1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION) 44 #include <linux/fscrypt.h> 45 46 /* 47 * The fourth extended filesystem constants/structures 48 */ 49 50 /* 51 * with AGGRESSIVE_CHECK allocator runs consistency checks over 52 * structures. these checks slow things down a lot 53 */ 54 #define AGGRESSIVE_CHECK__ 55 56 /* 57 * with DOUBLE_CHECK defined mballoc creates persistent in-core 58 * bitmaps, maintains and uses them to check for double allocations 59 */ 60 #define DOUBLE_CHECK__ 61 62 /* 63 * Define EXT4FS_DEBUG to produce debug messages 64 */ 65 #undef EXT4FS_DEBUG 66 67 /* 68 * Debug code 69 */ 70 #ifdef EXT4FS_DEBUG 71 #define ext4_debug(f, a...) \ 72 do { \ 73 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 74 __FILE__, __LINE__, __func__); \ 75 printk(KERN_DEBUG f, ## a); \ 76 } while (0) 77 #else 78 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 79 #endif 80 81 /* 82 * Turn on EXT_DEBUG to get lots of info about extents operations. 83 */ 84 #define EXT_DEBUG__ 85 #ifdef EXT_DEBUG 86 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 87 #else 88 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 89 #endif 90 91 /* data type for block offset of block group */ 92 typedef int ext4_grpblk_t; 93 94 /* data type for filesystem-wide blocks number */ 95 typedef unsigned long long ext4_fsblk_t; 96 97 /* data type for file logical block number */ 98 typedef __u32 ext4_lblk_t; 99 100 /* data type for block group number */ 101 typedef unsigned int ext4_group_t; 102 103 enum SHIFT_DIRECTION { 104 SHIFT_LEFT = 0, 105 SHIFT_RIGHT, 106 }; 107 108 /* 109 * Flags used in mballoc's allocation_context flags field. 110 * 111 * Also used to show what's going on for debugging purposes when the 112 * flag field is exported via the traceport interface 113 */ 114 115 /* prefer goal again. length */ 116 #define EXT4_MB_HINT_MERGE 0x0001 117 /* blocks already reserved */ 118 #define EXT4_MB_HINT_RESERVED 0x0002 119 /* metadata is being allocated */ 120 #define EXT4_MB_HINT_METADATA 0x0004 121 /* first blocks in the file */ 122 #define EXT4_MB_HINT_FIRST 0x0008 123 /* search for the best chunk */ 124 #define EXT4_MB_HINT_BEST 0x0010 125 /* data is being allocated */ 126 #define EXT4_MB_HINT_DATA 0x0020 127 /* don't preallocate (for tails) */ 128 #define EXT4_MB_HINT_NOPREALLOC 0x0040 129 /* allocate for locality group */ 130 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 131 /* allocate goal blocks or none */ 132 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 133 /* goal is meaningful */ 134 #define EXT4_MB_HINT_TRY_GOAL 0x0200 135 /* blocks already pre-reserved by delayed allocation */ 136 #define EXT4_MB_DELALLOC_RESERVED 0x0400 137 /* We are doing stream allocation */ 138 #define EXT4_MB_STREAM_ALLOC 0x0800 139 /* Use reserved root blocks if needed */ 140 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 141 /* Use blocks from reserved pool */ 142 #define EXT4_MB_USE_RESERVED 0x2000 143 144 struct ext4_allocation_request { 145 /* target inode for block we're allocating */ 146 struct inode *inode; 147 /* how many blocks we want to allocate */ 148 unsigned int len; 149 /* logical block in target inode */ 150 ext4_lblk_t logical; 151 /* the closest logical allocated block to the left */ 152 ext4_lblk_t lleft; 153 /* the closest logical allocated block to the right */ 154 ext4_lblk_t lright; 155 /* phys. target (a hint) */ 156 ext4_fsblk_t goal; 157 /* phys. block for the closest logical allocated block to the left */ 158 ext4_fsblk_t pleft; 159 /* phys. block for the closest logical allocated block to the right */ 160 ext4_fsblk_t pright; 161 /* flags. see above EXT4_MB_HINT_* */ 162 unsigned int flags; 163 }; 164 165 /* 166 * Logical to physical block mapping, used by ext4_map_blocks() 167 * 168 * This structure is used to pass requests into ext4_map_blocks() as 169 * well as to store the information returned by ext4_map_blocks(). It 170 * takes less room on the stack than a struct buffer_head. 171 */ 172 #define EXT4_MAP_NEW (1 << BH_New) 173 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 174 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 175 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 176 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 177 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 178 179 struct ext4_map_blocks { 180 ext4_fsblk_t m_pblk; 181 ext4_lblk_t m_lblk; 182 unsigned int m_len; 183 unsigned int m_flags; 184 }; 185 186 /* 187 * Flags for ext4_io_end->flags 188 */ 189 #define EXT4_IO_END_UNWRITTEN 0x0001 190 191 /* 192 * For converting unwritten extents on a work queue. 'handle' is used for 193 * buffered writeback. 194 */ 195 typedef struct ext4_io_end { 196 struct list_head list; /* per-file finished IO list */ 197 handle_t *handle; /* handle reserved for extent 198 * conversion */ 199 struct inode *inode; /* file being written to */ 200 struct bio *bio; /* Linked list of completed 201 * bios covering the extent */ 202 unsigned int flag; /* unwritten or not */ 203 atomic_t count; /* reference counter */ 204 loff_t offset; /* offset in the file */ 205 ssize_t size; /* size of the extent */ 206 } ext4_io_end_t; 207 208 struct ext4_io_submit { 209 struct writeback_control *io_wbc; 210 struct bio *io_bio; 211 ext4_io_end_t *io_end; 212 sector_t io_next_block; 213 }; 214 215 /* 216 * Special inodes numbers 217 */ 218 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 219 #define EXT4_ROOT_INO 2 /* Root inode */ 220 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 221 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 222 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 223 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 224 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 225 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 226 227 /* First non-reserved inode for old ext4 filesystems */ 228 #define EXT4_GOOD_OLD_FIRST_INO 11 229 230 /* 231 * Maximal count of links to a file 232 */ 233 #define EXT4_LINK_MAX 65000 234 235 /* 236 * Macro-instructions used to manage several block sizes 237 */ 238 #define EXT4_MIN_BLOCK_SIZE 1024 239 #define EXT4_MAX_BLOCK_SIZE 65536 240 #define EXT4_MIN_BLOCK_LOG_SIZE 10 241 #define EXT4_MAX_BLOCK_LOG_SIZE 16 242 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 243 #ifdef __KERNEL__ 244 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 245 #else 246 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 247 #endif 248 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 249 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 250 EXT4_SB(s)->s_cluster_bits) 251 #ifdef __KERNEL__ 252 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 253 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 254 #else 255 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 256 #endif 257 #ifdef __KERNEL__ 258 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 259 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 260 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 261 #else 262 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 263 EXT4_GOOD_OLD_INODE_SIZE : \ 264 (s)->s_inode_size) 265 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 266 EXT4_GOOD_OLD_FIRST_INO : \ 267 (s)->s_first_ino) 268 #endif 269 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 270 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 271 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 272 blkbits)) 273 274 /* Translate a block number to a cluster number */ 275 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 276 /* Translate a cluster number to a block number */ 277 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 278 /* Translate # of blks to # of clusters */ 279 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 280 (sbi)->s_cluster_bits) 281 /* Mask out the low bits to get the starting block of the cluster */ 282 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 283 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 284 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 285 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 286 /* Get the cluster offset */ 287 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 288 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 289 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 290 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 291 292 /* 293 * Structure of a blocks group descriptor 294 */ 295 struct ext4_group_desc 296 { 297 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 298 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 299 __le32 bg_inode_table_lo; /* Inodes table block */ 300 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 301 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 302 __le16 bg_used_dirs_count_lo; /* Directories count */ 303 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 304 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 305 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 306 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 307 __le16 bg_itable_unused_lo; /* Unused inodes count */ 308 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 309 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 310 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 311 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 312 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 313 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 314 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 315 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 316 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 317 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 318 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 319 __u32 bg_reserved; 320 }; 321 322 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 323 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 324 sizeof(__le16)) 325 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 326 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 327 sizeof(__le16)) 328 329 /* 330 * Structure of a flex block group info 331 */ 332 333 struct flex_groups { 334 atomic64_t free_clusters; 335 atomic_t free_inodes; 336 atomic_t used_dirs; 337 }; 338 339 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 340 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 341 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 342 343 /* 344 * Macro-instructions used to manage group descriptors 345 */ 346 #define EXT4_MIN_DESC_SIZE 32 347 #define EXT4_MIN_DESC_SIZE_64BIT 64 348 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 349 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 350 #ifdef __KERNEL__ 351 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 352 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 353 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 354 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 355 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 356 #else 357 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 358 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 359 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 360 #endif 361 362 /* 363 * Constants relative to the data blocks 364 */ 365 #define EXT4_NDIR_BLOCKS 12 366 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 367 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 368 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 369 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 370 371 /* 372 * Inode flags 373 */ 374 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 375 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 376 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 377 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 378 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 379 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 380 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 381 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 382 /* Reserved for compression usage... */ 383 #define EXT4_DIRTY_FL 0x00000100 384 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 385 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 386 /* nb: was previously EXT2_ECOMPR_FL */ 387 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 388 /* End compression flags --- maybe not all used */ 389 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 390 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 391 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 392 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 393 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 394 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 395 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 396 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 397 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 398 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 399 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 400 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 401 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 402 403 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 404 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 405 406 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 407 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 408 EXT4_IMMUTABLE_FL | \ 409 EXT4_APPEND_FL | \ 410 EXT4_NODUMP_FL | \ 411 EXT4_NOATIME_FL | \ 412 EXT4_PROJINHERIT_FL) 413 414 /* Flags that should be inherited by new inodes from their parent. */ 415 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 416 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 417 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 418 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 419 EXT4_PROJINHERIT_FL) 420 421 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 422 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 423 424 /* Flags that are appropriate for non-directories/regular files. */ 425 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 426 427 /* Mask out flags that are inappropriate for the given type of inode. */ 428 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 429 { 430 if (S_ISDIR(mode)) 431 return flags; 432 else if (S_ISREG(mode)) 433 return flags & EXT4_REG_FLMASK; 434 else 435 return flags & EXT4_OTHER_FLMASK; 436 } 437 438 /* 439 * Inode flags used for atomic set/get 440 */ 441 enum { 442 EXT4_INODE_SECRM = 0, /* Secure deletion */ 443 EXT4_INODE_UNRM = 1, /* Undelete */ 444 EXT4_INODE_COMPR = 2, /* Compress file */ 445 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 446 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 447 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 448 EXT4_INODE_NODUMP = 6, /* do not dump file */ 449 EXT4_INODE_NOATIME = 7, /* do not update atime */ 450 /* Reserved for compression usage... */ 451 EXT4_INODE_DIRTY = 8, 452 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 453 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 454 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 455 /* End compression flags --- maybe not all used */ 456 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 457 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 458 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 459 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 460 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 461 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 462 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 463 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 464 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 465 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 466 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 467 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 468 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 469 }; 470 471 /* 472 * Since it's pretty easy to mix up bit numbers and hex values, we use a 473 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 474 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 475 * any extra space in the compiled kernel image, otherwise, the build will fail. 476 * It's important that these values are the same, since we are using 477 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 478 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 479 * values found in ext2, ext3 and ext4 filesystems, and of course the values 480 * defined in e2fsprogs. 481 * 482 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 483 */ 484 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 485 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 486 487 static inline void ext4_check_flag_values(void) 488 { 489 CHECK_FLAG_VALUE(SECRM); 490 CHECK_FLAG_VALUE(UNRM); 491 CHECK_FLAG_VALUE(COMPR); 492 CHECK_FLAG_VALUE(SYNC); 493 CHECK_FLAG_VALUE(IMMUTABLE); 494 CHECK_FLAG_VALUE(APPEND); 495 CHECK_FLAG_VALUE(NODUMP); 496 CHECK_FLAG_VALUE(NOATIME); 497 CHECK_FLAG_VALUE(DIRTY); 498 CHECK_FLAG_VALUE(COMPRBLK); 499 CHECK_FLAG_VALUE(NOCOMPR); 500 CHECK_FLAG_VALUE(ENCRYPT); 501 CHECK_FLAG_VALUE(INDEX); 502 CHECK_FLAG_VALUE(IMAGIC); 503 CHECK_FLAG_VALUE(JOURNAL_DATA); 504 CHECK_FLAG_VALUE(NOTAIL); 505 CHECK_FLAG_VALUE(DIRSYNC); 506 CHECK_FLAG_VALUE(TOPDIR); 507 CHECK_FLAG_VALUE(HUGE_FILE); 508 CHECK_FLAG_VALUE(EXTENTS); 509 CHECK_FLAG_VALUE(EA_INODE); 510 CHECK_FLAG_VALUE(EOFBLOCKS); 511 CHECK_FLAG_VALUE(INLINE_DATA); 512 CHECK_FLAG_VALUE(PROJINHERIT); 513 CHECK_FLAG_VALUE(RESERVED); 514 } 515 516 /* Used to pass group descriptor data when online resize is done */ 517 struct ext4_new_group_input { 518 __u32 group; /* Group number for this data */ 519 __u64 block_bitmap; /* Absolute block number of block bitmap */ 520 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 521 __u64 inode_table; /* Absolute block number of inode table start */ 522 __u32 blocks_count; /* Total number of blocks in this group */ 523 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 524 __u16 unused; 525 }; 526 527 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 528 struct compat_ext4_new_group_input { 529 u32 group; 530 compat_u64 block_bitmap; 531 compat_u64 inode_bitmap; 532 compat_u64 inode_table; 533 u32 blocks_count; 534 u16 reserved_blocks; 535 u16 unused; 536 }; 537 #endif 538 539 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 540 struct ext4_new_group_data { 541 __u32 group; 542 __u64 block_bitmap; 543 __u64 inode_bitmap; 544 __u64 inode_table; 545 __u32 blocks_count; 546 __u16 reserved_blocks; 547 __u16 mdata_blocks; 548 __u32 free_clusters_count; 549 }; 550 551 /* Indexes used to index group tables in ext4_new_group_data */ 552 enum { 553 BLOCK_BITMAP = 0, /* block bitmap */ 554 INODE_BITMAP, /* inode bitmap */ 555 INODE_TABLE, /* inode tables */ 556 GROUP_TABLE_COUNT, 557 }; 558 559 /* 560 * Flags used by ext4_map_blocks() 561 */ 562 /* Allocate any needed blocks and/or convert an unwritten 563 extent to be an initialized ext4 */ 564 #define EXT4_GET_BLOCKS_CREATE 0x0001 565 /* Request the creation of an unwritten extent */ 566 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 567 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 568 EXT4_GET_BLOCKS_CREATE) 569 /* Caller is from the delayed allocation writeout path 570 * finally doing the actual allocation of delayed blocks */ 571 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 572 /* caller is from the direct IO path, request to creation of an 573 unwritten extents if not allocated, split the unwritten 574 extent if blocks has been preallocated already*/ 575 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 576 #define EXT4_GET_BLOCKS_CONVERT 0x0010 577 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 578 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 579 /* Convert extent to initialized after IO complete */ 580 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 581 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 582 /* Eventual metadata allocation (due to growing extent tree) 583 * should not fail, so try to use reserved blocks for that.*/ 584 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 585 /* Don't normalize allocation size (used for fallocate) */ 586 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 587 /* Request will not result in inode size update (user for fallocate) */ 588 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 589 /* Convert written extents to unwritten */ 590 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 591 /* Write zeros to newly created written extents */ 592 #define EXT4_GET_BLOCKS_ZERO 0x0200 593 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 594 EXT4_GET_BLOCKS_ZERO) 595 /* Caller will submit data before dropping transaction handle. This 596 * allows jbd2 to avoid submitting data before commit. */ 597 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 598 599 /* 600 * The bit position of these flags must not overlap with any of the 601 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 602 * read_extent_tree_block(), ext4_split_extent_at(), 603 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 604 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 605 * caching the extents when reading from the extent tree while a 606 * truncate or punch hole operation is in progress. 607 */ 608 #define EXT4_EX_NOCACHE 0x40000000 609 #define EXT4_EX_FORCE_CACHE 0x20000000 610 611 /* 612 * Flags used by ext4_free_blocks 613 */ 614 #define EXT4_FREE_BLOCKS_METADATA 0x0001 615 #define EXT4_FREE_BLOCKS_FORGET 0x0002 616 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 617 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 618 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 619 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 620 621 /* 622 * ioctl commands 623 */ 624 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 625 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 626 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 627 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 628 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 629 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 630 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 631 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 632 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 633 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 634 #define EXT4_IOC_MIGRATE _IO('f', 9) 635 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 636 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 637 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 638 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 639 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 640 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 641 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 642 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 643 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 644 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 645 646 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 647 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 648 649 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 650 651 /* 652 * Flags for going down operation 653 */ 654 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 655 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 656 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 657 658 659 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 660 /* 661 * ioctl commands in 32 bit emulation 662 */ 663 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 664 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 665 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 666 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 667 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 668 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 669 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 670 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 671 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 672 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 673 #endif 674 675 /* Max physical block we can address w/o extents */ 676 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 677 678 /* 679 * Structure of an inode on the disk 680 */ 681 struct ext4_inode { 682 __le16 i_mode; /* File mode */ 683 __le16 i_uid; /* Low 16 bits of Owner Uid */ 684 __le32 i_size_lo; /* Size in bytes */ 685 __le32 i_atime; /* Access time */ 686 __le32 i_ctime; /* Inode Change time */ 687 __le32 i_mtime; /* Modification time */ 688 __le32 i_dtime; /* Deletion Time */ 689 __le16 i_gid; /* Low 16 bits of Group Id */ 690 __le16 i_links_count; /* Links count */ 691 __le32 i_blocks_lo; /* Blocks count */ 692 __le32 i_flags; /* File flags */ 693 union { 694 struct { 695 __le32 l_i_version; 696 } linux1; 697 struct { 698 __u32 h_i_translator; 699 } hurd1; 700 struct { 701 __u32 m_i_reserved1; 702 } masix1; 703 } osd1; /* OS dependent 1 */ 704 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 705 __le32 i_generation; /* File version (for NFS) */ 706 __le32 i_file_acl_lo; /* File ACL */ 707 __le32 i_size_high; 708 __le32 i_obso_faddr; /* Obsoleted fragment address */ 709 union { 710 struct { 711 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 712 __le16 l_i_file_acl_high; 713 __le16 l_i_uid_high; /* these 2 fields */ 714 __le16 l_i_gid_high; /* were reserved2[0] */ 715 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 716 __le16 l_i_reserved; 717 } linux2; 718 struct { 719 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 720 __u16 h_i_mode_high; 721 __u16 h_i_uid_high; 722 __u16 h_i_gid_high; 723 __u32 h_i_author; 724 } hurd2; 725 struct { 726 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 727 __le16 m_i_file_acl_high; 728 __u32 m_i_reserved2[2]; 729 } masix2; 730 } osd2; /* OS dependent 2 */ 731 __le16 i_extra_isize; 732 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 733 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 734 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 735 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 736 __le32 i_crtime; /* File Creation time */ 737 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 738 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 739 __le32 i_projid; /* Project ID */ 740 }; 741 742 struct move_extent { 743 __u32 reserved; /* should be zero */ 744 __u32 donor_fd; /* donor file descriptor */ 745 __u64 orig_start; /* logical start offset in block for orig */ 746 __u64 donor_start; /* logical start offset in block for donor */ 747 __u64 len; /* block length to be moved */ 748 __u64 moved_len; /* moved block length */ 749 }; 750 751 #define EXT4_EPOCH_BITS 2 752 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 753 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 754 755 /* 756 * Extended fields will fit into an inode if the filesystem was formatted 757 * with large inodes (-I 256 or larger) and there are not currently any EAs 758 * consuming all of the available space. For new inodes we always reserve 759 * enough space for the kernel's known extended fields, but for inodes 760 * created with an old kernel this might not have been the case. None of 761 * the extended inode fields is critical for correct filesystem operation. 762 * This macro checks if a certain field fits in the inode. Note that 763 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 764 */ 765 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 766 ((offsetof(typeof(*ext4_inode), field) + \ 767 sizeof((ext4_inode)->field)) \ 768 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 769 (einode)->i_extra_isize)) \ 770 771 /* 772 * We use an encoding that preserves the times for extra epoch "00": 773 * 774 * extra msb of adjust for signed 775 * epoch 32-bit 32-bit tv_sec to 776 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 777 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 778 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 779 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 780 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 781 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 782 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 783 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 784 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 785 * 786 * Note that previous versions of the kernel on 64-bit systems would 787 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 788 * 1970. e2fsck will correct this, assuming that it is run on the 789 * affected filesystem before 2242. 790 */ 791 792 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 793 { 794 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 795 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 796 } 797 798 static inline void ext4_decode_extra_time(struct timespec64 *time, 799 __le32 extra) 800 { 801 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 802 803 #if 1 804 /* Handle legacy encoding of pre-1970 dates with epoch 805 * bits 1,1. (This backwards compatibility may be removed 806 * at the discretion of the ext4 developers.) 807 */ 808 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 809 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 810 extra_bits = 0; 811 time->tv_sec += extra_bits << 32; 812 #else 813 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 814 #endif 815 } 816 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 817 } 818 819 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 820 do { \ 821 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 822 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 823 (raw_inode)->xtime ## _extra = \ 824 ext4_encode_extra_time(&(inode)->xtime); \ 825 } \ 826 } while (0) 827 828 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 829 do { \ 830 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 831 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 832 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 833 (raw_inode)->xtime ## _extra = \ 834 ext4_encode_extra_time(&(einode)->xtime); \ 835 } while (0) 836 837 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 838 do { \ 839 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 840 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 841 ext4_decode_extra_time(&(inode)->xtime, \ 842 raw_inode->xtime ## _extra); \ 843 } \ 844 else \ 845 (inode)->xtime.tv_nsec = 0; \ 846 } while (0) 847 848 849 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 850 do { \ 851 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 852 (einode)->xtime.tv_sec = \ 853 (signed)le32_to_cpu((raw_inode)->xtime); \ 854 else \ 855 (einode)->xtime.tv_sec = 0; \ 856 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 857 ext4_decode_extra_time(&(einode)->xtime, \ 858 raw_inode->xtime ## _extra); \ 859 else \ 860 (einode)->xtime.tv_nsec = 0; \ 861 } while (0) 862 863 #define i_disk_version osd1.linux1.l_i_version 864 865 #if defined(__KERNEL__) || defined(__linux__) 866 #define i_reserved1 osd1.linux1.l_i_reserved1 867 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 868 #define i_blocks_high osd2.linux2.l_i_blocks_high 869 #define i_uid_low i_uid 870 #define i_gid_low i_gid 871 #define i_uid_high osd2.linux2.l_i_uid_high 872 #define i_gid_high osd2.linux2.l_i_gid_high 873 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 874 875 #elif defined(__GNU__) 876 877 #define i_translator osd1.hurd1.h_i_translator 878 #define i_uid_high osd2.hurd2.h_i_uid_high 879 #define i_gid_high osd2.hurd2.h_i_gid_high 880 #define i_author osd2.hurd2.h_i_author 881 882 #elif defined(__masix__) 883 884 #define i_reserved1 osd1.masix1.m_i_reserved1 885 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 886 #define i_reserved2 osd2.masix2.m_i_reserved2 887 888 #endif /* defined(__KERNEL__) || defined(__linux__) */ 889 890 #include "extents_status.h" 891 892 /* 893 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 894 * 895 * These are needed to avoid lockdep false positives when we need to 896 * allocate blocks to the quota inode during ext4_map_blocks(), while 897 * holding i_data_sem for a normal (non-quota) inode. Since we don't 898 * do quota tracking for the quota inode, this avoids deadlock (as 899 * well as infinite recursion, since it isn't turtles all the way 900 * down...) 901 * 902 * I_DATA_SEM_NORMAL - Used for most inodes 903 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 904 * where the second inode has larger inode number 905 * than the first 906 * I_DATA_SEM_QUOTA - Used for quota inodes only 907 */ 908 enum { 909 I_DATA_SEM_NORMAL = 0, 910 I_DATA_SEM_OTHER, 911 I_DATA_SEM_QUOTA, 912 }; 913 914 915 /* 916 * fourth extended file system inode data in memory 917 */ 918 struct ext4_inode_info { 919 __le32 i_data[15]; /* unconverted */ 920 __u32 i_dtime; 921 ext4_fsblk_t i_file_acl; 922 923 /* 924 * i_block_group is the number of the block group which contains 925 * this file's inode. Constant across the lifetime of the inode, 926 * it is used for making block allocation decisions - we try to 927 * place a file's data blocks near its inode block, and new inodes 928 * near to their parent directory's inode. 929 */ 930 ext4_group_t i_block_group; 931 ext4_lblk_t i_dir_start_lookup; 932 #if (BITS_PER_LONG < 64) 933 unsigned long i_state_flags; /* Dynamic state flags */ 934 #endif 935 unsigned long i_flags; 936 937 /* 938 * Extended attributes can be read independently of the main file 939 * data. Taking i_mutex even when reading would cause contention 940 * between readers of EAs and writers of regular file data, so 941 * instead we synchronize on xattr_sem when reading or changing 942 * EAs. 943 */ 944 struct rw_semaphore xattr_sem; 945 946 struct list_head i_orphan; /* unlinked but open inodes */ 947 948 /* 949 * i_disksize keeps track of what the inode size is ON DISK, not 950 * in memory. During truncate, i_size is set to the new size by 951 * the VFS prior to calling ext4_truncate(), but the filesystem won't 952 * set i_disksize to 0 until the truncate is actually under way. 953 * 954 * The intent is that i_disksize always represents the blocks which 955 * are used by this file. This allows recovery to restart truncate 956 * on orphans if we crash during truncate. We actually write i_disksize 957 * into the on-disk inode when writing inodes out, instead of i_size. 958 * 959 * The only time when i_disksize and i_size may be different is when 960 * a truncate is in progress. The only things which change i_disksize 961 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 962 */ 963 loff_t i_disksize; 964 965 /* 966 * i_data_sem is for serialising ext4_truncate() against 967 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 968 * data tree are chopped off during truncate. We can't do that in 969 * ext4 because whenever we perform intermediate commits during 970 * truncate, the inode and all the metadata blocks *must* be in a 971 * consistent state which allows truncation of the orphans to restart 972 * during recovery. Hence we must fix the get_block-vs-truncate race 973 * by other means, so we have i_data_sem. 974 */ 975 struct rw_semaphore i_data_sem; 976 /* 977 * i_mmap_sem is for serializing page faults with truncate / punch hole 978 * operations. We have to make sure that new page cannot be faulted in 979 * a section of the inode that is being punched. We cannot easily use 980 * i_data_sem for this since we need protection for the whole punch 981 * operation and i_data_sem ranks below transaction start so we have 982 * to occasionally drop it. 983 */ 984 struct rw_semaphore i_mmap_sem; 985 struct inode vfs_inode; 986 struct jbd2_inode *jinode; 987 988 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 989 990 /* 991 * File creation time. Its function is same as that of 992 * struct timespec64 i_{a,c,m}time in the generic inode. 993 */ 994 struct timespec64 i_crtime; 995 996 /* mballoc */ 997 struct list_head i_prealloc_list; 998 spinlock_t i_prealloc_lock; 999 1000 /* extents status tree */ 1001 struct ext4_es_tree i_es_tree; 1002 rwlock_t i_es_lock; 1003 struct list_head i_es_list; 1004 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1005 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1006 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1007 extents to shrink. Protected by 1008 i_es_lock */ 1009 1010 /* ialloc */ 1011 ext4_group_t i_last_alloc_group; 1012 1013 /* allocation reservation info for delalloc */ 1014 /* In case of bigalloc, this refer to clusters rather than blocks */ 1015 unsigned int i_reserved_data_blocks; 1016 ext4_lblk_t i_da_metadata_calc_last_lblock; 1017 int i_da_metadata_calc_len; 1018 1019 /* on-disk additional length */ 1020 __u16 i_extra_isize; 1021 1022 /* Indicate the inline data space. */ 1023 u16 i_inline_off; 1024 u16 i_inline_size; 1025 1026 #ifdef CONFIG_QUOTA 1027 /* quota space reservation, managed internally by quota code */ 1028 qsize_t i_reserved_quota; 1029 #endif 1030 1031 /* Lock protecting lists below */ 1032 spinlock_t i_completed_io_lock; 1033 /* 1034 * Completed IOs that need unwritten extents handling and have 1035 * transaction reserved 1036 */ 1037 struct list_head i_rsv_conversion_list; 1038 struct work_struct i_rsv_conversion_work; 1039 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1040 1041 spinlock_t i_block_reservation_lock; 1042 1043 /* 1044 * Transactions that contain inode's metadata needed to complete 1045 * fsync and fdatasync, respectively. 1046 */ 1047 tid_t i_sync_tid; 1048 tid_t i_datasync_tid; 1049 1050 #ifdef CONFIG_QUOTA 1051 struct dquot *i_dquot[MAXQUOTAS]; 1052 #endif 1053 1054 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1055 __u32 i_csum_seed; 1056 1057 kprojid_t i_projid; 1058 }; 1059 1060 /* 1061 * File system states 1062 */ 1063 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1064 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1065 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1066 1067 /* 1068 * Misc. filesystem flags 1069 */ 1070 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1071 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1072 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1073 1074 /* 1075 * Mount flags set via mount options or defaults 1076 */ 1077 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1078 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1079 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1080 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1081 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1082 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1083 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1084 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1085 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1086 #ifdef CONFIG_FS_DAX 1087 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1088 #else 1089 #define EXT4_MOUNT_DAX 0 1090 #endif 1091 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1092 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1093 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1094 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1095 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1096 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1097 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1098 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1099 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1100 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1101 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1102 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1103 * enable enforcement for hidden 1104 * quota files */ 1105 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1106 * enforcement for hidden quota 1107 * files */ 1108 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1109 * enforcement */ 1110 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1111 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1112 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1113 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1114 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1115 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1116 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1117 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1118 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1119 1120 /* 1121 * Mount flags set either automatically (could not be set by mount option) 1122 * based on per file system feature or property or in special cases such as 1123 * distinguishing between explicit mount option definition and default. 1124 */ 1125 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1126 specified delalloc */ 1127 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1128 size of blocksize * 8 1129 blocks */ 1130 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1131 file systems */ 1132 1133 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1134 specified journal checksum */ 1135 1136 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1137 ~EXT4_MOUNT_##opt 1138 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1139 EXT4_MOUNT_##opt 1140 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1141 EXT4_MOUNT_##opt) 1142 1143 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1144 ~EXT4_MOUNT2_##opt 1145 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1146 EXT4_MOUNT2_##opt 1147 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1148 EXT4_MOUNT2_##opt) 1149 1150 #define ext4_test_and_set_bit __test_and_set_bit_le 1151 #define ext4_set_bit __set_bit_le 1152 #define ext4_set_bit_atomic ext2_set_bit_atomic 1153 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1154 #define ext4_clear_bit __clear_bit_le 1155 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1156 #define ext4_test_bit test_bit_le 1157 #define ext4_find_next_zero_bit find_next_zero_bit_le 1158 #define ext4_find_next_bit find_next_bit_le 1159 1160 extern void ext4_set_bits(void *bm, int cur, int len); 1161 1162 /* 1163 * Maximal mount counts between two filesystem checks 1164 */ 1165 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1166 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1167 1168 /* 1169 * Behaviour when detecting errors 1170 */ 1171 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1172 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1173 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1174 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1175 1176 /* Metadata checksum algorithm codes */ 1177 #define EXT4_CRC32C_CHKSUM 1 1178 1179 /* 1180 * Structure of the super block 1181 */ 1182 struct ext4_super_block { 1183 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1184 __le32 s_blocks_count_lo; /* Blocks count */ 1185 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1186 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1187 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1188 __le32 s_first_data_block; /* First Data Block */ 1189 __le32 s_log_block_size; /* Block size */ 1190 __le32 s_log_cluster_size; /* Allocation cluster size */ 1191 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1192 __le32 s_clusters_per_group; /* # Clusters per group */ 1193 __le32 s_inodes_per_group; /* # Inodes per group */ 1194 __le32 s_mtime; /* Mount time */ 1195 /*30*/ __le32 s_wtime; /* Write time */ 1196 __le16 s_mnt_count; /* Mount count */ 1197 __le16 s_max_mnt_count; /* Maximal mount count */ 1198 __le16 s_magic; /* Magic signature */ 1199 __le16 s_state; /* File system state */ 1200 __le16 s_errors; /* Behaviour when detecting errors */ 1201 __le16 s_minor_rev_level; /* minor revision level */ 1202 /*40*/ __le32 s_lastcheck; /* time of last check */ 1203 __le32 s_checkinterval; /* max. time between checks */ 1204 __le32 s_creator_os; /* OS */ 1205 __le32 s_rev_level; /* Revision level */ 1206 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1207 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1208 /* 1209 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1210 * 1211 * Note: the difference between the compatible feature set and 1212 * the incompatible feature set is that if there is a bit set 1213 * in the incompatible feature set that the kernel doesn't 1214 * know about, it should refuse to mount the filesystem. 1215 * 1216 * e2fsck's requirements are more strict; if it doesn't know 1217 * about a feature in either the compatible or incompatible 1218 * feature set, it must abort and not try to meddle with 1219 * things it doesn't understand... 1220 */ 1221 __le32 s_first_ino; /* First non-reserved inode */ 1222 __le16 s_inode_size; /* size of inode structure */ 1223 __le16 s_block_group_nr; /* block group # of this superblock */ 1224 __le32 s_feature_compat; /* compatible feature set */ 1225 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1226 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1227 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1228 /*78*/ char s_volume_name[16]; /* volume name */ 1229 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1230 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1231 /* 1232 * Performance hints. Directory preallocation should only 1233 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1234 */ 1235 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1236 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1237 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1238 /* 1239 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1240 */ 1241 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1242 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1243 __le32 s_journal_dev; /* device number of journal file */ 1244 __le32 s_last_orphan; /* start of list of inodes to delete */ 1245 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1246 __u8 s_def_hash_version; /* Default hash version to use */ 1247 __u8 s_jnl_backup_type; 1248 __le16 s_desc_size; /* size of group descriptor */ 1249 /*100*/ __le32 s_default_mount_opts; 1250 __le32 s_first_meta_bg; /* First metablock block group */ 1251 __le32 s_mkfs_time; /* When the filesystem was created */ 1252 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1253 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1254 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1255 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1256 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1257 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1258 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1259 __le32 s_flags; /* Miscellaneous flags */ 1260 __le16 s_raid_stride; /* RAID stride */ 1261 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1262 __le64 s_mmp_block; /* Block for multi-mount protection */ 1263 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1264 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1265 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1266 __u8 s_encryption_level; /* versioning level for encryption */ 1267 __u8 s_reserved_pad; /* Padding to next 32bits */ 1268 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1269 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1270 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1271 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1272 snapshot's future use */ 1273 __le32 s_snapshot_list; /* inode number of the head of the 1274 on-disk snapshot list */ 1275 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1276 __le32 s_error_count; /* number of fs errors */ 1277 __le32 s_first_error_time; /* first time an error happened */ 1278 __le32 s_first_error_ino; /* inode involved in first error */ 1279 __le64 s_first_error_block; /* block involved of first error */ 1280 __u8 s_first_error_func[32]; /* function where the error happened */ 1281 __le32 s_first_error_line; /* line number where error happened */ 1282 __le32 s_last_error_time; /* most recent time of an error */ 1283 __le32 s_last_error_ino; /* inode involved in last error */ 1284 __le32 s_last_error_line; /* line number where error happened */ 1285 __le64 s_last_error_block; /* block involved of last error */ 1286 __u8 s_last_error_func[32]; /* function where the error happened */ 1287 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1288 __u8 s_mount_opts[64]; 1289 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1290 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1291 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1292 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1293 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1294 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1295 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1296 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1297 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1298 __le32 s_reserved[98]; /* Padding to the end of the block */ 1299 __le32 s_checksum; /* crc32c(superblock) */ 1300 }; 1301 1302 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1303 1304 #ifdef __KERNEL__ 1305 1306 /* 1307 * run-time mount flags 1308 */ 1309 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1310 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1311 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1312 1313 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1314 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1315 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1316 #else 1317 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1318 #endif 1319 1320 /* Number of quota types we support */ 1321 #define EXT4_MAXQUOTAS 3 1322 1323 /* 1324 * fourth extended-fs super-block data in memory 1325 */ 1326 struct ext4_sb_info { 1327 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1328 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1329 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1330 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1331 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1332 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1333 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1334 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1335 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1336 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1337 unsigned long s_overhead; /* # of fs overhead clusters */ 1338 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1339 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1340 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1341 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1342 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1343 struct buffer_head **s_group_desc; 1344 unsigned int s_mount_opt; 1345 unsigned int s_mount_opt2; 1346 unsigned int s_mount_flags; 1347 unsigned int s_def_mount_opt; 1348 ext4_fsblk_t s_sb_block; 1349 atomic64_t s_resv_clusters; 1350 kuid_t s_resuid; 1351 kgid_t s_resgid; 1352 unsigned short s_mount_state; 1353 unsigned short s_pad; 1354 int s_addr_per_block_bits; 1355 int s_desc_per_block_bits; 1356 int s_inode_size; 1357 int s_first_ino; 1358 unsigned int s_inode_readahead_blks; 1359 unsigned int s_inode_goal; 1360 u32 s_hash_seed[4]; 1361 int s_def_hash_version; 1362 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1363 struct percpu_counter s_freeclusters_counter; 1364 struct percpu_counter s_freeinodes_counter; 1365 struct percpu_counter s_dirs_counter; 1366 struct percpu_counter s_dirtyclusters_counter; 1367 struct blockgroup_lock *s_blockgroup_lock; 1368 struct proc_dir_entry *s_proc; 1369 struct kobject s_kobj; 1370 struct completion s_kobj_unregister; 1371 struct super_block *s_sb; 1372 1373 /* Journaling */ 1374 struct journal_s *s_journal; 1375 struct list_head s_orphan; 1376 struct mutex s_orphan_lock; 1377 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1378 unsigned long s_commit_interval; 1379 u32 s_max_batch_time; 1380 u32 s_min_batch_time; 1381 struct block_device *journal_bdev; 1382 #ifdef CONFIG_QUOTA 1383 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1384 int s_jquota_fmt; /* Format of quota to use */ 1385 #endif 1386 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1387 struct rb_root system_blks; 1388 1389 #ifdef EXTENTS_STATS 1390 /* ext4 extents stats */ 1391 unsigned long s_ext_min; 1392 unsigned long s_ext_max; 1393 unsigned long s_depth_max; 1394 spinlock_t s_ext_stats_lock; 1395 unsigned long s_ext_blocks; 1396 unsigned long s_ext_extents; 1397 #endif 1398 1399 /* for buddy allocator */ 1400 struct ext4_group_info ***s_group_info; 1401 struct inode *s_buddy_cache; 1402 spinlock_t s_md_lock; 1403 unsigned short *s_mb_offsets; 1404 unsigned int *s_mb_maxs; 1405 unsigned int s_group_info_size; 1406 unsigned int s_mb_free_pending; 1407 struct list_head s_freed_data_list; /* List of blocks to be freed 1408 after commit completed */ 1409 1410 /* tunables */ 1411 unsigned long s_stripe; 1412 unsigned int s_mb_stream_request; 1413 unsigned int s_mb_max_to_scan; 1414 unsigned int s_mb_min_to_scan; 1415 unsigned int s_mb_stats; 1416 unsigned int s_mb_order2_reqs; 1417 unsigned int s_mb_group_prealloc; 1418 unsigned int s_max_dir_size_kb; 1419 /* where last allocation was done - for stream allocation */ 1420 unsigned long s_mb_last_group; 1421 unsigned long s_mb_last_start; 1422 1423 /* stats for buddy allocator */ 1424 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1425 atomic_t s_bal_success; /* we found long enough chunks */ 1426 atomic_t s_bal_allocated; /* in blocks */ 1427 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1428 atomic_t s_bal_goals; /* goal hits */ 1429 atomic_t s_bal_breaks; /* too long searches */ 1430 atomic_t s_bal_2orders; /* 2^order hits */ 1431 spinlock_t s_bal_lock; 1432 unsigned long s_mb_buddies_generated; 1433 unsigned long long s_mb_generation_time; 1434 atomic_t s_mb_lost_chunks; 1435 atomic_t s_mb_preallocated; 1436 atomic_t s_mb_discarded; 1437 atomic_t s_lock_busy; 1438 1439 /* locality groups */ 1440 struct ext4_locality_group __percpu *s_locality_groups; 1441 1442 /* for write statistics */ 1443 unsigned long s_sectors_written_start; 1444 u64 s_kbytes_written; 1445 1446 /* the size of zero-out chunk */ 1447 unsigned int s_extent_max_zeroout_kb; 1448 1449 unsigned int s_log_groups_per_flex; 1450 struct flex_groups *s_flex_groups; 1451 ext4_group_t s_flex_groups_allocated; 1452 1453 /* workqueue for reserved extent conversions (buffered io) */ 1454 struct workqueue_struct *rsv_conversion_wq; 1455 1456 /* timer for periodic error stats printing */ 1457 struct timer_list s_err_report; 1458 1459 /* Lazy inode table initialization info */ 1460 struct ext4_li_request *s_li_request; 1461 /* Wait multiplier for lazy initialization thread */ 1462 unsigned int s_li_wait_mult; 1463 1464 /* Kernel thread for multiple mount protection */ 1465 struct task_struct *s_mmp_tsk; 1466 1467 /* record the last minlen when FITRIM is called. */ 1468 atomic_t s_last_trim_minblks; 1469 1470 /* Reference to checksum algorithm driver via cryptoapi */ 1471 struct crypto_shash *s_chksum_driver; 1472 1473 /* Precomputed FS UUID checksum for seeding other checksums */ 1474 __u32 s_csum_seed; 1475 1476 /* Reclaim extents from extent status tree */ 1477 struct shrinker s_es_shrinker; 1478 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1479 long s_es_nr_inode; 1480 struct ext4_es_stats s_es_stats; 1481 struct mb_cache *s_ea_block_cache; 1482 struct mb_cache *s_ea_inode_cache; 1483 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1484 1485 /* Ratelimit ext4 messages. */ 1486 struct ratelimit_state s_err_ratelimit_state; 1487 struct ratelimit_state s_warning_ratelimit_state; 1488 struct ratelimit_state s_msg_ratelimit_state; 1489 1490 /* Barrier between changing inodes' journal flags and writepages ops. */ 1491 struct percpu_rw_semaphore s_journal_flag_rwsem; 1492 struct dax_device *s_daxdev; 1493 }; 1494 1495 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1496 { 1497 return sb->s_fs_info; 1498 } 1499 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1500 { 1501 return container_of(inode, struct ext4_inode_info, vfs_inode); 1502 } 1503 1504 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1505 { 1506 return ino == EXT4_ROOT_INO || 1507 (ino >= EXT4_FIRST_INO(sb) && 1508 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1509 } 1510 1511 /* 1512 * Inode dynamic state flags 1513 */ 1514 enum { 1515 EXT4_STATE_JDATA, /* journaled data exists */ 1516 EXT4_STATE_NEW, /* inode is newly created */ 1517 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1518 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1519 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1520 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1521 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1522 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1523 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1524 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1525 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1526 }; 1527 1528 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1529 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1530 { \ 1531 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1532 } \ 1533 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1534 { \ 1535 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1536 } \ 1537 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1538 { \ 1539 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1540 } 1541 1542 /* Add these declarations here only so that these functions can be 1543 * found by name. Otherwise, they are very hard to locate. */ 1544 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1545 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1546 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1547 EXT4_INODE_BIT_FNS(flag, flags, 0) 1548 1549 /* Add these declarations here only so that these functions can be 1550 * found by name. Otherwise, they are very hard to locate. */ 1551 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1552 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1553 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1554 #if (BITS_PER_LONG < 64) 1555 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1556 1557 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1558 { 1559 (ei)->i_state_flags = 0; 1560 } 1561 #else 1562 EXT4_INODE_BIT_FNS(state, flags, 32) 1563 1564 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1565 { 1566 /* We depend on the fact that callers will set i_flags */ 1567 } 1568 #endif 1569 #else 1570 /* Assume that user mode programs are passing in an ext4fs superblock, not 1571 * a kernel struct super_block. This will allow us to call the feature-test 1572 * macros from user land. */ 1573 #define EXT4_SB(sb) (sb) 1574 #endif 1575 1576 /* 1577 * Returns true if the inode is inode is encrypted 1578 */ 1579 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1580 1581 /* 1582 * Codes for operating systems 1583 */ 1584 #define EXT4_OS_LINUX 0 1585 #define EXT4_OS_HURD 1 1586 #define EXT4_OS_MASIX 2 1587 #define EXT4_OS_FREEBSD 3 1588 #define EXT4_OS_LITES 4 1589 1590 /* 1591 * Revision levels 1592 */ 1593 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1594 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1595 1596 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1597 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1598 1599 #define EXT4_GOOD_OLD_INODE_SIZE 128 1600 1601 /* 1602 * Feature set definitions 1603 */ 1604 1605 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1606 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1607 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1608 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1609 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1610 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1611 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1612 1613 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1614 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1615 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1616 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1617 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1618 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1619 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1620 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1621 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1622 /* 1623 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1624 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1625 * all other data structures' checksums. However, the METADATA_CSUM and 1626 * GDT_CSUM bits are mutually exclusive. 1627 */ 1628 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1629 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1630 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1631 1632 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1633 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1634 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1635 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1636 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1637 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1638 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1639 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1640 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1641 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1642 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1643 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1644 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1645 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1646 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1647 1648 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1649 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1650 { \ 1651 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1652 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1653 } \ 1654 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1655 { \ 1656 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1657 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1658 } \ 1659 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1660 { \ 1661 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1662 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1663 } 1664 1665 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1666 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1667 { \ 1668 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1669 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1670 } \ 1671 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1672 { \ 1673 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1674 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1675 } \ 1676 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1677 { \ 1678 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1679 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1680 } 1681 1682 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1683 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1684 { \ 1685 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1686 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1687 } \ 1688 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1689 { \ 1690 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1691 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1692 } \ 1693 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1694 { \ 1695 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1696 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1697 } 1698 1699 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1700 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1701 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1702 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1703 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1704 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1705 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1706 1707 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1708 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1709 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1710 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1711 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1712 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1713 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1714 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1715 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1716 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1717 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1718 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1719 1720 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1721 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1722 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1723 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1724 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1725 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1726 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1727 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1728 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1729 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1730 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1731 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1732 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1733 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1734 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1735 1736 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1737 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1738 EXT4_FEATURE_INCOMPAT_META_BG) 1739 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1740 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1741 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1742 1743 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1744 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1745 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1746 EXT4_FEATURE_INCOMPAT_META_BG) 1747 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1748 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1749 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1750 1751 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1752 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1753 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1754 EXT4_FEATURE_INCOMPAT_META_BG| \ 1755 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1756 EXT4_FEATURE_INCOMPAT_64BIT| \ 1757 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1758 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1759 EXT4_FEATURE_INCOMPAT_MMP | \ 1760 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1761 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1762 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1763 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1764 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1765 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1766 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1767 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1768 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1769 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1770 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1771 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1772 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1773 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1774 EXT4_FEATURE_RO_COMPAT_PROJECT) 1775 1776 #define EXTN_FEATURE_FUNCS(ver) \ 1777 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1778 { \ 1779 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1780 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1781 } \ 1782 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1783 { \ 1784 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1785 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1786 } \ 1787 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1788 { \ 1789 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1790 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1791 } 1792 1793 EXTN_FEATURE_FUNCS(2) 1794 EXTN_FEATURE_FUNCS(3) 1795 EXTN_FEATURE_FUNCS(4) 1796 1797 static inline bool ext4_has_compat_features(struct super_block *sb) 1798 { 1799 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1800 } 1801 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1802 { 1803 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1804 } 1805 static inline bool ext4_has_incompat_features(struct super_block *sb) 1806 { 1807 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1808 } 1809 1810 /* 1811 * Superblock flags 1812 */ 1813 #define EXT4_FLAGS_RESIZING 0 1814 #define EXT4_FLAGS_SHUTDOWN 1 1815 1816 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1817 { 1818 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1819 } 1820 1821 1822 /* 1823 * Default values for user and/or group using reserved blocks 1824 */ 1825 #define EXT4_DEF_RESUID 0 1826 #define EXT4_DEF_RESGID 0 1827 1828 /* 1829 * Default project ID 1830 */ 1831 #define EXT4_DEF_PROJID 0 1832 1833 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1834 1835 /* 1836 * Default mount options 1837 */ 1838 #define EXT4_DEFM_DEBUG 0x0001 1839 #define EXT4_DEFM_BSDGROUPS 0x0002 1840 #define EXT4_DEFM_XATTR_USER 0x0004 1841 #define EXT4_DEFM_ACL 0x0008 1842 #define EXT4_DEFM_UID16 0x0010 1843 #define EXT4_DEFM_JMODE 0x0060 1844 #define EXT4_DEFM_JMODE_DATA 0x0020 1845 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1846 #define EXT4_DEFM_JMODE_WBACK 0x0060 1847 #define EXT4_DEFM_NOBARRIER 0x0100 1848 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1849 #define EXT4_DEFM_DISCARD 0x0400 1850 #define EXT4_DEFM_NODELALLOC 0x0800 1851 1852 /* 1853 * Default journal batch times 1854 */ 1855 #define EXT4_DEF_MIN_BATCH_TIME 0 1856 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1857 1858 /* 1859 * Minimum number of groups in a flexgroup before we separate out 1860 * directories into the first block group of a flexgroup 1861 */ 1862 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1863 1864 /* 1865 * Structure of a directory entry 1866 */ 1867 #define EXT4_NAME_LEN 255 1868 1869 struct ext4_dir_entry { 1870 __le32 inode; /* Inode number */ 1871 __le16 rec_len; /* Directory entry length */ 1872 __le16 name_len; /* Name length */ 1873 char name[EXT4_NAME_LEN]; /* File name */ 1874 }; 1875 1876 /* 1877 * The new version of the directory entry. Since EXT4 structures are 1878 * stored in intel byte order, and the name_len field could never be 1879 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1880 * file_type field. 1881 */ 1882 struct ext4_dir_entry_2 { 1883 __le32 inode; /* Inode number */ 1884 __le16 rec_len; /* Directory entry length */ 1885 __u8 name_len; /* Name length */ 1886 __u8 file_type; 1887 char name[EXT4_NAME_LEN]; /* File name */ 1888 }; 1889 1890 /* 1891 * This is a bogus directory entry at the end of each leaf block that 1892 * records checksums. 1893 */ 1894 struct ext4_dir_entry_tail { 1895 __le32 det_reserved_zero1; /* Pretend to be unused */ 1896 __le16 det_rec_len; /* 12 */ 1897 __u8 det_reserved_zero2; /* Zero name length */ 1898 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1899 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1900 }; 1901 1902 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1903 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1904 ((blocksize) - \ 1905 sizeof(struct ext4_dir_entry_tail)))) 1906 1907 /* 1908 * Ext4 directory file types. Only the low 3 bits are used. The 1909 * other bits are reserved for now. 1910 */ 1911 #define EXT4_FT_UNKNOWN 0 1912 #define EXT4_FT_REG_FILE 1 1913 #define EXT4_FT_DIR 2 1914 #define EXT4_FT_CHRDEV 3 1915 #define EXT4_FT_BLKDEV 4 1916 #define EXT4_FT_FIFO 5 1917 #define EXT4_FT_SOCK 6 1918 #define EXT4_FT_SYMLINK 7 1919 1920 #define EXT4_FT_MAX 8 1921 1922 #define EXT4_FT_DIR_CSUM 0xDE 1923 1924 /* 1925 * EXT4_DIR_PAD defines the directory entries boundaries 1926 * 1927 * NOTE: It must be a multiple of 4 1928 */ 1929 #define EXT4_DIR_PAD 4 1930 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1931 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1932 ~EXT4_DIR_ROUND) 1933 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1934 1935 /* 1936 * If we ever get support for fs block sizes > page_size, we'll need 1937 * to remove the #if statements in the next two functions... 1938 */ 1939 static inline unsigned int 1940 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1941 { 1942 unsigned len = le16_to_cpu(dlen); 1943 1944 #if (PAGE_SIZE >= 65536) 1945 if (len == EXT4_MAX_REC_LEN || len == 0) 1946 return blocksize; 1947 return (len & 65532) | ((len & 3) << 16); 1948 #else 1949 return len; 1950 #endif 1951 } 1952 1953 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1954 { 1955 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1956 BUG(); 1957 #if (PAGE_SIZE >= 65536) 1958 if (len < 65536) 1959 return cpu_to_le16(len); 1960 if (len == blocksize) { 1961 if (blocksize == 65536) 1962 return cpu_to_le16(EXT4_MAX_REC_LEN); 1963 else 1964 return cpu_to_le16(0); 1965 } 1966 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1967 #else 1968 return cpu_to_le16(len); 1969 #endif 1970 } 1971 1972 /* 1973 * Hash Tree Directory indexing 1974 * (c) Daniel Phillips, 2001 1975 */ 1976 1977 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1978 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1979 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 1980 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 1981 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1982 1983 /* Legal values for the dx_root hash_version field: */ 1984 1985 #define DX_HASH_LEGACY 0 1986 #define DX_HASH_HALF_MD4 1 1987 #define DX_HASH_TEA 2 1988 #define DX_HASH_LEGACY_UNSIGNED 3 1989 #define DX_HASH_HALF_MD4_UNSIGNED 4 1990 #define DX_HASH_TEA_UNSIGNED 5 1991 1992 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1993 const void *address, unsigned int length) 1994 { 1995 struct { 1996 struct shash_desc shash; 1997 char ctx[4]; 1998 } desc; 1999 2000 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2001 2002 desc.shash.tfm = sbi->s_chksum_driver; 2003 desc.shash.flags = 0; 2004 *(u32 *)desc.ctx = crc; 2005 2006 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2007 2008 return *(u32 *)desc.ctx; 2009 } 2010 2011 #ifdef __KERNEL__ 2012 2013 /* hash info structure used by the directory hash */ 2014 struct dx_hash_info 2015 { 2016 u32 hash; 2017 u32 minor_hash; 2018 int hash_version; 2019 u32 *seed; 2020 }; 2021 2022 2023 /* 32 and 64 bit signed EOF for dx directories */ 2024 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2025 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2026 2027 2028 /* 2029 * Control parameters used by ext4_htree_next_block 2030 */ 2031 #define HASH_NB_ALWAYS 1 2032 2033 struct ext4_filename { 2034 const struct qstr *usr_fname; 2035 struct fscrypt_str disk_name; 2036 struct dx_hash_info hinfo; 2037 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2038 struct fscrypt_str crypto_buf; 2039 #endif 2040 }; 2041 2042 #define fname_name(p) ((p)->disk_name.name) 2043 #define fname_len(p) ((p)->disk_name.len) 2044 2045 /* 2046 * Describe an inode's exact location on disk and in memory 2047 */ 2048 struct ext4_iloc 2049 { 2050 struct buffer_head *bh; 2051 unsigned long offset; 2052 ext4_group_t block_group; 2053 }; 2054 2055 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2056 { 2057 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2058 } 2059 2060 static inline bool ext4_is_quota_file(struct inode *inode) 2061 { 2062 return IS_NOQUOTA(inode) && 2063 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2064 } 2065 2066 /* 2067 * This structure is stuffed into the struct file's private_data field 2068 * for directories. It is where we put information so that we can do 2069 * readdir operations in hash tree order. 2070 */ 2071 struct dir_private_info { 2072 struct rb_root root; 2073 struct rb_node *curr_node; 2074 struct fname *extra_fname; 2075 loff_t last_pos; 2076 __u32 curr_hash; 2077 __u32 curr_minor_hash; 2078 __u32 next_hash; 2079 }; 2080 2081 /* calculate the first block number of the group */ 2082 static inline ext4_fsblk_t 2083 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2084 { 2085 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2086 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2087 } 2088 2089 /* 2090 * Special error return code only used by dx_probe() and its callers. 2091 */ 2092 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2093 2094 /* htree levels for ext4 */ 2095 #define EXT4_HTREE_LEVEL_COMPAT 2 2096 #define EXT4_HTREE_LEVEL 3 2097 2098 static inline int ext4_dir_htree_level(struct super_block *sb) 2099 { 2100 return ext4_has_feature_largedir(sb) ? 2101 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2102 } 2103 2104 /* 2105 * Timeout and state flag for lazy initialization inode thread. 2106 */ 2107 #define EXT4_DEF_LI_WAIT_MULT 10 2108 #define EXT4_DEF_LI_MAX_START_DELAY 5 2109 #define EXT4_LAZYINIT_QUIT 0x0001 2110 #define EXT4_LAZYINIT_RUNNING 0x0002 2111 2112 /* 2113 * Lazy inode table initialization info 2114 */ 2115 struct ext4_lazy_init { 2116 unsigned long li_state; 2117 struct list_head li_request_list; 2118 struct mutex li_list_mtx; 2119 }; 2120 2121 struct ext4_li_request { 2122 struct super_block *lr_super; 2123 struct ext4_sb_info *lr_sbi; 2124 ext4_group_t lr_next_group; 2125 struct list_head lr_request; 2126 unsigned long lr_next_sched; 2127 unsigned long lr_timeout; 2128 }; 2129 2130 struct ext4_features { 2131 struct kobject f_kobj; 2132 struct completion f_kobj_unregister; 2133 }; 2134 2135 /* 2136 * This structure will be used for multiple mount protection. It will be 2137 * written into the block number saved in the s_mmp_block field in the 2138 * superblock. Programs that check MMP should assume that if 2139 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2140 * to use the filesystem, regardless of how old the timestamp is. 2141 */ 2142 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2143 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2144 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2145 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2146 2147 struct mmp_struct { 2148 __le32 mmp_magic; /* Magic number for MMP */ 2149 __le32 mmp_seq; /* Sequence no. updated periodically */ 2150 2151 /* 2152 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2153 * purposes and do not affect the correctness of the algorithm 2154 */ 2155 __le64 mmp_time; /* Time last updated */ 2156 char mmp_nodename[64]; /* Node which last updated MMP block */ 2157 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2158 2159 /* 2160 * mmp_check_interval is used to verify if the MMP block has been 2161 * updated on the block device. The value is updated based on the 2162 * maximum time to write the MMP block during an update cycle. 2163 */ 2164 __le16 mmp_check_interval; 2165 2166 __le16 mmp_pad1; 2167 __le32 mmp_pad2[226]; 2168 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2169 }; 2170 2171 /* arguments passed to the mmp thread */ 2172 struct mmpd_data { 2173 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2174 struct super_block *sb; /* super block of the fs */ 2175 }; 2176 2177 /* 2178 * Check interval multiplier 2179 * The MMP block is written every update interval and initially checked every 2180 * update interval x the multiplier (the value is then adapted based on the 2181 * write latency). The reason is that writes can be delayed under load and we 2182 * don't want readers to incorrectly assume that the filesystem is no longer 2183 * in use. 2184 */ 2185 #define EXT4_MMP_CHECK_MULT 2UL 2186 2187 /* 2188 * Minimum interval for MMP checking in seconds. 2189 */ 2190 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2191 2192 /* 2193 * Maximum interval for MMP checking in seconds. 2194 */ 2195 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2196 2197 /* 2198 * Function prototypes 2199 */ 2200 2201 /* 2202 * Ok, these declarations are also in <linux/kernel.h> but none of the 2203 * ext4 source programs needs to include it so they are duplicated here. 2204 */ 2205 # define NORET_TYPE /**/ 2206 # define ATTRIB_NORET __attribute__((noreturn)) 2207 # define NORET_AND noreturn, 2208 2209 /* bitmap.c */ 2210 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2211 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2212 struct ext4_group_desc *gdp, 2213 struct buffer_head *bh, int sz); 2214 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2215 struct ext4_group_desc *gdp, 2216 struct buffer_head *bh, int sz); 2217 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2218 struct ext4_group_desc *gdp, 2219 struct buffer_head *bh); 2220 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2221 struct ext4_group_desc *gdp, 2222 struct buffer_head *bh); 2223 2224 /* balloc.c */ 2225 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2226 ext4_fsblk_t blocknr, 2227 ext4_group_t *blockgrpp, 2228 ext4_grpblk_t *offsetp); 2229 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2230 ext4_fsblk_t block); 2231 2232 extern unsigned int ext4_block_group(struct super_block *sb, 2233 ext4_fsblk_t blocknr); 2234 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2235 ext4_fsblk_t blocknr); 2236 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2237 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2238 ext4_group_t group); 2239 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2240 ext4_fsblk_t goal, 2241 unsigned int flags, 2242 unsigned long *count, 2243 int *errp); 2244 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2245 s64 nclusters, unsigned int flags); 2246 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2247 extern void ext4_check_blocks_bitmap(struct super_block *); 2248 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2249 ext4_group_t block_group, 2250 struct buffer_head ** bh); 2251 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2252 2253 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2254 ext4_group_t block_group); 2255 extern int ext4_wait_block_bitmap(struct super_block *sb, 2256 ext4_group_t block_group, 2257 struct buffer_head *bh); 2258 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2259 ext4_group_t block_group); 2260 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2261 ext4_group_t block_group, 2262 struct ext4_group_desc *gdp); 2263 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2264 2265 static inline bool ext4_encrypted_inode(struct inode *inode) 2266 { 2267 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2268 } 2269 2270 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2271 static inline int ext4_fname_setup_filename(struct inode *dir, 2272 const struct qstr *iname, 2273 int lookup, struct ext4_filename *fname) 2274 { 2275 struct fscrypt_name name; 2276 int err; 2277 2278 memset(fname, 0, sizeof(struct ext4_filename)); 2279 2280 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2281 2282 fname->usr_fname = name.usr_fname; 2283 fname->disk_name = name.disk_name; 2284 fname->hinfo.hash = name.hash; 2285 fname->hinfo.minor_hash = name.minor_hash; 2286 fname->crypto_buf = name.crypto_buf; 2287 return err; 2288 } 2289 2290 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2291 { 2292 struct fscrypt_name name; 2293 2294 name.crypto_buf = fname->crypto_buf; 2295 fscrypt_free_filename(&name); 2296 2297 fname->crypto_buf.name = NULL; 2298 fname->usr_fname = NULL; 2299 fname->disk_name.name = NULL; 2300 } 2301 #else 2302 static inline int ext4_fname_setup_filename(struct inode *dir, 2303 const struct qstr *iname, 2304 int lookup, struct ext4_filename *fname) 2305 { 2306 fname->usr_fname = iname; 2307 fname->disk_name.name = (unsigned char *) iname->name; 2308 fname->disk_name.len = iname->len; 2309 return 0; 2310 } 2311 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2312 2313 #endif 2314 2315 /* dir.c */ 2316 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2317 struct file *, 2318 struct ext4_dir_entry_2 *, 2319 struct buffer_head *, char *, int, 2320 unsigned int); 2321 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2322 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2323 (de), (bh), (buf), (size), (offset))) 2324 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2325 __u32 minor_hash, 2326 struct ext4_dir_entry_2 *dirent, 2327 struct fscrypt_str *ent_name); 2328 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2329 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2330 struct buffer_head *bh, 2331 void *buf, int buf_size, 2332 struct ext4_filename *fname, 2333 struct ext4_dir_entry_2 **dest_de); 2334 void ext4_insert_dentry(struct inode *inode, 2335 struct ext4_dir_entry_2 *de, 2336 int buf_size, 2337 struct ext4_filename *fname); 2338 static inline void ext4_update_dx_flag(struct inode *inode) 2339 { 2340 if (!ext4_has_feature_dir_index(inode->i_sb)) 2341 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2342 } 2343 static const unsigned char ext4_filetype_table[] = { 2344 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2345 }; 2346 2347 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2348 { 2349 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2350 return DT_UNKNOWN; 2351 2352 return ext4_filetype_table[filetype]; 2353 } 2354 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2355 void *buf, int buf_size); 2356 2357 /* fsync.c */ 2358 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2359 2360 /* hash.c */ 2361 extern int ext4fs_dirhash(const char *name, int len, struct 2362 dx_hash_info *hinfo); 2363 2364 /* ialloc.c */ 2365 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2366 const struct qstr *qstr, __u32 goal, 2367 uid_t *owner, __u32 i_flags, 2368 int handle_type, unsigned int line_no, 2369 int nblocks); 2370 2371 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2372 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2373 i_flags, 0, 0, 0) 2374 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2375 type, nblocks) \ 2376 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2377 0, (type), __LINE__, (nblocks)) 2378 2379 2380 extern void ext4_free_inode(handle_t *, struct inode *); 2381 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2382 extern unsigned long ext4_count_free_inodes(struct super_block *); 2383 extern unsigned long ext4_count_dirs(struct super_block *); 2384 extern void ext4_check_inodes_bitmap(struct super_block *); 2385 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2386 extern int ext4_init_inode_table(struct super_block *sb, 2387 ext4_group_t group, int barrier); 2388 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2389 2390 /* mballoc.c */ 2391 extern const struct seq_operations ext4_mb_seq_groups_ops; 2392 extern long ext4_mb_stats; 2393 extern long ext4_mb_max_to_scan; 2394 extern int ext4_mb_init(struct super_block *); 2395 extern int ext4_mb_release(struct super_block *); 2396 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2397 struct ext4_allocation_request *, int *); 2398 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2399 extern void ext4_discard_preallocations(struct inode *); 2400 extern int __init ext4_init_mballoc(void); 2401 extern void ext4_exit_mballoc(void); 2402 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2403 struct buffer_head *bh, ext4_fsblk_t block, 2404 unsigned long count, int flags); 2405 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2406 ext4_group_t ngroups); 2407 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2408 ext4_group_t i, struct ext4_group_desc *desc); 2409 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2410 ext4_fsblk_t block, unsigned long count); 2411 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2412 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2413 2414 /* inode.c */ 2415 int ext4_inode_is_fast_symlink(struct inode *inode); 2416 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2417 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2418 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2419 bool wait, struct buffer_head **bhs); 2420 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2421 struct buffer_head *bh_result, int create); 2422 int ext4_get_block(struct inode *inode, sector_t iblock, 2423 struct buffer_head *bh_result, int create); 2424 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2425 struct buffer_head *bh_result, int create); 2426 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2427 struct buffer_head *bh, int create); 2428 int ext4_walk_page_buffers(handle_t *handle, 2429 struct buffer_head *head, 2430 unsigned from, 2431 unsigned to, 2432 int *partial, 2433 int (*fn)(handle_t *handle, 2434 struct buffer_head *bh)); 2435 int do_journal_get_write_access(handle_t *handle, 2436 struct buffer_head *bh); 2437 #define FALL_BACK_TO_NONDELALLOC 1 2438 #define CONVERT_INLINE_DATA 2 2439 2440 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2441 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2442 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2443 extern int ext4_setattr(struct dentry *, struct iattr *); 2444 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2445 extern void ext4_evict_inode(struct inode *); 2446 extern void ext4_clear_inode(struct inode *); 2447 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2448 extern int ext4_sync_inode(handle_t *, struct inode *); 2449 extern void ext4_dirty_inode(struct inode *, int); 2450 extern int ext4_change_inode_journal_flag(struct inode *, int); 2451 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2452 extern int ext4_inode_attach_jinode(struct inode *inode); 2453 extern int ext4_can_truncate(struct inode *inode); 2454 extern int ext4_truncate(struct inode *); 2455 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2456 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2457 extern void ext4_set_inode_flags(struct inode *); 2458 extern int ext4_alloc_da_blocks(struct inode *inode); 2459 extern void ext4_set_aops(struct inode *inode); 2460 extern int ext4_writepage_trans_blocks(struct inode *); 2461 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2462 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2463 loff_t lstart, loff_t lend); 2464 extern int ext4_page_mkwrite(struct vm_fault *vmf); 2465 extern int ext4_filemap_fault(struct vm_fault *vmf); 2466 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2467 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2468 extern void ext4_da_update_reserve_space(struct inode *inode, 2469 int used, int quota_claim); 2470 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2471 ext4_fsblk_t pblk, ext4_lblk_t len); 2472 2473 /* indirect.c */ 2474 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2475 struct ext4_map_blocks *map, int flags); 2476 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2477 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2478 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2479 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2480 ext4_lblk_t start, ext4_lblk_t end); 2481 2482 /* ioctl.c */ 2483 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2484 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2485 2486 /* migrate.c */ 2487 extern int ext4_ext_migrate(struct inode *); 2488 extern int ext4_ind_migrate(struct inode *inode); 2489 2490 /* namei.c */ 2491 extern int ext4_dirent_csum_verify(struct inode *inode, 2492 struct ext4_dir_entry *dirent); 2493 extern int ext4_orphan_add(handle_t *, struct inode *); 2494 extern int ext4_orphan_del(handle_t *, struct inode *); 2495 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2496 __u32 start_minor_hash, __u32 *next_hash); 2497 extern int ext4_search_dir(struct buffer_head *bh, 2498 char *search_buf, 2499 int buf_size, 2500 struct inode *dir, 2501 struct ext4_filename *fname, 2502 unsigned int offset, 2503 struct ext4_dir_entry_2 **res_dir); 2504 extern int ext4_generic_delete_entry(handle_t *handle, 2505 struct inode *dir, 2506 struct ext4_dir_entry_2 *de_del, 2507 struct buffer_head *bh, 2508 void *entry_buf, 2509 int buf_size, 2510 int csum_size); 2511 extern bool ext4_empty_dir(struct inode *inode); 2512 2513 /* resize.c */ 2514 extern int ext4_group_add(struct super_block *sb, 2515 struct ext4_new_group_data *input); 2516 extern int ext4_group_extend(struct super_block *sb, 2517 struct ext4_super_block *es, 2518 ext4_fsblk_t n_blocks_count); 2519 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2520 2521 /* super.c */ 2522 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2523 extern int ext4_calculate_overhead(struct super_block *sb); 2524 extern void ext4_superblock_csum_set(struct super_block *sb); 2525 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2526 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2527 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2528 ext4_group_t ngroup); 2529 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2530 char nbuf[16]); 2531 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2532 ext4_group_t block_group, 2533 unsigned int flags); 2534 2535 extern __printf(4, 5) 2536 void __ext4_error(struct super_block *, const char *, unsigned int, 2537 const char *, ...); 2538 extern __printf(5, 6) 2539 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2540 const char *, ...); 2541 extern __printf(5, 6) 2542 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2543 const char *, ...); 2544 extern void __ext4_std_error(struct super_block *, const char *, 2545 unsigned int, int); 2546 extern __printf(4, 5) 2547 void __ext4_abort(struct super_block *, const char *, unsigned int, 2548 const char *, ...); 2549 extern __printf(4, 5) 2550 void __ext4_warning(struct super_block *, const char *, unsigned int, 2551 const char *, ...); 2552 extern __printf(4, 5) 2553 void __ext4_warning_inode(const struct inode *inode, const char *function, 2554 unsigned int line, const char *fmt, ...); 2555 extern __printf(3, 4) 2556 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2557 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2558 const char *, unsigned int, const char *); 2559 extern __printf(7, 8) 2560 void __ext4_grp_locked_error(const char *, unsigned int, 2561 struct super_block *, ext4_group_t, 2562 unsigned long, ext4_fsblk_t, 2563 const char *, ...); 2564 2565 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2566 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2567 2568 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2569 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2570 2571 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2572 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2573 2574 #ifdef CONFIG_PRINTK 2575 2576 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2577 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2578 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2579 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2580 #define ext4_error(sb, fmt, ...) \ 2581 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2582 #define ext4_abort(sb, fmt, ...) \ 2583 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2584 #define ext4_warning(sb, fmt, ...) \ 2585 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2586 #define ext4_warning_inode(inode, fmt, ...) \ 2587 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2588 #define ext4_msg(sb, level, fmt, ...) \ 2589 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2590 #define dump_mmp_msg(sb, mmp, msg) \ 2591 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2592 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2593 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2594 fmt, ##__VA_ARGS__) 2595 2596 #else 2597 2598 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2599 do { \ 2600 no_printk(fmt, ##__VA_ARGS__); \ 2601 __ext4_error_inode(inode, "", 0, block, " "); \ 2602 } while (0) 2603 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2604 do { \ 2605 no_printk(fmt, ##__VA_ARGS__); \ 2606 __ext4_error_file(file, "", 0, block, " "); \ 2607 } while (0) 2608 #define ext4_error(sb, fmt, ...) \ 2609 do { \ 2610 no_printk(fmt, ##__VA_ARGS__); \ 2611 __ext4_error(sb, "", 0, " "); \ 2612 } while (0) 2613 #define ext4_abort(sb, fmt, ...) \ 2614 do { \ 2615 no_printk(fmt, ##__VA_ARGS__); \ 2616 __ext4_abort(sb, "", 0, " "); \ 2617 } while (0) 2618 #define ext4_warning(sb, fmt, ...) \ 2619 do { \ 2620 no_printk(fmt, ##__VA_ARGS__); \ 2621 __ext4_warning(sb, "", 0, " "); \ 2622 } while (0) 2623 #define ext4_warning_inode(inode, fmt, ...) \ 2624 do { \ 2625 no_printk(fmt, ##__VA_ARGS__); \ 2626 __ext4_warning_inode(inode, "", 0, " "); \ 2627 } while (0) 2628 #define ext4_msg(sb, level, fmt, ...) \ 2629 do { \ 2630 no_printk(fmt, ##__VA_ARGS__); \ 2631 __ext4_msg(sb, "", " "); \ 2632 } while (0) 2633 #define dump_mmp_msg(sb, mmp, msg) \ 2634 __dump_mmp_msg(sb, mmp, "", 0, "") 2635 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2636 do { \ 2637 no_printk(fmt, ##__VA_ARGS__); \ 2638 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2639 } while (0) 2640 2641 #endif 2642 2643 extern void ext4_update_dynamic_rev(struct super_block *sb); 2644 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2645 __u32 compat); 2646 extern int ext4_update_rocompat_feature(handle_t *handle, 2647 struct super_block *sb, __u32 rocompat); 2648 extern int ext4_update_incompat_feature(handle_t *handle, 2649 struct super_block *sb, __u32 incompat); 2650 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2651 struct ext4_group_desc *bg); 2652 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2653 struct ext4_group_desc *bg); 2654 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2655 struct ext4_group_desc *bg); 2656 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2657 struct ext4_group_desc *bg); 2658 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2659 struct ext4_group_desc *bg); 2660 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2661 struct ext4_group_desc *bg); 2662 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2663 struct ext4_group_desc *bg); 2664 extern void ext4_block_bitmap_set(struct super_block *sb, 2665 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2666 extern void ext4_inode_bitmap_set(struct super_block *sb, 2667 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2668 extern void ext4_inode_table_set(struct super_block *sb, 2669 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2670 extern void ext4_free_group_clusters_set(struct super_block *sb, 2671 struct ext4_group_desc *bg, 2672 __u32 count); 2673 extern void ext4_free_inodes_set(struct super_block *sb, 2674 struct ext4_group_desc *bg, __u32 count); 2675 extern void ext4_used_dirs_set(struct super_block *sb, 2676 struct ext4_group_desc *bg, __u32 count); 2677 extern void ext4_itable_unused_set(struct super_block *sb, 2678 struct ext4_group_desc *bg, __u32 count); 2679 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2680 struct ext4_group_desc *gdp); 2681 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2682 struct ext4_group_desc *gdp); 2683 extern int ext4_register_li_request(struct super_block *sb, 2684 ext4_group_t first_not_zeroed); 2685 2686 static inline int ext4_has_metadata_csum(struct super_block *sb) 2687 { 2688 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2689 !EXT4_SB(sb)->s_chksum_driver); 2690 2691 return ext4_has_feature_metadata_csum(sb) && 2692 (EXT4_SB(sb)->s_chksum_driver != NULL); 2693 } 2694 2695 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2696 { 2697 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2698 } 2699 2700 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2701 { 2702 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2703 le32_to_cpu(es->s_blocks_count_lo); 2704 } 2705 2706 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2707 { 2708 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2709 le32_to_cpu(es->s_r_blocks_count_lo); 2710 } 2711 2712 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2713 { 2714 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2715 le32_to_cpu(es->s_free_blocks_count_lo); 2716 } 2717 2718 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2719 ext4_fsblk_t blk) 2720 { 2721 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2722 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2723 } 2724 2725 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2726 ext4_fsblk_t blk) 2727 { 2728 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2729 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2730 } 2731 2732 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2733 ext4_fsblk_t blk) 2734 { 2735 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2736 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2737 } 2738 2739 static inline loff_t ext4_isize(struct super_block *sb, 2740 struct ext4_inode *raw_inode) 2741 { 2742 if (ext4_has_feature_largedir(sb) || 2743 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2744 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2745 le32_to_cpu(raw_inode->i_size_lo); 2746 2747 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2748 } 2749 2750 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2751 { 2752 raw_inode->i_size_lo = cpu_to_le32(i_size); 2753 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2754 } 2755 2756 static inline 2757 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2758 ext4_group_t group) 2759 { 2760 struct ext4_group_info ***grp_info; 2761 long indexv, indexh; 2762 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2763 grp_info = EXT4_SB(sb)->s_group_info; 2764 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2765 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2766 return grp_info[indexv][indexh]; 2767 } 2768 2769 /* 2770 * Reading s_groups_count requires using smp_rmb() afterwards. See 2771 * the locking protocol documented in the comments of ext4_group_add() 2772 * in resize.c 2773 */ 2774 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2775 { 2776 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2777 2778 smp_rmb(); 2779 return ngroups; 2780 } 2781 2782 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2783 ext4_group_t block_group) 2784 { 2785 return block_group >> sbi->s_log_groups_per_flex; 2786 } 2787 2788 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2789 { 2790 return 1 << sbi->s_log_groups_per_flex; 2791 } 2792 2793 #define ext4_std_error(sb, errno) \ 2794 do { \ 2795 if ((errno)) \ 2796 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2797 } while (0) 2798 2799 #ifdef CONFIG_SMP 2800 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2801 * counters. So we need to make sure we have free clusters more 2802 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2803 */ 2804 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2805 #else 2806 #define EXT4_FREECLUSTERS_WATERMARK 0 2807 #endif 2808 2809 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2810 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2811 { 2812 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2813 !inode_is_locked(inode)); 2814 down_write(&EXT4_I(inode)->i_data_sem); 2815 if (newsize > EXT4_I(inode)->i_disksize) 2816 EXT4_I(inode)->i_disksize = newsize; 2817 up_write(&EXT4_I(inode)->i_data_sem); 2818 } 2819 2820 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2821 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2822 { 2823 int changed = 0; 2824 2825 if (newsize > inode->i_size) { 2826 i_size_write(inode, newsize); 2827 changed = 1; 2828 } 2829 if (newsize > EXT4_I(inode)->i_disksize) { 2830 ext4_update_i_disksize(inode, newsize); 2831 changed |= 2; 2832 } 2833 return changed; 2834 } 2835 2836 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2837 loff_t len); 2838 2839 struct ext4_group_info { 2840 unsigned long bb_state; 2841 struct rb_root bb_free_root; 2842 ext4_grpblk_t bb_first_free; /* first free block */ 2843 ext4_grpblk_t bb_free; /* total free blocks */ 2844 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2845 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2846 struct list_head bb_prealloc_list; 2847 #ifdef DOUBLE_CHECK 2848 void *bb_bitmap; 2849 #endif 2850 struct rw_semaphore alloc_sem; 2851 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2852 * regions, index is order. 2853 * bb_counters[3] = 5 means 2854 * 5 free 8-block regions. */ 2855 }; 2856 2857 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2858 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2859 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2860 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2861 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2862 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2863 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2864 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2865 2866 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2867 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2868 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2869 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2870 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2871 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2872 2873 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2874 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2875 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2876 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2877 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2878 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2879 2880 #define EXT4_MAX_CONTENTION 8 2881 #define EXT4_CONTENTION_THRESHOLD 2 2882 2883 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2884 ext4_group_t group) 2885 { 2886 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2887 } 2888 2889 /* 2890 * Returns true if the filesystem is busy enough that attempts to 2891 * access the block group locks has run into contention. 2892 */ 2893 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2894 { 2895 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2896 } 2897 2898 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2899 { 2900 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2901 if (spin_trylock(lock)) 2902 /* 2903 * We're able to grab the lock right away, so drop the 2904 * lock contention counter. 2905 */ 2906 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2907 else { 2908 /* 2909 * The lock is busy, so bump the contention counter, 2910 * and then wait on the spin lock. 2911 */ 2912 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2913 EXT4_MAX_CONTENTION); 2914 spin_lock(lock); 2915 } 2916 } 2917 2918 static inline void ext4_unlock_group(struct super_block *sb, 2919 ext4_group_t group) 2920 { 2921 spin_unlock(ext4_group_lock_ptr(sb, group)); 2922 } 2923 2924 /* 2925 * Block validity checking 2926 */ 2927 #define ext4_check_indirect_blockref(inode, bh) \ 2928 ext4_check_blockref(__func__, __LINE__, inode, \ 2929 (__le32 *)(bh)->b_data, \ 2930 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2931 2932 #define ext4_ind_check_inode(inode) \ 2933 ext4_check_blockref(__func__, __LINE__, inode, \ 2934 EXT4_I(inode)->i_data, \ 2935 EXT4_NDIR_BLOCKS) 2936 2937 /* 2938 * Inodes and files operations 2939 */ 2940 2941 /* dir.c */ 2942 extern const struct file_operations ext4_dir_operations; 2943 2944 /* file.c */ 2945 extern const struct inode_operations ext4_file_inode_operations; 2946 extern const struct file_operations ext4_file_operations; 2947 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2948 2949 /* inline.c */ 2950 extern int ext4_get_max_inline_size(struct inode *inode); 2951 extern int ext4_find_inline_data_nolock(struct inode *inode); 2952 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2953 unsigned int len); 2954 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2955 2956 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2957 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2958 struct inode *inode, 2959 loff_t pos, unsigned len, 2960 unsigned flags, 2961 struct page **pagep); 2962 extern int ext4_write_inline_data_end(struct inode *inode, 2963 loff_t pos, unsigned len, 2964 unsigned copied, 2965 struct page *page); 2966 extern struct buffer_head * 2967 ext4_journalled_write_inline_data(struct inode *inode, 2968 unsigned len, 2969 struct page *page); 2970 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2971 struct inode *inode, 2972 loff_t pos, unsigned len, 2973 unsigned flags, 2974 struct page **pagep, 2975 void **fsdata); 2976 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2977 unsigned len, unsigned copied, 2978 struct page *page); 2979 extern int ext4_try_add_inline_entry(handle_t *handle, 2980 struct ext4_filename *fname, 2981 struct inode *dir, struct inode *inode); 2982 extern int ext4_try_create_inline_dir(handle_t *handle, 2983 struct inode *parent, 2984 struct inode *inode); 2985 extern int ext4_read_inline_dir(struct file *filp, 2986 struct dir_context *ctx, 2987 int *has_inline_data); 2988 extern int htree_inlinedir_to_tree(struct file *dir_file, 2989 struct inode *dir, ext4_lblk_t block, 2990 struct dx_hash_info *hinfo, 2991 __u32 start_hash, __u32 start_minor_hash, 2992 int *has_inline_data); 2993 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2994 struct ext4_filename *fname, 2995 struct ext4_dir_entry_2 **res_dir, 2996 int *has_inline_data); 2997 extern int ext4_delete_inline_entry(handle_t *handle, 2998 struct inode *dir, 2999 struct ext4_dir_entry_2 *de_del, 3000 struct buffer_head *bh, 3001 int *has_inline_data); 3002 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3003 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3004 struct ext4_dir_entry_2 **parent_de, 3005 int *retval); 3006 extern int ext4_inline_data_fiemap(struct inode *inode, 3007 struct fiemap_extent_info *fieinfo, 3008 int *has_inline, __u64 start, __u64 len); 3009 3010 struct iomap; 3011 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3012 3013 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3014 3015 extern int ext4_convert_inline_data(struct inode *inode); 3016 3017 static inline int ext4_has_inline_data(struct inode *inode) 3018 { 3019 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3020 EXT4_I(inode)->i_inline_off; 3021 } 3022 3023 /* namei.c */ 3024 extern const struct inode_operations ext4_dir_inode_operations; 3025 extern const struct inode_operations ext4_special_inode_operations; 3026 extern struct dentry *ext4_get_parent(struct dentry *child); 3027 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3028 struct ext4_dir_entry_2 *de, 3029 int blocksize, int csum_size, 3030 unsigned int parent_ino, int dotdot_real_len); 3031 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3032 unsigned int blocksize); 3033 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3034 struct inode *inode, 3035 struct buffer_head *bh); 3036 #define S_SHIFT 12 3037 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3038 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3039 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3040 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3041 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3042 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3043 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3044 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3045 }; 3046 3047 static inline void ext4_set_de_type(struct super_block *sb, 3048 struct ext4_dir_entry_2 *de, 3049 umode_t mode) { 3050 if (ext4_has_feature_filetype(sb)) 3051 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3052 } 3053 3054 /* readpages.c */ 3055 extern int ext4_mpage_readpages(struct address_space *mapping, 3056 struct list_head *pages, struct page *page, 3057 unsigned nr_pages); 3058 3059 /* symlink.c */ 3060 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3061 extern const struct inode_operations ext4_symlink_inode_operations; 3062 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3063 3064 /* sysfs.c */ 3065 extern int ext4_register_sysfs(struct super_block *sb); 3066 extern void ext4_unregister_sysfs(struct super_block *sb); 3067 extern int __init ext4_init_sysfs(void); 3068 extern void ext4_exit_sysfs(void); 3069 3070 /* block_validity */ 3071 extern void ext4_release_system_zone(struct super_block *sb); 3072 extern int ext4_setup_system_zone(struct super_block *sb); 3073 extern int __init ext4_init_system_zone(void); 3074 extern void ext4_exit_system_zone(void); 3075 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3076 ext4_fsblk_t start_blk, 3077 unsigned int count); 3078 extern int ext4_check_blockref(const char *, unsigned int, 3079 struct inode *, __le32 *, unsigned int); 3080 3081 /* extents.c */ 3082 struct ext4_ext_path; 3083 struct ext4_extent; 3084 3085 /* 3086 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3087 * __le32. 3088 */ 3089 #define EXT_MAX_BLOCKS 0xffffffff 3090 3091 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3092 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3093 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3094 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3095 struct ext4_map_blocks *map, int flags); 3096 extern int ext4_ext_truncate(handle_t *, struct inode *); 3097 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3098 ext4_lblk_t end); 3099 extern void ext4_ext_init(struct super_block *); 3100 extern void ext4_ext_release(struct super_block *); 3101 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3102 loff_t len); 3103 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3104 loff_t offset, ssize_t len); 3105 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3106 struct ext4_map_blocks *map, int flags); 3107 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3108 ext4_lblk_t lblocks); 3109 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3110 int num, 3111 struct ext4_ext_path *path); 3112 extern int ext4_can_extents_be_merged(struct inode *inode, 3113 struct ext4_extent *ex1, 3114 struct ext4_extent *ex2); 3115 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3116 struct ext4_ext_path **, 3117 struct ext4_extent *, int); 3118 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3119 struct ext4_ext_path **, 3120 int flags); 3121 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3122 extern int ext4_ext_check_inode(struct inode *inode); 3123 extern int ext4_find_delalloc_range(struct inode *inode, 3124 ext4_lblk_t lblk_start, 3125 ext4_lblk_t lblk_end); 3126 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3127 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3128 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3129 __u64 start, __u64 len); 3130 extern int ext4_ext_precache(struct inode *inode); 3131 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3132 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3133 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3134 struct inode *inode2, ext4_lblk_t lblk1, 3135 ext4_lblk_t lblk2, ext4_lblk_t count, 3136 int mark_unwritten,int *err); 3137 3138 /* move_extent.c */ 3139 extern void ext4_double_down_write_data_sem(struct inode *first, 3140 struct inode *second); 3141 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3142 struct inode *donor_inode); 3143 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3144 __u64 start_orig, __u64 start_donor, 3145 __u64 len, __u64 *moved_len); 3146 3147 /* page-io.c */ 3148 extern int __init ext4_init_pageio(void); 3149 extern void ext4_exit_pageio(void); 3150 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3151 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3152 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3153 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3154 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3155 struct writeback_control *wbc); 3156 extern void ext4_end_io_rsv_work(struct work_struct *work); 3157 extern void ext4_io_submit(struct ext4_io_submit *io); 3158 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3159 struct page *page, 3160 int len, 3161 struct writeback_control *wbc, 3162 bool keep_towrite); 3163 3164 /* mmp.c */ 3165 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3166 3167 /* 3168 * Add new method to test whether block and inode bitmaps are properly 3169 * initialized. With uninit_bg reading the block from disk is not enough 3170 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3171 */ 3172 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3173 3174 static inline int bitmap_uptodate(struct buffer_head *bh) 3175 { 3176 return (buffer_uptodate(bh) && 3177 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3178 } 3179 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3180 { 3181 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3182 } 3183 3184 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3185 3186 /* For ioend & aio unwritten conversion wait queues */ 3187 #define EXT4_WQ_HASH_SZ 37 3188 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3189 EXT4_WQ_HASH_SZ]) 3190 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3191 3192 extern int ext4_resize_begin(struct super_block *sb); 3193 extern void ext4_resize_end(struct super_block *sb); 3194 3195 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3196 struct ext4_io_end *io_end) 3197 { 3198 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3199 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3200 atomic_inc(&EXT4_I(inode)->i_unwritten); 3201 } 3202 } 3203 3204 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3205 { 3206 struct inode *inode = io_end->inode; 3207 3208 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3209 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3210 /* Wake up anyone waiting on unwritten extent conversion */ 3211 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3212 wake_up_all(ext4_ioend_wq(inode)); 3213 } 3214 } 3215 3216 extern const struct iomap_ops ext4_iomap_ops; 3217 3218 #endif /* __KERNEL__ */ 3219 3220 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3221 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3222 3223 #endif /* _EXT4_H */ 3224