 __ext4_warning_inode(inode, "", 0, " "); \ 2750 } while (0) 2751 #define ext4_msg(sb, level, fmt, ...) \ 2752 do { \ 2753 no_printk(fmt, ##__VA_ARGS__); \ 2754 __ext4_msg(sb, "", " "); \ 2755 } while (0) 2756 #define dump_mmp_msg(sb, mmp, msg) \ 2757 __dump_mmp_msg(sb, mmp, "", 0, "") 2758 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2759 do { \ 2760 no_printk(fmt, ##__VA_ARGS__); \ 2761 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2762 } while (0) 2763 2764 #endif 2765 2766 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2767 __u32 compat); 2768 extern int ext4_update_rocompat_feature(handle_t *handle, 2769 struct super_block *sb, __u32 rocompat); 2770 extern int ext4_update_incompat_feature(handle_t *handle, 2771 struct super_block *sb, __u32 incompat); 2772 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2773 struct ext4_group_desc *bg); 2774 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2775 struct ext4_group_desc *bg); 2776 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2777 struct ext4_group_desc *bg); 2778 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2779 struct ext4_group_desc *bg); 2780 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2781 struct ext4_group_desc *bg); 2782 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2783 struct ext4_group_desc *bg); 2784 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2785 struct ext4_group_desc *bg); 2786 extern void ext4_block_bitmap_set(struct super_block *sb, 2787 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2788 extern void ext4_inode_bitmap_set(struct super_block *sb, 2789 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2790 extern void ext4_inode_table_set(struct super_block *sb, 2791 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2792 extern void ext4_free_group_clusters_set(struct super_block *sb, 2793 struct ext4_group_desc *bg, 2794 __u32 count); 2795 extern void ext4_free_inodes_set(struct super_block *sb, 2796 struct ext4_group_desc *bg, __u32 count); 2797 extern void ext4_used_dirs_set(struct super_block *sb, 2798 struct ext4_group_desc *bg, __u32 count); 2799 extern void ext4_itable_unused_set(struct super_block *sb, 2800 struct ext4_group_desc *bg, __u32 count); 2801 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2802 struct ext4_group_desc *gdp); 2803 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2804 struct ext4_group_desc *gdp); 2805 extern int ext4_register_li_request(struct super_block *sb, 2806 ext4_group_t first_not_zeroed); 2807 2808 static inline int ext4_has_metadata_csum(struct super_block *sb) 2809 { 2810 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2811 !EXT4_SB(sb)->s_chksum_driver); 2812 2813 return ext4_has_feature_metadata_csum(sb) && 2814 (EXT4_SB(sb)->s_chksum_driver != NULL); 2815 } 2816 2817 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2818 { 2819 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2820 } 2821 2822 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2823 { 2824 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2825 le32_to_cpu(es->s_blocks_count_lo); 2826 } 2827 2828 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2829 { 2830 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2831 le32_to_cpu(es->s_r_blocks_count_lo); 2832 } 2833 2834 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2835 { 2836 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2837 le32_to_cpu(es->s_free_blocks_count_lo); 2838 } 2839 2840 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2841 ext4_fsblk_t blk) 2842 { 2843 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2844 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2845 } 2846 2847 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2848 ext4_fsblk_t blk) 2849 { 2850 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2851 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2852 } 2853 2854 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2855 ext4_fsblk_t blk) 2856 { 2857 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2858 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2859 } 2860 2861 static inline loff_t ext4_isize(struct super_block *sb, 2862 struct ext4_inode *raw_inode) 2863 { 2864 if (ext4_has_feature_largedir(sb) || 2865 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2866 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2867 le32_to_cpu(raw_inode->i_size_lo); 2868 2869 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2870 } 2871 2872 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2873 { 2874 raw_inode->i_size_lo = cpu_to_le32(i_size); 2875 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2876 } 2877 2878 static inline 2879 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2880 ext4_group_t group) 2881 { 2882 struct ext4_group_info ***grp_info; 2883 long indexv, indexh; 2884 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2885 grp_info = EXT4_SB(sb)->s_group_info; 2886 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2887 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2888 return grp_info[indexv][indexh]; 2889 } 2890 2891 /* 2892 * Reading s_groups_count requires using smp_rmb() afterwards. See 2893 * the locking protocol documented in the comments of ext4_group_add() 2894 * in resize.c 2895 */ 2896 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2897 { 2898 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2899 2900 smp_rmb(); 2901 return ngroups; 2902 } 2903 2904 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2905 ext4_group_t block_group) 2906 { 2907 return block_group >> sbi->s_log_groups_per_flex; 2908 } 2909 2910 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2911 { 2912 return 1 << sbi->s_log_groups_per_flex; 2913 } 2914 2915 #define ext4_std_error(sb, errno) \ 2916 do { \ 2917 if ((errno)) \ 2918 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2919 } while (0) 2920 2921 #ifdef CONFIG_SMP 2922 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2923 * counters. So we need to make sure we have free clusters more 2924 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2925 */ 2926 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2927 #else 2928 #define EXT4_FREECLUSTERS_WATERMARK 0 2929 #endif 2930 2931 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2932 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2933 { 2934 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2935 !inode_is_locked(inode)); 2936 down_write(&EXT4_I(inode)->i_data_sem); 2937 if (newsize > EXT4_I(inode)->i_disksize) 2938 EXT4_I(inode)->i_disksize = newsize; 2939 up_write(&EXT4_I(inode)->i_data_sem); 2940 } 2941 2942 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2943 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2944 { 2945 int changed = 0; 2946 2947 if (newsize > inode->i_size) { 2948 i_size_write(inode, newsize); 2949 changed = 1; 2950 } 2951 if (newsize > EXT4_I(inode)->i_disksize) { 2952 ext4_update_i_disksize(inode, newsize); 2953 changed |= 2; 2954 } 2955 return changed; 2956 } 2957 2958 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2959 loff_t len); 2960 2961 struct ext4_group_info { 2962 unsigned long bb_state; 2963 struct rb_root bb_free_root; 2964 ext4_grpblk_t bb_first_free; /* first free block */ 2965 ext4_grpblk_t bb_free; /* total free blocks */ 2966 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2967 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2968 struct list_head bb_prealloc_list; 2969 #ifdef DOUBLE_CHECK 2970 void *bb_bitmap; 2971 #endif 2972 struct rw_semaphore alloc_sem; 2973 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2974 * regions, index is order. 2975 * bb_counters[3] = 5 means 2976 * 5 free 8-block regions. */ 2977 }; 2978 2979 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2980 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2981 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2982 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2983 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2984 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2985 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2986 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2987 2988 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2989 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2990 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2991 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2992 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2993 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2994 2995 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2996 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2997 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2998 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2999 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3000 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3001 3002 #define EXT4_MAX_CONTENTION 8 3003 #define EXT4_CONTENTION_THRESHOLD 2 3004 3005 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3006 ext4_group_t group) 3007 { 3008 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3009 } 3010 3011 /* 3012 * Returns true if the filesystem is busy enough that attempts to 3013 * access the block group locks has run into contention. 3014 */ 3015 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3016 { 3017 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3018 } 3019 3020 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3021 { 3022 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 3023 if (spin_trylock(lock)) 3024 /* 3025 * We're able to grab the lock right away, so drop the 3026 * lock contention counter. 3027 */ 3028 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3029 else { 3030 /* 3031 * The lock is busy, so bump the contention counter, 3032 * and then wait on the spin lock. 3033 */ 3034 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3035 EXT4_MAX_CONTENTION); 3036 spin_lock(lock); 3037 } 3038 } 3039 3040 static inline void ext4_unlock_group(struct super_block *sb, 3041 ext4_group_t group) 3042 { 3043 spin_unlock(ext4_group_lock_ptr(sb, group)); 3044 } 3045 3046 /* 3047 * Block validity checking 3048 */ 3049 #define ext4_check_indirect_blockref(inode, bh) \ 3050 ext4_check_blockref(__func__, __LINE__, inode, \ 3051 (__le32 *)(bh)->b_data, \ 3052 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3053 3054 #define ext4_ind_check_inode(inode) \ 3055 ext4_check_blockref(__func__, __LINE__, inode, \ 3056 EXT4_I(inode)->i_data, \ 3057 EXT4_NDIR_BLOCKS) 3058 3059 /* 3060 * Inodes and files operations 3061 */ 3062 3063 /* dir.c */ 3064 extern const struct file_operations ext4_dir_operations; 3065 3066 #ifdef CONFIG_UNICODE 3067 extern const struct dentry_operations ext4_dentry_ops; 3068 #endif 3069 3070 /* file.c */ 3071 extern const struct inode_operations ext4_file_inode_operations; 3072 extern const struct file_operations ext4_file_operations; 3073 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3074 3075 /* inline.c */ 3076 extern int ext4_get_max_inline_size(struct inode *inode); 3077 extern int ext4_find_inline_data_nolock(struct inode *inode); 3078 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 3079 unsigned int len); 3080 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3081 3082 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3083 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3084 struct inode *inode, 3085 loff_t pos, unsigned len, 3086 unsigned flags, 3087 struct page **pagep); 3088 extern int ext4_write_inline_data_end(struct inode *inode, 3089 loff_t pos, unsigned len, 3090 unsigned copied, 3091 struct page *page); 3092 extern struct buffer_head * 3093 ext4_journalled_write_inline_data(struct inode *inode, 3094 unsigned len, 3095 struct page *page); 3096 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3097 struct inode *inode, 3098 loff_t pos, unsigned len, 3099 unsigned flags, 3100 struct page **pagep, 3101 void **fsdata); 3102 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3103 unsigned len, unsigned copied, 3104 struct page *page); 3105 extern int ext4_try_add_inline_entry(handle_t *handle, 3106 struct ext4_filename *fname, 3107 struct inode *dir, struct inode *inode); 3108 extern int ext4_try_create_inline_dir(handle_t *handle, 3109 struct inode *parent, 3110 struct inode *inode); 3111 extern int ext4_read_inline_dir(struct file *filp, 3112 struct dir_context *ctx, 3113 int *has_inline_data); 3114 extern int ext4_inlinedir_to_tree(struct file *dir_file, 3115 struct inode *dir, ext4_lblk_t block, 3116 struct dx_hash_info *hinfo, 3117 __u32 start_hash, __u32 start_minor_hash, 3118 int *has_inline_data); 3119 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3120 struct ext4_filename *fname, 3121 struct ext4_dir_entry_2 **res_dir, 3122 int *has_inline_data); 3123 extern int ext4_delete_inline_entry(handle_t *handle, 3124 struct inode *dir, 3125 struct ext4_dir_entry_2 *de_del, 3126 struct buffer_head *bh, 3127 int *has_inline_data); 3128 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3129 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3130 struct ext4_dir_entry_2 **parent_de, 3131 int *retval); 3132 extern int ext4_inline_data_fiemap(struct inode *inode, 3133 struct fiemap_extent_info *fieinfo, 3134 int *has_inline, __u64 start, __u64 len); 3135 3136 struct iomap; 3137 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3138 3139 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3140 3141 extern int ext4_convert_inline_data(struct inode *inode); 3142 3143 static inline int ext4_has_inline_data(struct inode *inode) 3144 { 3145 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3146 EXT4_I(inode)->i_inline_off; 3147 } 3148 3149 /* namei.c */ 3150 extern const struct inode_operations ext4_dir_inode_operations; 3151 extern const struct inode_operations ext4_special_inode_operations; 3152 extern struct dentry *ext4_get_parent(struct dentry *child); 3153 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3154 struct ext4_dir_entry_2 *de, 3155 int blocksize, int csum_size, 3156 unsigned int parent_ino, int dotdot_real_len); 3157 extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3158 unsigned int blocksize); 3159 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3160 struct buffer_head *bh); 3161 extern int ext4_ci_compare(const struct inode *parent, 3162 const struct qstr *fname, 3163 const struct qstr *entry, bool quick); 3164 3165 #define S_SHIFT 12 3166 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3167 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3168 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3169 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3170 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3171 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3172 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3173 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3174 }; 3175 3176 static inline void ext4_set_de_type(struct super_block *sb, 3177 struct ext4_dir_entry_2 *de, 3178 umode_t mode) { 3179 if (ext4_has_feature_filetype(sb)) 3180 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3181 } 3182 3183 /* readpages.c */ 3184 extern int ext4_mpage_readpages(struct address_space *mapping, 3185 struct list_head *pages, struct page *page, 3186 unsigned nr_pages, bool is_readahead); 3187 3188 /* symlink.c */ 3189 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3190 extern const struct inode_operations ext4_symlink_inode_operations; 3191 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3192 3193 /* sysfs.c */ 3194 extern int ext4_register_sysfs(struct super_block *sb); 3195 extern void ext4_unregister_sysfs(struct super_block *sb); 3196 extern int __init ext4_init_sysfs(void); 3197 extern void ext4_exit_sysfs(void); 3198 3199 /* block_validity */ 3200 extern void ext4_release_system_zone(struct super_block *sb); 3201 extern int ext4_setup_system_zone(struct super_block *sb); 3202 extern int __init ext4_init_system_zone(void); 3203 extern void ext4_exit_system_zone(void); 3204 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3205 ext4_fsblk_t start_blk, 3206 unsigned int count); 3207 extern int ext4_check_blockref(const char *, unsigned int, 3208 struct inode *, __le32 *, unsigned int); 3209 3210 /* extents.c */ 3211 struct ext4_ext_path; 3212 struct ext4_extent; 3213 3214 /* 3215 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3216 * __le32. 3217 */ 3218 #define EXT_MAX_BLOCKS 0xffffffff 3219 3220 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3221 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3222 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3223 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3224 struct ext4_map_blocks *map, int flags); 3225 extern int ext4_ext_truncate(handle_t *, struct inode *); 3226 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3227 ext4_lblk_t end); 3228 extern void ext4_ext_init(struct super_block *); 3229 extern void ext4_ext_release(struct super_block *); 3230 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3231 loff_t len); 3232 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3233 loff_t offset, ssize_t len); 3234 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3235 struct ext4_map_blocks *map, int flags); 3236 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3237 ext4_lblk_t lblocks); 3238 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3239 int num, 3240 struct ext4_ext_path *path); 3241 extern int ext4_can_extents_be_merged(struct inode *inode, 3242 struct ext4_extent *ex1, 3243 struct ext4_extent *ex2); 3244 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3245 struct ext4_ext_path **, 3246 struct ext4_extent *, int); 3247 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3248 struct ext4_ext_path **, 3249 int flags); 3250 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3251 extern int ext4_ext_check_inode(struct inode *inode); 3252 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3253 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3254 __u64 start, __u64 len); 3255 extern int ext4_get_es_cache(struct inode *inode, 3256 struct fiemap_extent_info *fieinfo, 3257 __u64 start, __u64 len); 3258 extern int ext4_ext_precache(struct inode *inode); 3259 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3260 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3261 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3262 struct inode *inode2, ext4_lblk_t lblk1, 3263 ext4_lblk_t lblk2, ext4_lblk_t count, 3264 int mark_unwritten,int *err); 3265 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3266 3267 /* move_extent.c */ 3268 extern void ext4_double_down_write_data_sem(struct inode *first, 3269 struct inode *second); 3270 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3271 struct inode *donor_inode); 3272 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3273 __u64 start_orig, __u64 start_donor, 3274 __u64 len, __u64 *moved_len); 3275 3276 /* page-io.c */ 3277 extern int __init ext4_init_pageio(void); 3278 extern void ext4_exit_pageio(void); 3279 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3280 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3281 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3282 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3283 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3284 struct writeback_control *wbc); 3285 extern void ext4_end_io_rsv_work(struct work_struct *work); 3286 extern void ext4_io_submit(struct ext4_io_submit *io); 3287 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3288 struct page *page, 3289 int len, 3290 struct writeback_control *wbc, 3291 bool keep_towrite); 3292 3293 /* mmp.c */ 3294 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3295 3296 /* 3297 * Add new method to test whether block and inode bitmaps are properly 3298 * initialized. With uninit_bg reading the block from disk is not enough 3299 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3300 */ 3301 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3302 3303 static inline int bitmap_uptodate(struct buffer_head *bh) 3304 { 3305 return (buffer_uptodate(bh) && 3306 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3307 } 3308 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3309 { 3310 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3311 } 3312 3313 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3314 3315 /* For ioend & aio unwritten conversion wait queues */ 3316 #define EXT4_WQ_HASH_SZ 37 3317 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3318 EXT4_WQ_HASH_SZ]) 3319 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3320 3321 extern int ext4_resize_begin(struct super_block *sb); 3322 extern void ext4_resize_end(struct super_block *sb); 3323 3324 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3325 struct ext4_io_end *io_end) 3326 { 3327 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3328 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3329 atomic_inc(&EXT4_I(inode)->i_unwritten); 3330 } 3331 } 3332 3333 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3334 { 3335 struct inode *inode = io_end->inode; 3336 3337 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3338 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3339 /* Wake up anyone waiting on unwritten extent conversion */ 3340 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3341 wake_up_all(ext4_ioend_wq(inode)); 3342 } 3343 } 3344 3345 extern const struct iomap_ops ext4_iomap_ops; 3346 3347 static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3348 { 3349 /* 3350 * If the buffer has the write error flag, we have failed 3351 * to write out data in the block. In this case, we don't 3352 * have to read the block because we may read the old data 3353 * successfully. 3354 */ 3355 if (!buffer_uptodate(bh) && buffer_write_io_error(bh)) 3356 set_buffer_uptodate(bh); 3357 return buffer_uptodate(bh); 3358 } 3359 3360 #endif /* __KERNEL__ */ 3361 3362 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3363 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3364 3365 #endif /* _EXT4_H */ 3366