1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION) 44 #include <linux/fscrypt.h> 45 46 #include <linux/compiler.h> 47 48 /* Until this gets included into linux/compiler-gcc.h */ 49 #ifndef __nonstring 50 #if defined(GCC_VERSION) && (GCC_VERSION >= 80000) 51 #define __nonstring __attribute__((nonstring)) 52 #else 53 #define __nonstring 54 #endif 55 #endif 56 57 /* 58 * The fourth extended filesystem constants/structures 59 */ 60 61 /* 62 * with AGGRESSIVE_CHECK allocator runs consistency checks over 63 * structures. these checks slow things down a lot 64 */ 65 #define AGGRESSIVE_CHECK__ 66 67 /* 68 * with DOUBLE_CHECK defined mballoc creates persistent in-core 69 * bitmaps, maintains and uses them to check for double allocations 70 */ 71 #define DOUBLE_CHECK__ 72 73 /* 74 * Define EXT4FS_DEBUG to produce debug messages 75 */ 76 #undef EXT4FS_DEBUG 77 78 /* 79 * Debug code 80 */ 81 #ifdef EXT4FS_DEBUG 82 #define ext4_debug(f, a...) \ 83 do { \ 84 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 85 __FILE__, __LINE__, __func__); \ 86 printk(KERN_DEBUG f, ## a); \ 87 } while (0) 88 #else 89 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 90 #endif 91 92 /* 93 * Turn on EXT_DEBUG to get lots of info about extents operations. 94 */ 95 #define EXT_DEBUG__ 96 #ifdef EXT_DEBUG 97 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 98 #else 99 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 100 #endif 101 102 /* data type for block offset of block group */ 103 typedef int ext4_grpblk_t; 104 105 /* data type for filesystem-wide blocks number */ 106 typedef unsigned long long ext4_fsblk_t; 107 108 /* data type for file logical block number */ 109 typedef __u32 ext4_lblk_t; 110 111 /* data type for block group number */ 112 typedef unsigned int ext4_group_t; 113 114 enum SHIFT_DIRECTION { 115 SHIFT_LEFT = 0, 116 SHIFT_RIGHT, 117 }; 118 119 /* 120 * Flags used in mballoc's allocation_context flags field. 121 * 122 * Also used to show what's going on for debugging purposes when the 123 * flag field is exported via the traceport interface 124 */ 125 126 /* prefer goal again. length */ 127 #define EXT4_MB_HINT_MERGE 0x0001 128 /* blocks already reserved */ 129 #define EXT4_MB_HINT_RESERVED 0x0002 130 /* metadata is being allocated */ 131 #define EXT4_MB_HINT_METADATA 0x0004 132 /* first blocks in the file */ 133 #define EXT4_MB_HINT_FIRST 0x0008 134 /* search for the best chunk */ 135 #define EXT4_MB_HINT_BEST 0x0010 136 /* data is being allocated */ 137 #define EXT4_MB_HINT_DATA 0x0020 138 /* don't preallocate (for tails) */ 139 #define EXT4_MB_HINT_NOPREALLOC 0x0040 140 /* allocate for locality group */ 141 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 142 /* allocate goal blocks or none */ 143 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 144 /* goal is meaningful */ 145 #define EXT4_MB_HINT_TRY_GOAL 0x0200 146 /* blocks already pre-reserved by delayed allocation */ 147 #define EXT4_MB_DELALLOC_RESERVED 0x0400 148 /* We are doing stream allocation */ 149 #define EXT4_MB_STREAM_ALLOC 0x0800 150 /* Use reserved root blocks if needed */ 151 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 152 /* Use blocks from reserved pool */ 153 #define EXT4_MB_USE_RESERVED 0x2000 154 155 struct ext4_allocation_request { 156 /* target inode for block we're allocating */ 157 struct inode *inode; 158 /* how many blocks we want to allocate */ 159 unsigned int len; 160 /* logical block in target inode */ 161 ext4_lblk_t logical; 162 /* the closest logical allocated block to the left */ 163 ext4_lblk_t lleft; 164 /* the closest logical allocated block to the right */ 165 ext4_lblk_t lright; 166 /* phys. target (a hint) */ 167 ext4_fsblk_t goal; 168 /* phys. block for the closest logical allocated block to the left */ 169 ext4_fsblk_t pleft; 170 /* phys. block for the closest logical allocated block to the right */ 171 ext4_fsblk_t pright; 172 /* flags. see above EXT4_MB_HINT_* */ 173 unsigned int flags; 174 }; 175 176 /* 177 * Logical to physical block mapping, used by ext4_map_blocks() 178 * 179 * This structure is used to pass requests into ext4_map_blocks() as 180 * well as to store the information returned by ext4_map_blocks(). It 181 * takes less room on the stack than a struct buffer_head. 182 */ 183 #define EXT4_MAP_NEW (1 << BH_New) 184 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 185 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 186 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 187 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 188 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 189 190 struct ext4_map_blocks { 191 ext4_fsblk_t m_pblk; 192 ext4_lblk_t m_lblk; 193 unsigned int m_len; 194 unsigned int m_flags; 195 }; 196 197 /* 198 * Flags for ext4_io_end->flags 199 */ 200 #define EXT4_IO_END_UNWRITTEN 0x0001 201 202 /* 203 * For converting unwritten extents on a work queue. 'handle' is used for 204 * buffered writeback. 205 */ 206 typedef struct ext4_io_end { 207 struct list_head list; /* per-file finished IO list */ 208 handle_t *handle; /* handle reserved for extent 209 * conversion */ 210 struct inode *inode; /* file being written to */ 211 struct bio *bio; /* Linked list of completed 212 * bios covering the extent */ 213 unsigned int flag; /* unwritten or not */ 214 atomic_t count; /* reference counter */ 215 loff_t offset; /* offset in the file */ 216 ssize_t size; /* size of the extent */ 217 } ext4_io_end_t; 218 219 struct ext4_io_submit { 220 struct writeback_control *io_wbc; 221 struct bio *io_bio; 222 ext4_io_end_t *io_end; 223 sector_t io_next_block; 224 }; 225 226 /* 227 * Special inodes numbers 228 */ 229 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 230 #define EXT4_ROOT_INO 2 /* Root inode */ 231 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 232 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 233 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 234 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 235 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 236 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 237 238 /* First non-reserved inode for old ext4 filesystems */ 239 #define EXT4_GOOD_OLD_FIRST_INO 11 240 241 /* 242 * Maximal count of links to a file 243 */ 244 #define EXT4_LINK_MAX 65000 245 246 /* 247 * Macro-instructions used to manage several block sizes 248 */ 249 #define EXT4_MIN_BLOCK_SIZE 1024 250 #define EXT4_MAX_BLOCK_SIZE 65536 251 #define EXT4_MIN_BLOCK_LOG_SIZE 10 252 #define EXT4_MAX_BLOCK_LOG_SIZE 16 253 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 254 #ifdef __KERNEL__ 255 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 256 #else 257 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 258 #endif 259 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 260 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 261 EXT4_SB(s)->s_cluster_bits) 262 #ifdef __KERNEL__ 263 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 264 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 265 #else 266 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 267 #endif 268 #ifdef __KERNEL__ 269 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 270 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 271 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 272 #else 273 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 274 EXT4_GOOD_OLD_INODE_SIZE : \ 275 (s)->s_inode_size) 276 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 277 EXT4_GOOD_OLD_FIRST_INO : \ 278 (s)->s_first_ino) 279 #endif 280 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 281 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 282 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 283 blkbits)) 284 285 /* Translate a block number to a cluster number */ 286 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 287 /* Translate a cluster number to a block number */ 288 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 289 /* Translate # of blks to # of clusters */ 290 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 291 (sbi)->s_cluster_bits) 292 /* Mask out the low bits to get the starting block of the cluster */ 293 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 294 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 295 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 296 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 297 /* Get the cluster offset */ 298 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 299 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 300 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 301 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 302 303 /* 304 * Structure of a blocks group descriptor 305 */ 306 struct ext4_group_desc 307 { 308 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 309 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 310 __le32 bg_inode_table_lo; /* Inodes table block */ 311 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 312 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 313 __le16 bg_used_dirs_count_lo; /* Directories count */ 314 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 315 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 316 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 317 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 318 __le16 bg_itable_unused_lo; /* Unused inodes count */ 319 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 320 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 321 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 322 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 323 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 324 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 325 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 326 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 327 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 328 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 329 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 330 __u32 bg_reserved; 331 }; 332 333 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 334 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 335 sizeof(__le16)) 336 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 337 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 338 sizeof(__le16)) 339 340 /* 341 * Structure of a flex block group info 342 */ 343 344 struct flex_groups { 345 atomic64_t free_clusters; 346 atomic_t free_inodes; 347 atomic_t used_dirs; 348 }; 349 350 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 351 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 352 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 353 354 /* 355 * Macro-instructions used to manage group descriptors 356 */ 357 #define EXT4_MIN_DESC_SIZE 32 358 #define EXT4_MIN_DESC_SIZE_64BIT 64 359 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 360 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 361 #ifdef __KERNEL__ 362 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 363 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 364 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 365 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 366 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 367 #else 368 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 369 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 370 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 371 #endif 372 373 /* 374 * Constants relative to the data blocks 375 */ 376 #define EXT4_NDIR_BLOCKS 12 377 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 378 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 379 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 380 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 381 382 /* 383 * Inode flags 384 */ 385 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 386 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 387 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 388 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 389 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 390 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 391 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 392 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 393 /* Reserved for compression usage... */ 394 #define EXT4_DIRTY_FL 0x00000100 395 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 396 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 397 /* nb: was previously EXT2_ECOMPR_FL */ 398 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 399 /* End compression flags --- maybe not all used */ 400 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 401 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 402 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 403 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 404 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 405 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 406 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 407 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 408 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 409 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 410 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 411 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 412 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 413 414 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 415 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 416 417 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 418 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 419 EXT4_IMMUTABLE_FL | \ 420 EXT4_APPEND_FL | \ 421 EXT4_NODUMP_FL | \ 422 EXT4_NOATIME_FL | \ 423 EXT4_PROJINHERIT_FL) 424 425 /* Flags that should be inherited by new inodes from their parent. */ 426 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 427 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 428 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 429 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 430 EXT4_PROJINHERIT_FL) 431 432 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 433 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 434 435 /* Flags that are appropriate for non-directories/regular files. */ 436 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 437 438 /* Mask out flags that are inappropriate for the given type of inode. */ 439 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 440 { 441 if (S_ISDIR(mode)) 442 return flags; 443 else if (S_ISREG(mode)) 444 return flags & EXT4_REG_FLMASK; 445 else 446 return flags & EXT4_OTHER_FLMASK; 447 } 448 449 /* 450 * Inode flags used for atomic set/get 451 */ 452 enum { 453 EXT4_INODE_SECRM = 0, /* Secure deletion */ 454 EXT4_INODE_UNRM = 1, /* Undelete */ 455 EXT4_INODE_COMPR = 2, /* Compress file */ 456 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 457 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 458 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 459 EXT4_INODE_NODUMP = 6, /* do not dump file */ 460 EXT4_INODE_NOATIME = 7, /* do not update atime */ 461 /* Reserved for compression usage... */ 462 EXT4_INODE_DIRTY = 8, 463 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 464 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 465 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 466 /* End compression flags --- maybe not all used */ 467 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 468 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 469 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 470 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 471 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 472 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 473 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 474 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 475 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 476 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 477 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 478 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 479 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 480 }; 481 482 /* 483 * Since it's pretty easy to mix up bit numbers and hex values, we use a 484 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 485 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 486 * any extra space in the compiled kernel image, otherwise, the build will fail. 487 * It's important that these values are the same, since we are using 488 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 489 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 490 * values found in ext2, ext3 and ext4 filesystems, and of course the values 491 * defined in e2fsprogs. 492 * 493 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 494 */ 495 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 496 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 497 498 static inline void ext4_check_flag_values(void) 499 { 500 CHECK_FLAG_VALUE(SECRM); 501 CHECK_FLAG_VALUE(UNRM); 502 CHECK_FLAG_VALUE(COMPR); 503 CHECK_FLAG_VALUE(SYNC); 504 CHECK_FLAG_VALUE(IMMUTABLE); 505 CHECK_FLAG_VALUE(APPEND); 506 CHECK_FLAG_VALUE(NODUMP); 507 CHECK_FLAG_VALUE(NOATIME); 508 CHECK_FLAG_VALUE(DIRTY); 509 CHECK_FLAG_VALUE(COMPRBLK); 510 CHECK_FLAG_VALUE(NOCOMPR); 511 CHECK_FLAG_VALUE(ENCRYPT); 512 CHECK_FLAG_VALUE(INDEX); 513 CHECK_FLAG_VALUE(IMAGIC); 514 CHECK_FLAG_VALUE(JOURNAL_DATA); 515 CHECK_FLAG_VALUE(NOTAIL); 516 CHECK_FLAG_VALUE(DIRSYNC); 517 CHECK_FLAG_VALUE(TOPDIR); 518 CHECK_FLAG_VALUE(HUGE_FILE); 519 CHECK_FLAG_VALUE(EXTENTS); 520 CHECK_FLAG_VALUE(EA_INODE); 521 CHECK_FLAG_VALUE(EOFBLOCKS); 522 CHECK_FLAG_VALUE(INLINE_DATA); 523 CHECK_FLAG_VALUE(PROJINHERIT); 524 CHECK_FLAG_VALUE(RESERVED); 525 } 526 527 /* Used to pass group descriptor data when online resize is done */ 528 struct ext4_new_group_input { 529 __u32 group; /* Group number for this data */ 530 __u64 block_bitmap; /* Absolute block number of block bitmap */ 531 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 532 __u64 inode_table; /* Absolute block number of inode table start */ 533 __u32 blocks_count; /* Total number of blocks in this group */ 534 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 535 __u16 unused; 536 }; 537 538 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 539 struct compat_ext4_new_group_input { 540 u32 group; 541 compat_u64 block_bitmap; 542 compat_u64 inode_bitmap; 543 compat_u64 inode_table; 544 u32 blocks_count; 545 u16 reserved_blocks; 546 u16 unused; 547 }; 548 #endif 549 550 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 551 struct ext4_new_group_data { 552 __u32 group; 553 __u64 block_bitmap; 554 __u64 inode_bitmap; 555 __u64 inode_table; 556 __u32 blocks_count; 557 __u16 reserved_blocks; 558 __u16 mdata_blocks; 559 __u32 free_clusters_count; 560 }; 561 562 /* Indexes used to index group tables in ext4_new_group_data */ 563 enum { 564 BLOCK_BITMAP = 0, /* block bitmap */ 565 INODE_BITMAP, /* inode bitmap */ 566 INODE_TABLE, /* inode tables */ 567 GROUP_TABLE_COUNT, 568 }; 569 570 /* 571 * Flags used by ext4_map_blocks() 572 */ 573 /* Allocate any needed blocks and/or convert an unwritten 574 extent to be an initialized ext4 */ 575 #define EXT4_GET_BLOCKS_CREATE 0x0001 576 /* Request the creation of an unwritten extent */ 577 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 578 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 579 EXT4_GET_BLOCKS_CREATE) 580 /* Caller is from the delayed allocation writeout path 581 * finally doing the actual allocation of delayed blocks */ 582 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 583 /* caller is from the direct IO path, request to creation of an 584 unwritten extents if not allocated, split the unwritten 585 extent if blocks has been preallocated already*/ 586 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 587 #define EXT4_GET_BLOCKS_CONVERT 0x0010 588 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 589 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 590 /* Convert extent to initialized after IO complete */ 591 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 592 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 593 /* Eventual metadata allocation (due to growing extent tree) 594 * should not fail, so try to use reserved blocks for that.*/ 595 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 596 /* Don't normalize allocation size (used for fallocate) */ 597 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 598 /* Request will not result in inode size update (user for fallocate) */ 599 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 600 /* Convert written extents to unwritten */ 601 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 602 /* Write zeros to newly created written extents */ 603 #define EXT4_GET_BLOCKS_ZERO 0x0200 604 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 605 EXT4_GET_BLOCKS_ZERO) 606 /* Caller will submit data before dropping transaction handle. This 607 * allows jbd2 to avoid submitting data before commit. */ 608 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 609 610 /* 611 * The bit position of these flags must not overlap with any of the 612 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 613 * read_extent_tree_block(), ext4_split_extent_at(), 614 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 615 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 616 * caching the extents when reading from the extent tree while a 617 * truncate or punch hole operation is in progress. 618 */ 619 #define EXT4_EX_NOCACHE 0x40000000 620 #define EXT4_EX_FORCE_CACHE 0x20000000 621 622 /* 623 * Flags used by ext4_free_blocks 624 */ 625 #define EXT4_FREE_BLOCKS_METADATA 0x0001 626 #define EXT4_FREE_BLOCKS_FORGET 0x0002 627 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 628 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 629 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 630 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 631 632 /* 633 * ioctl commands 634 */ 635 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 636 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 637 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 638 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 639 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 640 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 641 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 642 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 643 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 644 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 645 #define EXT4_IOC_MIGRATE _IO('f', 9) 646 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 647 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 648 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 649 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 650 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 651 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 652 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 653 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 654 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 655 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 656 657 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 658 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 659 660 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 661 662 /* 663 * Flags for going down operation 664 */ 665 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 666 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 667 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 668 669 670 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 671 /* 672 * ioctl commands in 32 bit emulation 673 */ 674 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 675 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 676 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 677 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 678 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 679 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 680 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 681 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 682 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 683 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 684 #endif 685 686 /* Max physical block we can address w/o extents */ 687 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 688 689 /* 690 * Structure of an inode on the disk 691 */ 692 struct ext4_inode { 693 __le16 i_mode; /* File mode */ 694 __le16 i_uid; /* Low 16 bits of Owner Uid */ 695 __le32 i_size_lo; /* Size in bytes */ 696 __le32 i_atime; /* Access time */ 697 __le32 i_ctime; /* Inode Change time */ 698 __le32 i_mtime; /* Modification time */ 699 __le32 i_dtime; /* Deletion Time */ 700 __le16 i_gid; /* Low 16 bits of Group Id */ 701 __le16 i_links_count; /* Links count */ 702 __le32 i_blocks_lo; /* Blocks count */ 703 __le32 i_flags; /* File flags */ 704 union { 705 struct { 706 __le32 l_i_version; 707 } linux1; 708 struct { 709 __u32 h_i_translator; 710 } hurd1; 711 struct { 712 __u32 m_i_reserved1; 713 } masix1; 714 } osd1; /* OS dependent 1 */ 715 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 716 __le32 i_generation; /* File version (for NFS) */ 717 __le32 i_file_acl_lo; /* File ACL */ 718 __le32 i_size_high; 719 __le32 i_obso_faddr; /* Obsoleted fragment address */ 720 union { 721 struct { 722 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 723 __le16 l_i_file_acl_high; 724 __le16 l_i_uid_high; /* these 2 fields */ 725 __le16 l_i_gid_high; /* were reserved2[0] */ 726 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 727 __le16 l_i_reserved; 728 } linux2; 729 struct { 730 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 731 __u16 h_i_mode_high; 732 __u16 h_i_uid_high; 733 __u16 h_i_gid_high; 734 __u32 h_i_author; 735 } hurd2; 736 struct { 737 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 738 __le16 m_i_file_acl_high; 739 __u32 m_i_reserved2[2]; 740 } masix2; 741 } osd2; /* OS dependent 2 */ 742 __le16 i_extra_isize; 743 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 744 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 745 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 746 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 747 __le32 i_crtime; /* File Creation time */ 748 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 749 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 750 __le32 i_projid; /* Project ID */ 751 }; 752 753 struct move_extent { 754 __u32 reserved; /* should be zero */ 755 __u32 donor_fd; /* donor file descriptor */ 756 __u64 orig_start; /* logical start offset in block for orig */ 757 __u64 donor_start; /* logical start offset in block for donor */ 758 __u64 len; /* block length to be moved */ 759 __u64 moved_len; /* moved block length */ 760 }; 761 762 #define EXT4_EPOCH_BITS 2 763 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 764 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 765 766 /* 767 * Extended fields will fit into an inode if the filesystem was formatted 768 * with large inodes (-I 256 or larger) and there are not currently any EAs 769 * consuming all of the available space. For new inodes we always reserve 770 * enough space for the kernel's known extended fields, but for inodes 771 * created with an old kernel this might not have been the case. None of 772 * the extended inode fields is critical for correct filesystem operation. 773 * This macro checks if a certain field fits in the inode. Note that 774 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 775 */ 776 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 777 ((offsetof(typeof(*ext4_inode), field) + \ 778 sizeof((ext4_inode)->field)) \ 779 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 780 (einode)->i_extra_isize)) \ 781 782 /* 783 * We use an encoding that preserves the times for extra epoch "00": 784 * 785 * extra msb of adjust for signed 786 * epoch 32-bit 32-bit tv_sec to 787 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 788 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 789 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 790 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 791 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 792 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 793 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 794 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 795 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 796 * 797 * Note that previous versions of the kernel on 64-bit systems would 798 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 799 * 1970. e2fsck will correct this, assuming that it is run on the 800 * affected filesystem before 2242. 801 */ 802 803 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 804 { 805 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 806 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 807 } 808 809 static inline void ext4_decode_extra_time(struct timespec64 *time, 810 __le32 extra) 811 { 812 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 813 814 #if 1 815 /* Handle legacy encoding of pre-1970 dates with epoch 816 * bits 1,1. (This backwards compatibility may be removed 817 * at the discretion of the ext4 developers.) 818 */ 819 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 820 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 821 extra_bits = 0; 822 time->tv_sec += extra_bits << 32; 823 #else 824 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 825 #endif 826 } 827 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 828 } 829 830 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 831 do { \ 832 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 833 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 834 (raw_inode)->xtime ## _extra = \ 835 ext4_encode_extra_time(&(inode)->xtime); \ 836 } \ 837 } while (0) 838 839 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 840 do { \ 841 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 842 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 843 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 844 (raw_inode)->xtime ## _extra = \ 845 ext4_encode_extra_time(&(einode)->xtime); \ 846 } while (0) 847 848 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 849 do { \ 850 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 851 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 852 ext4_decode_extra_time(&(inode)->xtime, \ 853 raw_inode->xtime ## _extra); \ 854 } \ 855 else \ 856 (inode)->xtime.tv_nsec = 0; \ 857 } while (0) 858 859 860 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 861 do { \ 862 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 863 (einode)->xtime.tv_sec = \ 864 (signed)le32_to_cpu((raw_inode)->xtime); \ 865 else \ 866 (einode)->xtime.tv_sec = 0; \ 867 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 868 ext4_decode_extra_time(&(einode)->xtime, \ 869 raw_inode->xtime ## _extra); \ 870 else \ 871 (einode)->xtime.tv_nsec = 0; \ 872 } while (0) 873 874 #define i_disk_version osd1.linux1.l_i_version 875 876 #if defined(__KERNEL__) || defined(__linux__) 877 #define i_reserved1 osd1.linux1.l_i_reserved1 878 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 879 #define i_blocks_high osd2.linux2.l_i_blocks_high 880 #define i_uid_low i_uid 881 #define i_gid_low i_gid 882 #define i_uid_high osd2.linux2.l_i_uid_high 883 #define i_gid_high osd2.linux2.l_i_gid_high 884 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 885 886 #elif defined(__GNU__) 887 888 #define i_translator osd1.hurd1.h_i_translator 889 #define i_uid_high osd2.hurd2.h_i_uid_high 890 #define i_gid_high osd2.hurd2.h_i_gid_high 891 #define i_author osd2.hurd2.h_i_author 892 893 #elif defined(__masix__) 894 895 #define i_reserved1 osd1.masix1.m_i_reserved1 896 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 897 #define i_reserved2 osd2.masix2.m_i_reserved2 898 899 #endif /* defined(__KERNEL__) || defined(__linux__) */ 900 901 #include "extents_status.h" 902 903 /* 904 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 905 * 906 * These are needed to avoid lockdep false positives when we need to 907 * allocate blocks to the quota inode during ext4_map_blocks(), while 908 * holding i_data_sem for a normal (non-quota) inode. Since we don't 909 * do quota tracking for the quota inode, this avoids deadlock (as 910 * well as infinite recursion, since it isn't turtles all the way 911 * down...) 912 * 913 * I_DATA_SEM_NORMAL - Used for most inodes 914 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 915 * where the second inode has larger inode number 916 * than the first 917 * I_DATA_SEM_QUOTA - Used for quota inodes only 918 */ 919 enum { 920 I_DATA_SEM_NORMAL = 0, 921 I_DATA_SEM_OTHER, 922 I_DATA_SEM_QUOTA, 923 }; 924 925 926 /* 927 * fourth extended file system inode data in memory 928 */ 929 struct ext4_inode_info { 930 __le32 i_data[15]; /* unconverted */ 931 __u32 i_dtime; 932 ext4_fsblk_t i_file_acl; 933 934 /* 935 * i_block_group is the number of the block group which contains 936 * this file's inode. Constant across the lifetime of the inode, 937 * it is used for making block allocation decisions - we try to 938 * place a file's data blocks near its inode block, and new inodes 939 * near to their parent directory's inode. 940 */ 941 ext4_group_t i_block_group; 942 ext4_lblk_t i_dir_start_lookup; 943 #if (BITS_PER_LONG < 64) 944 unsigned long i_state_flags; /* Dynamic state flags */ 945 #endif 946 unsigned long i_flags; 947 948 /* 949 * Extended attributes can be read independently of the main file 950 * data. Taking i_mutex even when reading would cause contention 951 * between readers of EAs and writers of regular file data, so 952 * instead we synchronize on xattr_sem when reading or changing 953 * EAs. 954 */ 955 struct rw_semaphore xattr_sem; 956 957 struct list_head i_orphan; /* unlinked but open inodes */ 958 959 /* 960 * i_disksize keeps track of what the inode size is ON DISK, not 961 * in memory. During truncate, i_size is set to the new size by 962 * the VFS prior to calling ext4_truncate(), but the filesystem won't 963 * set i_disksize to 0 until the truncate is actually under way. 964 * 965 * The intent is that i_disksize always represents the blocks which 966 * are used by this file. This allows recovery to restart truncate 967 * on orphans if we crash during truncate. We actually write i_disksize 968 * into the on-disk inode when writing inodes out, instead of i_size. 969 * 970 * The only time when i_disksize and i_size may be different is when 971 * a truncate is in progress. The only things which change i_disksize 972 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 973 */ 974 loff_t i_disksize; 975 976 /* 977 * i_data_sem is for serialising ext4_truncate() against 978 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 979 * data tree are chopped off during truncate. We can't do that in 980 * ext4 because whenever we perform intermediate commits during 981 * truncate, the inode and all the metadata blocks *must* be in a 982 * consistent state which allows truncation of the orphans to restart 983 * during recovery. Hence we must fix the get_block-vs-truncate race 984 * by other means, so we have i_data_sem. 985 */ 986 struct rw_semaphore i_data_sem; 987 /* 988 * i_mmap_sem is for serializing page faults with truncate / punch hole 989 * operations. We have to make sure that new page cannot be faulted in 990 * a section of the inode that is being punched. We cannot easily use 991 * i_data_sem for this since we need protection for the whole punch 992 * operation and i_data_sem ranks below transaction start so we have 993 * to occasionally drop it. 994 */ 995 struct rw_semaphore i_mmap_sem; 996 struct inode vfs_inode; 997 struct jbd2_inode *jinode; 998 999 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1000 1001 /* 1002 * File creation time. Its function is same as that of 1003 * struct timespec64 i_{a,c,m}time in the generic inode. 1004 */ 1005 struct timespec64 i_crtime; 1006 1007 /* mballoc */ 1008 struct list_head i_prealloc_list; 1009 spinlock_t i_prealloc_lock; 1010 1011 /* extents status tree */ 1012 struct ext4_es_tree i_es_tree; 1013 rwlock_t i_es_lock; 1014 struct list_head i_es_list; 1015 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1016 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1017 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1018 extents to shrink. Protected by 1019 i_es_lock */ 1020 1021 /* ialloc */ 1022 ext4_group_t i_last_alloc_group; 1023 1024 /* allocation reservation info for delalloc */ 1025 /* In case of bigalloc, this refer to clusters rather than blocks */ 1026 unsigned int i_reserved_data_blocks; 1027 ext4_lblk_t i_da_metadata_calc_last_lblock; 1028 int i_da_metadata_calc_len; 1029 1030 /* on-disk additional length */ 1031 __u16 i_extra_isize; 1032 1033 /* Indicate the inline data space. */ 1034 u16 i_inline_off; 1035 u16 i_inline_size; 1036 1037 #ifdef CONFIG_QUOTA 1038 /* quota space reservation, managed internally by quota code */ 1039 qsize_t i_reserved_quota; 1040 #endif 1041 1042 /* Lock protecting lists below */ 1043 spinlock_t i_completed_io_lock; 1044 /* 1045 * Completed IOs that need unwritten extents handling and have 1046 * transaction reserved 1047 */ 1048 struct list_head i_rsv_conversion_list; 1049 struct work_struct i_rsv_conversion_work; 1050 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1051 1052 spinlock_t i_block_reservation_lock; 1053 1054 /* 1055 * Transactions that contain inode's metadata needed to complete 1056 * fsync and fdatasync, respectively. 1057 */ 1058 tid_t i_sync_tid; 1059 tid_t i_datasync_tid; 1060 1061 #ifdef CONFIG_QUOTA 1062 struct dquot *i_dquot[MAXQUOTAS]; 1063 #endif 1064 1065 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1066 __u32 i_csum_seed; 1067 1068 kprojid_t i_projid; 1069 }; 1070 1071 /* 1072 * File system states 1073 */ 1074 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1075 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1076 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1077 1078 /* 1079 * Misc. filesystem flags 1080 */ 1081 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1082 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1083 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1084 1085 /* 1086 * Mount flags set via mount options or defaults 1087 */ 1088 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1089 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1090 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1091 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1092 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1093 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1094 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1095 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1096 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1097 #ifdef CONFIG_FS_DAX 1098 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1099 #else 1100 #define EXT4_MOUNT_DAX 0 1101 #endif 1102 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1103 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1104 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1105 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1106 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1107 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1108 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1109 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1110 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1111 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1112 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1113 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1114 * enable enforcement for hidden 1115 * quota files */ 1116 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1117 * enforcement for hidden quota 1118 * files */ 1119 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1120 * enforcement */ 1121 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1122 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1123 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1124 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1125 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1126 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1127 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1128 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1129 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1130 1131 /* 1132 * Mount flags set either automatically (could not be set by mount option) 1133 * based on per file system feature or property or in special cases such as 1134 * distinguishing between explicit mount option definition and default. 1135 */ 1136 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1137 specified delalloc */ 1138 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1139 size of blocksize * 8 1140 blocks */ 1141 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1142 file systems */ 1143 1144 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1145 specified journal checksum */ 1146 1147 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1148 ~EXT4_MOUNT_##opt 1149 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1150 EXT4_MOUNT_##opt 1151 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1152 EXT4_MOUNT_##opt) 1153 1154 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1155 ~EXT4_MOUNT2_##opt 1156 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1157 EXT4_MOUNT2_##opt 1158 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1159 EXT4_MOUNT2_##opt) 1160 1161 #define ext4_test_and_set_bit __test_and_set_bit_le 1162 #define ext4_set_bit __set_bit_le 1163 #define ext4_set_bit_atomic ext2_set_bit_atomic 1164 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1165 #define ext4_clear_bit __clear_bit_le 1166 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1167 #define ext4_test_bit test_bit_le 1168 #define ext4_find_next_zero_bit find_next_zero_bit_le 1169 #define ext4_find_next_bit find_next_bit_le 1170 1171 extern void ext4_set_bits(void *bm, int cur, int len); 1172 1173 /* 1174 * Maximal mount counts between two filesystem checks 1175 */ 1176 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1177 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1178 1179 /* 1180 * Behaviour when detecting errors 1181 */ 1182 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1183 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1184 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1185 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1186 1187 /* Metadata checksum algorithm codes */ 1188 #define EXT4_CRC32C_CHKSUM 1 1189 1190 /* 1191 * Structure of the super block 1192 */ 1193 struct ext4_super_block { 1194 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1195 __le32 s_blocks_count_lo; /* Blocks count */ 1196 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1197 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1198 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1199 __le32 s_first_data_block; /* First Data Block */ 1200 __le32 s_log_block_size; /* Block size */ 1201 __le32 s_log_cluster_size; /* Allocation cluster size */ 1202 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1203 __le32 s_clusters_per_group; /* # Clusters per group */ 1204 __le32 s_inodes_per_group; /* # Inodes per group */ 1205 __le32 s_mtime; /* Mount time */ 1206 /*30*/ __le32 s_wtime; /* Write time */ 1207 __le16 s_mnt_count; /* Mount count */ 1208 __le16 s_max_mnt_count; /* Maximal mount count */ 1209 __le16 s_magic; /* Magic signature */ 1210 __le16 s_state; /* File system state */ 1211 __le16 s_errors; /* Behaviour when detecting errors */ 1212 __le16 s_minor_rev_level; /* minor revision level */ 1213 /*40*/ __le32 s_lastcheck; /* time of last check */ 1214 __le32 s_checkinterval; /* max. time between checks */ 1215 __le32 s_creator_os; /* OS */ 1216 __le32 s_rev_level; /* Revision level */ 1217 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1218 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1219 /* 1220 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1221 * 1222 * Note: the difference between the compatible feature set and 1223 * the incompatible feature set is that if there is a bit set 1224 * in the incompatible feature set that the kernel doesn't 1225 * know about, it should refuse to mount the filesystem. 1226 * 1227 * e2fsck's requirements are more strict; if it doesn't know 1228 * about a feature in either the compatible or incompatible 1229 * feature set, it must abort and not try to meddle with 1230 * things it doesn't understand... 1231 */ 1232 __le32 s_first_ino; /* First non-reserved inode */ 1233 __le16 s_inode_size; /* size of inode structure */ 1234 __le16 s_block_group_nr; /* block group # of this superblock */ 1235 __le32 s_feature_compat; /* compatible feature set */ 1236 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1237 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1238 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1239 /*78*/ char s_volume_name[16]; /* volume name */ 1240 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1241 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1242 /* 1243 * Performance hints. Directory preallocation should only 1244 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1245 */ 1246 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1247 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1248 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1249 /* 1250 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1251 */ 1252 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1253 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1254 __le32 s_journal_dev; /* device number of journal file */ 1255 __le32 s_last_orphan; /* start of list of inodes to delete */ 1256 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1257 __u8 s_def_hash_version; /* Default hash version to use */ 1258 __u8 s_jnl_backup_type; 1259 __le16 s_desc_size; /* size of group descriptor */ 1260 /*100*/ __le32 s_default_mount_opts; 1261 __le32 s_first_meta_bg; /* First metablock block group */ 1262 __le32 s_mkfs_time; /* When the filesystem was created */ 1263 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1264 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1265 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1266 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1267 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1268 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1269 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1270 __le32 s_flags; /* Miscellaneous flags */ 1271 __le16 s_raid_stride; /* RAID stride */ 1272 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1273 __le64 s_mmp_block; /* Block for multi-mount protection */ 1274 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1275 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1276 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1277 __u8 s_encryption_level; /* versioning level for encryption */ 1278 __u8 s_reserved_pad; /* Padding to next 32bits */ 1279 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1280 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1281 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1282 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1283 snapshot's future use */ 1284 __le32 s_snapshot_list; /* inode number of the head of the 1285 on-disk snapshot list */ 1286 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1287 __le32 s_error_count; /* number of fs errors */ 1288 __le32 s_first_error_time; /* first time an error happened */ 1289 __le32 s_first_error_ino; /* inode involved in first error */ 1290 __le64 s_first_error_block; /* block involved of first error */ 1291 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1292 __le32 s_first_error_line; /* line number where error happened */ 1293 __le32 s_last_error_time; /* most recent time of an error */ 1294 __le32 s_last_error_ino; /* inode involved in last error */ 1295 __le32 s_last_error_line; /* line number where error happened */ 1296 __le64 s_last_error_block; /* block involved of last error */ 1297 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1298 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1299 __u8 s_mount_opts[64]; 1300 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1301 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1302 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1303 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1304 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1305 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1306 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1307 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1308 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1309 __u8 s_wtime_hi; 1310 __u8 s_mtime_hi; 1311 __u8 s_mkfs_time_hi; 1312 __u8 s_lastcheck_hi; 1313 __u8 s_first_error_time_hi; 1314 __u8 s_last_error_time_hi; 1315 __u8 s_pad[2]; 1316 __le32 s_reserved[96]; /* Padding to the end of the block */ 1317 __le32 s_checksum; /* crc32c(superblock) */ 1318 }; 1319 1320 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1321 1322 #ifdef __KERNEL__ 1323 1324 /* 1325 * run-time mount flags 1326 */ 1327 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1328 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1329 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1330 1331 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1332 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1333 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1334 #else 1335 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1336 #endif 1337 1338 /* Number of quota types we support */ 1339 #define EXT4_MAXQUOTAS 3 1340 1341 /* 1342 * fourth extended-fs super-block data in memory 1343 */ 1344 struct ext4_sb_info { 1345 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1346 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1347 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1348 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1349 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1350 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1351 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1352 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1353 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1354 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1355 unsigned long s_overhead; /* # of fs overhead clusters */ 1356 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1357 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1358 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1359 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1360 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1361 struct buffer_head **s_group_desc; 1362 unsigned int s_mount_opt; 1363 unsigned int s_mount_opt2; 1364 unsigned int s_mount_flags; 1365 unsigned int s_def_mount_opt; 1366 ext4_fsblk_t s_sb_block; 1367 atomic64_t s_resv_clusters; 1368 kuid_t s_resuid; 1369 kgid_t s_resgid; 1370 unsigned short s_mount_state; 1371 unsigned short s_pad; 1372 int s_addr_per_block_bits; 1373 int s_desc_per_block_bits; 1374 int s_inode_size; 1375 int s_first_ino; 1376 unsigned int s_inode_readahead_blks; 1377 unsigned int s_inode_goal; 1378 u32 s_hash_seed[4]; 1379 int s_def_hash_version; 1380 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1381 struct percpu_counter s_freeclusters_counter; 1382 struct percpu_counter s_freeinodes_counter; 1383 struct percpu_counter s_dirs_counter; 1384 struct percpu_counter s_dirtyclusters_counter; 1385 struct blockgroup_lock *s_blockgroup_lock; 1386 struct proc_dir_entry *s_proc; 1387 struct kobject s_kobj; 1388 struct completion s_kobj_unregister; 1389 struct super_block *s_sb; 1390 1391 /* Journaling */ 1392 struct journal_s *s_journal; 1393 struct list_head s_orphan; 1394 struct mutex s_orphan_lock; 1395 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1396 unsigned long s_commit_interval; 1397 u32 s_max_batch_time; 1398 u32 s_min_batch_time; 1399 struct block_device *journal_bdev; 1400 #ifdef CONFIG_QUOTA 1401 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1402 int s_jquota_fmt; /* Format of quota to use */ 1403 #endif 1404 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1405 struct rb_root system_blks; 1406 1407 #ifdef EXTENTS_STATS 1408 /* ext4 extents stats */ 1409 unsigned long s_ext_min; 1410 unsigned long s_ext_max; 1411 unsigned long s_depth_max; 1412 spinlock_t s_ext_stats_lock; 1413 unsigned long s_ext_blocks; 1414 unsigned long s_ext_extents; 1415 #endif 1416 1417 /* for buddy allocator */ 1418 struct ext4_group_info ***s_group_info; 1419 struct inode *s_buddy_cache; 1420 spinlock_t s_md_lock; 1421 unsigned short *s_mb_offsets; 1422 unsigned int *s_mb_maxs; 1423 unsigned int s_group_info_size; 1424 unsigned int s_mb_free_pending; 1425 struct list_head s_freed_data_list; /* List of blocks to be freed 1426 after commit completed */ 1427 1428 /* tunables */ 1429 unsigned long s_stripe; 1430 unsigned int s_mb_stream_request; 1431 unsigned int s_mb_max_to_scan; 1432 unsigned int s_mb_min_to_scan; 1433 unsigned int s_mb_stats; 1434 unsigned int s_mb_order2_reqs; 1435 unsigned int s_mb_group_prealloc; 1436 unsigned int s_max_dir_size_kb; 1437 /* where last allocation was done - for stream allocation */ 1438 unsigned long s_mb_last_group; 1439 unsigned long s_mb_last_start; 1440 1441 /* stats for buddy allocator */ 1442 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1443 atomic_t s_bal_success; /* we found long enough chunks */ 1444 atomic_t s_bal_allocated; /* in blocks */ 1445 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1446 atomic_t s_bal_goals; /* goal hits */ 1447 atomic_t s_bal_breaks; /* too long searches */ 1448 atomic_t s_bal_2orders; /* 2^order hits */ 1449 spinlock_t s_bal_lock; 1450 unsigned long s_mb_buddies_generated; 1451 unsigned long long s_mb_generation_time; 1452 atomic_t s_mb_lost_chunks; 1453 atomic_t s_mb_preallocated; 1454 atomic_t s_mb_discarded; 1455 atomic_t s_lock_busy; 1456 1457 /* locality groups */ 1458 struct ext4_locality_group __percpu *s_locality_groups; 1459 1460 /* for write statistics */ 1461 unsigned long s_sectors_written_start; 1462 u64 s_kbytes_written; 1463 1464 /* the size of zero-out chunk */ 1465 unsigned int s_extent_max_zeroout_kb; 1466 1467 unsigned int s_log_groups_per_flex; 1468 struct flex_groups *s_flex_groups; 1469 ext4_group_t s_flex_groups_allocated; 1470 1471 /* workqueue for reserved extent conversions (buffered io) */ 1472 struct workqueue_struct *rsv_conversion_wq; 1473 1474 /* timer for periodic error stats printing */ 1475 struct timer_list s_err_report; 1476 1477 /* Lazy inode table initialization info */ 1478 struct ext4_li_request *s_li_request; 1479 /* Wait multiplier for lazy initialization thread */ 1480 unsigned int s_li_wait_mult; 1481 1482 /* Kernel thread for multiple mount protection */ 1483 struct task_struct *s_mmp_tsk; 1484 1485 /* record the last minlen when FITRIM is called. */ 1486 atomic_t s_last_trim_minblks; 1487 1488 /* Reference to checksum algorithm driver via cryptoapi */ 1489 struct crypto_shash *s_chksum_driver; 1490 1491 /* Precomputed FS UUID checksum for seeding other checksums */ 1492 __u32 s_csum_seed; 1493 1494 /* Reclaim extents from extent status tree */ 1495 struct shrinker s_es_shrinker; 1496 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1497 long s_es_nr_inode; 1498 struct ext4_es_stats s_es_stats; 1499 struct mb_cache *s_ea_block_cache; 1500 struct mb_cache *s_ea_inode_cache; 1501 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1502 1503 /* Ratelimit ext4 messages. */ 1504 struct ratelimit_state s_err_ratelimit_state; 1505 struct ratelimit_state s_warning_ratelimit_state; 1506 struct ratelimit_state s_msg_ratelimit_state; 1507 1508 /* Barrier between changing inodes' journal flags and writepages ops. */ 1509 struct percpu_rw_semaphore s_journal_flag_rwsem; 1510 struct dax_device *s_daxdev; 1511 }; 1512 1513 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1514 { 1515 return sb->s_fs_info; 1516 } 1517 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1518 { 1519 return container_of(inode, struct ext4_inode_info, vfs_inode); 1520 } 1521 1522 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1523 { 1524 return ino == EXT4_ROOT_INO || 1525 (ino >= EXT4_FIRST_INO(sb) && 1526 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1527 } 1528 1529 /* 1530 * Inode dynamic state flags 1531 */ 1532 enum { 1533 EXT4_STATE_JDATA, /* journaled data exists */ 1534 EXT4_STATE_NEW, /* inode is newly created */ 1535 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1536 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1537 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1538 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1539 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1540 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1541 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1542 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1543 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1544 }; 1545 1546 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1547 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1548 { \ 1549 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1550 } \ 1551 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1552 { \ 1553 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1554 } \ 1555 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1556 { \ 1557 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1558 } 1559 1560 /* Add these declarations here only so that these functions can be 1561 * found by name. Otherwise, they are very hard to locate. */ 1562 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1563 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1564 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1565 EXT4_INODE_BIT_FNS(flag, flags, 0) 1566 1567 /* Add these declarations here only so that these functions can be 1568 * found by name. Otherwise, they are very hard to locate. */ 1569 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1570 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1571 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1572 #if (BITS_PER_LONG < 64) 1573 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1574 1575 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1576 { 1577 (ei)->i_state_flags = 0; 1578 } 1579 #else 1580 EXT4_INODE_BIT_FNS(state, flags, 32) 1581 1582 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1583 { 1584 /* We depend on the fact that callers will set i_flags */ 1585 } 1586 #endif 1587 #else 1588 /* Assume that user mode programs are passing in an ext4fs superblock, not 1589 * a kernel struct super_block. This will allow us to call the feature-test 1590 * macros from user land. */ 1591 #define EXT4_SB(sb) (sb) 1592 #endif 1593 1594 /* 1595 * Returns true if the inode is inode is encrypted 1596 */ 1597 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1598 1599 /* 1600 * Codes for operating systems 1601 */ 1602 #define EXT4_OS_LINUX 0 1603 #define EXT4_OS_HURD 1 1604 #define EXT4_OS_MASIX 2 1605 #define EXT4_OS_FREEBSD 3 1606 #define EXT4_OS_LITES 4 1607 1608 /* 1609 * Revision levels 1610 */ 1611 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1612 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1613 1614 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1615 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1616 1617 #define EXT4_GOOD_OLD_INODE_SIZE 128 1618 1619 /* 1620 * Feature set definitions 1621 */ 1622 1623 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1624 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1625 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1626 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1627 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1628 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1629 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1630 1631 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1632 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1633 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1634 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1635 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1636 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1637 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1638 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1639 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1640 /* 1641 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1642 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1643 * all other data structures' checksums. However, the METADATA_CSUM and 1644 * GDT_CSUM bits are mutually exclusive. 1645 */ 1646 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1647 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1648 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1649 1650 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1651 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1652 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1653 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1654 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1655 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1656 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1657 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1658 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1659 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1660 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1661 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1662 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1663 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1664 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1665 1666 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1667 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1668 { \ 1669 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1670 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1671 } \ 1672 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1673 { \ 1674 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1675 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1676 } \ 1677 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1678 { \ 1679 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1680 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1681 } 1682 1683 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1684 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1685 { \ 1686 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1687 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1688 } \ 1689 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1690 { \ 1691 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1692 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1693 } \ 1694 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1695 { \ 1696 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1697 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1698 } 1699 1700 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1701 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1702 { \ 1703 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1704 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1705 } \ 1706 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1707 { \ 1708 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1709 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1710 } \ 1711 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1712 { \ 1713 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1714 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1715 } 1716 1717 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1718 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1719 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1720 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1721 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1722 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1723 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1724 1725 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1726 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1727 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1728 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1729 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1730 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1731 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1732 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1733 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1734 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1735 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1736 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1737 1738 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1739 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1740 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1741 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1742 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1743 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1744 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1745 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1746 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1747 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1748 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1749 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1750 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1751 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1752 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1753 1754 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1755 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1756 EXT4_FEATURE_INCOMPAT_META_BG) 1757 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1758 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1759 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1760 1761 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1762 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1763 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1764 EXT4_FEATURE_INCOMPAT_META_BG) 1765 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1766 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1767 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1768 1769 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1770 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1771 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1772 EXT4_FEATURE_INCOMPAT_META_BG| \ 1773 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1774 EXT4_FEATURE_INCOMPAT_64BIT| \ 1775 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1776 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1777 EXT4_FEATURE_INCOMPAT_MMP | \ 1778 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1779 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1780 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1781 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1782 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1783 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1784 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1785 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1786 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1787 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1788 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1789 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1790 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1791 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1792 EXT4_FEATURE_RO_COMPAT_PROJECT) 1793 1794 #define EXTN_FEATURE_FUNCS(ver) \ 1795 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1796 { \ 1797 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1798 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1799 } \ 1800 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1801 { \ 1802 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1803 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1804 } \ 1805 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1806 { \ 1807 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1808 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1809 } 1810 1811 EXTN_FEATURE_FUNCS(2) 1812 EXTN_FEATURE_FUNCS(3) 1813 EXTN_FEATURE_FUNCS(4) 1814 1815 static inline bool ext4_has_compat_features(struct super_block *sb) 1816 { 1817 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1818 } 1819 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1820 { 1821 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1822 } 1823 static inline bool ext4_has_incompat_features(struct super_block *sb) 1824 { 1825 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1826 } 1827 1828 /* 1829 * Superblock flags 1830 */ 1831 #define EXT4_FLAGS_RESIZING 0 1832 #define EXT4_FLAGS_SHUTDOWN 1 1833 1834 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1835 { 1836 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1837 } 1838 1839 1840 /* 1841 * Default values for user and/or group using reserved blocks 1842 */ 1843 #define EXT4_DEF_RESUID 0 1844 #define EXT4_DEF_RESGID 0 1845 1846 /* 1847 * Default project ID 1848 */ 1849 #define EXT4_DEF_PROJID 0 1850 1851 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1852 1853 /* 1854 * Default mount options 1855 */ 1856 #define EXT4_DEFM_DEBUG 0x0001 1857 #define EXT4_DEFM_BSDGROUPS 0x0002 1858 #define EXT4_DEFM_XATTR_USER 0x0004 1859 #define EXT4_DEFM_ACL 0x0008 1860 #define EXT4_DEFM_UID16 0x0010 1861 #define EXT4_DEFM_JMODE 0x0060 1862 #define EXT4_DEFM_JMODE_DATA 0x0020 1863 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1864 #define EXT4_DEFM_JMODE_WBACK 0x0060 1865 #define EXT4_DEFM_NOBARRIER 0x0100 1866 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1867 #define EXT4_DEFM_DISCARD 0x0400 1868 #define EXT4_DEFM_NODELALLOC 0x0800 1869 1870 /* 1871 * Default journal batch times 1872 */ 1873 #define EXT4_DEF_MIN_BATCH_TIME 0 1874 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1875 1876 /* 1877 * Minimum number of groups in a flexgroup before we separate out 1878 * directories into the first block group of a flexgroup 1879 */ 1880 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1881 1882 /* 1883 * Structure of a directory entry 1884 */ 1885 #define EXT4_NAME_LEN 255 1886 1887 struct ext4_dir_entry { 1888 __le32 inode; /* Inode number */ 1889 __le16 rec_len; /* Directory entry length */ 1890 __le16 name_len; /* Name length */ 1891 char name[EXT4_NAME_LEN]; /* File name */ 1892 }; 1893 1894 /* 1895 * The new version of the directory entry. Since EXT4 structures are 1896 * stored in intel byte order, and the name_len field could never be 1897 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1898 * file_type field. 1899 */ 1900 struct ext4_dir_entry_2 { 1901 __le32 inode; /* Inode number */ 1902 __le16 rec_len; /* Directory entry length */ 1903 __u8 name_len; /* Name length */ 1904 __u8 file_type; 1905 char name[EXT4_NAME_LEN]; /* File name */ 1906 }; 1907 1908 /* 1909 * This is a bogus directory entry at the end of each leaf block that 1910 * records checksums. 1911 */ 1912 struct ext4_dir_entry_tail { 1913 __le32 det_reserved_zero1; /* Pretend to be unused */ 1914 __le16 det_rec_len; /* 12 */ 1915 __u8 det_reserved_zero2; /* Zero name length */ 1916 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1917 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1918 }; 1919 1920 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1921 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1922 ((blocksize) - \ 1923 sizeof(struct ext4_dir_entry_tail)))) 1924 1925 /* 1926 * Ext4 directory file types. Only the low 3 bits are used. The 1927 * other bits are reserved for now. 1928 */ 1929 #define EXT4_FT_UNKNOWN 0 1930 #define EXT4_FT_REG_FILE 1 1931 #define EXT4_FT_DIR 2 1932 #define EXT4_FT_CHRDEV 3 1933 #define EXT4_FT_BLKDEV 4 1934 #define EXT4_FT_FIFO 5 1935 #define EXT4_FT_SOCK 6 1936 #define EXT4_FT_SYMLINK 7 1937 1938 #define EXT4_FT_MAX 8 1939 1940 #define EXT4_FT_DIR_CSUM 0xDE 1941 1942 /* 1943 * EXT4_DIR_PAD defines the directory entries boundaries 1944 * 1945 * NOTE: It must be a multiple of 4 1946 */ 1947 #define EXT4_DIR_PAD 4 1948 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1949 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1950 ~EXT4_DIR_ROUND) 1951 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1952 1953 /* 1954 * If we ever get support for fs block sizes > page_size, we'll need 1955 * to remove the #if statements in the next two functions... 1956 */ 1957 static inline unsigned int 1958 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1959 { 1960 unsigned len = le16_to_cpu(dlen); 1961 1962 #if (PAGE_SIZE >= 65536) 1963 if (len == EXT4_MAX_REC_LEN || len == 0) 1964 return blocksize; 1965 return (len & 65532) | ((len & 3) << 16); 1966 #else 1967 return len; 1968 #endif 1969 } 1970 1971 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1972 { 1973 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1974 BUG(); 1975 #if (PAGE_SIZE >= 65536) 1976 if (len < 65536) 1977 return cpu_to_le16(len); 1978 if (len == blocksize) { 1979 if (blocksize == 65536) 1980 return cpu_to_le16(EXT4_MAX_REC_LEN); 1981 else 1982 return cpu_to_le16(0); 1983 } 1984 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1985 #else 1986 return cpu_to_le16(len); 1987 #endif 1988 } 1989 1990 /* 1991 * Hash Tree Directory indexing 1992 * (c) Daniel Phillips, 2001 1993 */ 1994 1995 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1996 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1997 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 1998 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 1999 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2000 2001 /* Legal values for the dx_root hash_version field: */ 2002 2003 #define DX_HASH_LEGACY 0 2004 #define DX_HASH_HALF_MD4 1 2005 #define DX_HASH_TEA 2 2006 #define DX_HASH_LEGACY_UNSIGNED 3 2007 #define DX_HASH_HALF_MD4_UNSIGNED 4 2008 #define DX_HASH_TEA_UNSIGNED 5 2009 2010 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2011 const void *address, unsigned int length) 2012 { 2013 struct { 2014 struct shash_desc shash; 2015 char ctx[4]; 2016 } desc; 2017 2018 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2019 2020 desc.shash.tfm = sbi->s_chksum_driver; 2021 desc.shash.flags = 0; 2022 *(u32 *)desc.ctx = crc; 2023 2024 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2025 2026 return *(u32 *)desc.ctx; 2027 } 2028 2029 #ifdef __KERNEL__ 2030 2031 /* hash info structure used by the directory hash */ 2032 struct dx_hash_info 2033 { 2034 u32 hash; 2035 u32 minor_hash; 2036 int hash_version; 2037 u32 *seed; 2038 }; 2039 2040 2041 /* 32 and 64 bit signed EOF for dx directories */ 2042 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2043 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2044 2045 2046 /* 2047 * Control parameters used by ext4_htree_next_block 2048 */ 2049 #define HASH_NB_ALWAYS 1 2050 2051 struct ext4_filename { 2052 const struct qstr *usr_fname; 2053 struct fscrypt_str disk_name; 2054 struct dx_hash_info hinfo; 2055 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2056 struct fscrypt_str crypto_buf; 2057 #endif 2058 }; 2059 2060 #define fname_name(p) ((p)->disk_name.name) 2061 #define fname_len(p) ((p)->disk_name.len) 2062 2063 /* 2064 * Describe an inode's exact location on disk and in memory 2065 */ 2066 struct ext4_iloc 2067 { 2068 struct buffer_head *bh; 2069 unsigned long offset; 2070 ext4_group_t block_group; 2071 }; 2072 2073 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2074 { 2075 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2076 } 2077 2078 static inline bool ext4_is_quota_file(struct inode *inode) 2079 { 2080 return IS_NOQUOTA(inode) && 2081 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2082 } 2083 2084 /* 2085 * This structure is stuffed into the struct file's private_data field 2086 * for directories. It is where we put information so that we can do 2087 * readdir operations in hash tree order. 2088 */ 2089 struct dir_private_info { 2090 struct rb_root root; 2091 struct rb_node *curr_node; 2092 struct fname *extra_fname; 2093 loff_t last_pos; 2094 __u32 curr_hash; 2095 __u32 curr_minor_hash; 2096 __u32 next_hash; 2097 }; 2098 2099 /* calculate the first block number of the group */ 2100 static inline ext4_fsblk_t 2101 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2102 { 2103 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2104 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2105 } 2106 2107 /* 2108 * Special error return code only used by dx_probe() and its callers. 2109 */ 2110 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2111 2112 /* htree levels for ext4 */ 2113 #define EXT4_HTREE_LEVEL_COMPAT 2 2114 #define EXT4_HTREE_LEVEL 3 2115 2116 static inline int ext4_dir_htree_level(struct super_block *sb) 2117 { 2118 return ext4_has_feature_largedir(sb) ? 2119 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2120 } 2121 2122 /* 2123 * Timeout and state flag for lazy initialization inode thread. 2124 */ 2125 #define EXT4_DEF_LI_WAIT_MULT 10 2126 #define EXT4_DEF_LI_MAX_START_DELAY 5 2127 #define EXT4_LAZYINIT_QUIT 0x0001 2128 #define EXT4_LAZYINIT_RUNNING 0x0002 2129 2130 /* 2131 * Lazy inode table initialization info 2132 */ 2133 struct ext4_lazy_init { 2134 unsigned long li_state; 2135 struct list_head li_request_list; 2136 struct mutex li_list_mtx; 2137 }; 2138 2139 struct ext4_li_request { 2140 struct super_block *lr_super; 2141 struct ext4_sb_info *lr_sbi; 2142 ext4_group_t lr_next_group; 2143 struct list_head lr_request; 2144 unsigned long lr_next_sched; 2145 unsigned long lr_timeout; 2146 }; 2147 2148 struct ext4_features { 2149 struct kobject f_kobj; 2150 struct completion f_kobj_unregister; 2151 }; 2152 2153 /* 2154 * This structure will be used for multiple mount protection. It will be 2155 * written into the block number saved in the s_mmp_block field in the 2156 * superblock. Programs that check MMP should assume that if 2157 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2158 * to use the filesystem, regardless of how old the timestamp is. 2159 */ 2160 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2161 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2162 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2163 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2164 2165 struct mmp_struct { 2166 __le32 mmp_magic; /* Magic number for MMP */ 2167 __le32 mmp_seq; /* Sequence no. updated periodically */ 2168 2169 /* 2170 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2171 * purposes and do not affect the correctness of the algorithm 2172 */ 2173 __le64 mmp_time; /* Time last updated */ 2174 char mmp_nodename[64]; /* Node which last updated MMP block */ 2175 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2176 2177 /* 2178 * mmp_check_interval is used to verify if the MMP block has been 2179 * updated on the block device. The value is updated based on the 2180 * maximum time to write the MMP block during an update cycle. 2181 */ 2182 __le16 mmp_check_interval; 2183 2184 __le16 mmp_pad1; 2185 __le32 mmp_pad2[226]; 2186 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2187 }; 2188 2189 /* arguments passed to the mmp thread */ 2190 struct mmpd_data { 2191 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2192 struct super_block *sb; /* super block of the fs */ 2193 }; 2194 2195 /* 2196 * Check interval multiplier 2197 * The MMP block is written every update interval and initially checked every 2198 * update interval x the multiplier (the value is then adapted based on the 2199 * write latency). The reason is that writes can be delayed under load and we 2200 * don't want readers to incorrectly assume that the filesystem is no longer 2201 * in use. 2202 */ 2203 #define EXT4_MMP_CHECK_MULT 2UL 2204 2205 /* 2206 * Minimum interval for MMP checking in seconds. 2207 */ 2208 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2209 2210 /* 2211 * Maximum interval for MMP checking in seconds. 2212 */ 2213 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2214 2215 /* 2216 * Function prototypes 2217 */ 2218 2219 /* 2220 * Ok, these declarations are also in <linux/kernel.h> but none of the 2221 * ext4 source programs needs to include it so they are duplicated here. 2222 */ 2223 # define NORET_TYPE /**/ 2224 # define ATTRIB_NORET __attribute__((noreturn)) 2225 # define NORET_AND noreturn, 2226 2227 /* bitmap.c */ 2228 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2229 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2230 struct ext4_group_desc *gdp, 2231 struct buffer_head *bh, int sz); 2232 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2233 struct ext4_group_desc *gdp, 2234 struct buffer_head *bh, int sz); 2235 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2236 struct ext4_group_desc *gdp, 2237 struct buffer_head *bh); 2238 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2239 struct ext4_group_desc *gdp, 2240 struct buffer_head *bh); 2241 2242 /* balloc.c */ 2243 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2244 ext4_fsblk_t blocknr, 2245 ext4_group_t *blockgrpp, 2246 ext4_grpblk_t *offsetp); 2247 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2248 ext4_fsblk_t block); 2249 2250 extern unsigned int ext4_block_group(struct super_block *sb, 2251 ext4_fsblk_t blocknr); 2252 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2253 ext4_fsblk_t blocknr); 2254 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2255 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2256 ext4_group_t group); 2257 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2258 ext4_fsblk_t goal, 2259 unsigned int flags, 2260 unsigned long *count, 2261 int *errp); 2262 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2263 s64 nclusters, unsigned int flags); 2264 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2265 extern void ext4_check_blocks_bitmap(struct super_block *); 2266 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2267 ext4_group_t block_group, 2268 struct buffer_head ** bh); 2269 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2270 2271 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2272 ext4_group_t block_group); 2273 extern int ext4_wait_block_bitmap(struct super_block *sb, 2274 ext4_group_t block_group, 2275 struct buffer_head *bh); 2276 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2277 ext4_group_t block_group); 2278 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2279 ext4_group_t block_group, 2280 struct ext4_group_desc *gdp); 2281 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2282 2283 static inline bool ext4_encrypted_inode(struct inode *inode) 2284 { 2285 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2286 } 2287 2288 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2289 static inline int ext4_fname_setup_filename(struct inode *dir, 2290 const struct qstr *iname, 2291 int lookup, struct ext4_filename *fname) 2292 { 2293 struct fscrypt_name name; 2294 int err; 2295 2296 memset(fname, 0, sizeof(struct ext4_filename)); 2297 2298 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2299 2300 fname->usr_fname = name.usr_fname; 2301 fname->disk_name = name.disk_name; 2302 fname->hinfo.hash = name.hash; 2303 fname->hinfo.minor_hash = name.minor_hash; 2304 fname->crypto_buf = name.crypto_buf; 2305 return err; 2306 } 2307 2308 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2309 { 2310 struct fscrypt_name name; 2311 2312 name.crypto_buf = fname->crypto_buf; 2313 fscrypt_free_filename(&name); 2314 2315 fname->crypto_buf.name = NULL; 2316 fname->usr_fname = NULL; 2317 fname->disk_name.name = NULL; 2318 } 2319 #else 2320 static inline int ext4_fname_setup_filename(struct inode *dir, 2321 const struct qstr *iname, 2322 int lookup, struct ext4_filename *fname) 2323 { 2324 fname->usr_fname = iname; 2325 fname->disk_name.name = (unsigned char *) iname->name; 2326 fname->disk_name.len = iname->len; 2327 return 0; 2328 } 2329 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2330 2331 #endif 2332 2333 /* dir.c */ 2334 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2335 struct file *, 2336 struct ext4_dir_entry_2 *, 2337 struct buffer_head *, char *, int, 2338 unsigned int); 2339 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2340 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2341 (de), (bh), (buf), (size), (offset))) 2342 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2343 __u32 minor_hash, 2344 struct ext4_dir_entry_2 *dirent, 2345 struct fscrypt_str *ent_name); 2346 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2347 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2348 struct buffer_head *bh, 2349 void *buf, int buf_size, 2350 struct ext4_filename *fname, 2351 struct ext4_dir_entry_2 **dest_de); 2352 void ext4_insert_dentry(struct inode *inode, 2353 struct ext4_dir_entry_2 *de, 2354 int buf_size, 2355 struct ext4_filename *fname); 2356 static inline void ext4_update_dx_flag(struct inode *inode) 2357 { 2358 if (!ext4_has_feature_dir_index(inode->i_sb)) 2359 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2360 } 2361 static const unsigned char ext4_filetype_table[] = { 2362 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2363 }; 2364 2365 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2366 { 2367 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2368 return DT_UNKNOWN; 2369 2370 return ext4_filetype_table[filetype]; 2371 } 2372 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2373 void *buf, int buf_size); 2374 2375 /* fsync.c */ 2376 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2377 2378 /* hash.c */ 2379 extern int ext4fs_dirhash(const char *name, int len, struct 2380 dx_hash_info *hinfo); 2381 2382 /* ialloc.c */ 2383 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2384 const struct qstr *qstr, __u32 goal, 2385 uid_t *owner, __u32 i_flags, 2386 int handle_type, unsigned int line_no, 2387 int nblocks); 2388 2389 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2390 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2391 i_flags, 0, 0, 0) 2392 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2393 type, nblocks) \ 2394 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2395 0, (type), __LINE__, (nblocks)) 2396 2397 2398 extern void ext4_free_inode(handle_t *, struct inode *); 2399 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2400 extern unsigned long ext4_count_free_inodes(struct super_block *); 2401 extern unsigned long ext4_count_dirs(struct super_block *); 2402 extern void ext4_check_inodes_bitmap(struct super_block *); 2403 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2404 extern int ext4_init_inode_table(struct super_block *sb, 2405 ext4_group_t group, int barrier); 2406 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2407 2408 /* mballoc.c */ 2409 extern const struct seq_operations ext4_mb_seq_groups_ops; 2410 extern long ext4_mb_stats; 2411 extern long ext4_mb_max_to_scan; 2412 extern int ext4_mb_init(struct super_block *); 2413 extern int ext4_mb_release(struct super_block *); 2414 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2415 struct ext4_allocation_request *, int *); 2416 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2417 extern void ext4_discard_preallocations(struct inode *); 2418 extern int __init ext4_init_mballoc(void); 2419 extern void ext4_exit_mballoc(void); 2420 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2421 struct buffer_head *bh, ext4_fsblk_t block, 2422 unsigned long count, int flags); 2423 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2424 ext4_group_t ngroups); 2425 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2426 ext4_group_t i, struct ext4_group_desc *desc); 2427 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2428 ext4_fsblk_t block, unsigned long count); 2429 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2430 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2431 2432 /* inode.c */ 2433 int ext4_inode_is_fast_symlink(struct inode *inode); 2434 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2435 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2436 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2437 bool wait, struct buffer_head **bhs); 2438 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2439 struct buffer_head *bh_result, int create); 2440 int ext4_get_block(struct inode *inode, sector_t iblock, 2441 struct buffer_head *bh_result, int create); 2442 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2443 struct buffer_head *bh_result, int create); 2444 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2445 struct buffer_head *bh, int create); 2446 int ext4_walk_page_buffers(handle_t *handle, 2447 struct buffer_head *head, 2448 unsigned from, 2449 unsigned to, 2450 int *partial, 2451 int (*fn)(handle_t *handle, 2452 struct buffer_head *bh)); 2453 int do_journal_get_write_access(handle_t *handle, 2454 struct buffer_head *bh); 2455 #define FALL_BACK_TO_NONDELALLOC 1 2456 #define CONVERT_INLINE_DATA 2 2457 2458 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2459 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2460 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2461 extern int ext4_setattr(struct dentry *, struct iattr *); 2462 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2463 extern void ext4_evict_inode(struct inode *); 2464 extern void ext4_clear_inode(struct inode *); 2465 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2466 extern int ext4_sync_inode(handle_t *, struct inode *); 2467 extern void ext4_dirty_inode(struct inode *, int); 2468 extern int ext4_change_inode_journal_flag(struct inode *, int); 2469 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2470 extern int ext4_inode_attach_jinode(struct inode *inode); 2471 extern int ext4_can_truncate(struct inode *inode); 2472 extern int ext4_truncate(struct inode *); 2473 extern int ext4_break_layouts(struct inode *); 2474 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2475 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2476 extern void ext4_set_inode_flags(struct inode *); 2477 extern int ext4_alloc_da_blocks(struct inode *inode); 2478 extern void ext4_set_aops(struct inode *inode); 2479 extern int ext4_writepage_trans_blocks(struct inode *); 2480 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2481 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2482 loff_t lstart, loff_t lend); 2483 extern int ext4_page_mkwrite(struct vm_fault *vmf); 2484 extern int ext4_filemap_fault(struct vm_fault *vmf); 2485 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2486 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2487 extern void ext4_da_update_reserve_space(struct inode *inode, 2488 int used, int quota_claim); 2489 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2490 ext4_fsblk_t pblk, ext4_lblk_t len); 2491 2492 /* indirect.c */ 2493 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2494 struct ext4_map_blocks *map, int flags); 2495 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2496 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2497 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2498 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2499 ext4_lblk_t start, ext4_lblk_t end); 2500 2501 /* ioctl.c */ 2502 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2503 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2504 2505 /* migrate.c */ 2506 extern int ext4_ext_migrate(struct inode *); 2507 extern int ext4_ind_migrate(struct inode *inode); 2508 2509 /* namei.c */ 2510 extern int ext4_dirent_csum_verify(struct inode *inode, 2511 struct ext4_dir_entry *dirent); 2512 extern int ext4_orphan_add(handle_t *, struct inode *); 2513 extern int ext4_orphan_del(handle_t *, struct inode *); 2514 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2515 __u32 start_minor_hash, __u32 *next_hash); 2516 extern int ext4_search_dir(struct buffer_head *bh, 2517 char *search_buf, 2518 int buf_size, 2519 struct inode *dir, 2520 struct ext4_filename *fname, 2521 unsigned int offset, 2522 struct ext4_dir_entry_2 **res_dir); 2523 extern int ext4_generic_delete_entry(handle_t *handle, 2524 struct inode *dir, 2525 struct ext4_dir_entry_2 *de_del, 2526 struct buffer_head *bh, 2527 void *entry_buf, 2528 int buf_size, 2529 int csum_size); 2530 extern bool ext4_empty_dir(struct inode *inode); 2531 2532 /* resize.c */ 2533 extern int ext4_group_add(struct super_block *sb, 2534 struct ext4_new_group_data *input); 2535 extern int ext4_group_extend(struct super_block *sb, 2536 struct ext4_super_block *es, 2537 ext4_fsblk_t n_blocks_count); 2538 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2539 2540 /* super.c */ 2541 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2542 extern int ext4_calculate_overhead(struct super_block *sb); 2543 extern void ext4_superblock_csum_set(struct super_block *sb); 2544 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2545 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2546 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2547 ext4_group_t ngroup); 2548 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2549 char nbuf[16]); 2550 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2551 ext4_group_t block_group, 2552 unsigned int flags); 2553 2554 extern __printf(4, 5) 2555 void __ext4_error(struct super_block *, const char *, unsigned int, 2556 const char *, ...); 2557 extern __printf(5, 6) 2558 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2559 const char *, ...); 2560 extern __printf(5, 6) 2561 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2562 const char *, ...); 2563 extern void __ext4_std_error(struct super_block *, const char *, 2564 unsigned int, int); 2565 extern __printf(4, 5) 2566 void __ext4_abort(struct super_block *, const char *, unsigned int, 2567 const char *, ...); 2568 extern __printf(4, 5) 2569 void __ext4_warning(struct super_block *, const char *, unsigned int, 2570 const char *, ...); 2571 extern __printf(4, 5) 2572 void __ext4_warning_inode(const struct inode *inode, const char *function, 2573 unsigned int line, const char *fmt, ...); 2574 extern __printf(3, 4) 2575 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2576 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2577 const char *, unsigned int, const char *); 2578 extern __printf(7, 8) 2579 void __ext4_grp_locked_error(const char *, unsigned int, 2580 struct super_block *, ext4_group_t, 2581 unsigned long, ext4_fsblk_t, 2582 const char *, ...); 2583 2584 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2585 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2586 2587 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2588 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2589 2590 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2591 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2592 2593 #ifdef CONFIG_PRINTK 2594 2595 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2596 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2597 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2598 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2599 #define ext4_error(sb, fmt, ...) \ 2600 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2601 #define ext4_abort(sb, fmt, ...) \ 2602 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2603 #define ext4_warning(sb, fmt, ...) \ 2604 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2605 #define ext4_warning_inode(inode, fmt, ...) \ 2606 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2607 #define ext4_msg(sb, level, fmt, ...) \ 2608 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2609 #define dump_mmp_msg(sb, mmp, msg) \ 2610 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2611 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2612 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2613 fmt, ##__VA_ARGS__) 2614 2615 #else 2616 2617 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2618 do { \ 2619 no_printk(fmt, ##__VA_ARGS__); \ 2620 __ext4_error_inode(inode, "", 0, block, " "); \ 2621 } while (0) 2622 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2623 do { \ 2624 no_printk(fmt, ##__VA_ARGS__); \ 2625 __ext4_error_file(file, "", 0, block, " "); \ 2626 } while (0) 2627 #define ext4_error(sb, fmt, ...) \ 2628 do { \ 2629 no_printk(fmt, ##__VA_ARGS__); \ 2630 __ext4_error(sb, "", 0, " "); \ 2631 } while (0) 2632 #define ext4_abort(sb, fmt, ...) \ 2633 do { \ 2634 no_printk(fmt, ##__VA_ARGS__); \ 2635 __ext4_abort(sb, "", 0, " "); \ 2636 } while (0) 2637 #define ext4_warning(sb, fmt, ...) \ 2638 do { \ 2639 no_printk(fmt, ##__VA_ARGS__); \ 2640 __ext4_warning(sb, "", 0, " "); \ 2641 } while (0) 2642 #define ext4_warning_inode(inode, fmt, ...) \ 2643 do { \ 2644 no_printk(fmt, ##__VA_ARGS__); \ 2645 __ext4_warning_inode(inode, "", 0, " "); \ 2646 } while (0) 2647 #define ext4_msg(sb, level, fmt, ...) \ 2648 do { \ 2649 no_printk(fmt, ##__VA_ARGS__); \ 2650 __ext4_msg(sb, "", " "); \ 2651 } while (0) 2652 #define dump_mmp_msg(sb, mmp, msg) \ 2653 __dump_mmp_msg(sb, mmp, "", 0, "") 2654 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2655 do { \ 2656 no_printk(fmt, ##__VA_ARGS__); \ 2657 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2658 } while (0) 2659 2660 #endif 2661 2662 extern void ext4_update_dynamic_rev(struct super_block *sb); 2663 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2664 __u32 compat); 2665 extern int ext4_update_rocompat_feature(handle_t *handle, 2666 struct super_block *sb, __u32 rocompat); 2667 extern int ext4_update_incompat_feature(handle_t *handle, 2668 struct super_block *sb, __u32 incompat); 2669 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2670 struct ext4_group_desc *bg); 2671 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2672 struct ext4_group_desc *bg); 2673 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2674 struct ext4_group_desc *bg); 2675 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2676 struct ext4_group_desc *bg); 2677 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2678 struct ext4_group_desc *bg); 2679 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2680 struct ext4_group_desc *bg); 2681 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2682 struct ext4_group_desc *bg); 2683 extern void ext4_block_bitmap_set(struct super_block *sb, 2684 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2685 extern void ext4_inode_bitmap_set(struct super_block *sb, 2686 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2687 extern void ext4_inode_table_set(struct super_block *sb, 2688 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2689 extern void ext4_free_group_clusters_set(struct super_block *sb, 2690 struct ext4_group_desc *bg, 2691 __u32 count); 2692 extern void ext4_free_inodes_set(struct super_block *sb, 2693 struct ext4_group_desc *bg, __u32 count); 2694 extern void ext4_used_dirs_set(struct super_block *sb, 2695 struct ext4_group_desc *bg, __u32 count); 2696 extern void ext4_itable_unused_set(struct super_block *sb, 2697 struct ext4_group_desc *bg, __u32 count); 2698 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2699 struct ext4_group_desc *gdp); 2700 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2701 struct ext4_group_desc *gdp); 2702 extern int ext4_register_li_request(struct super_block *sb, 2703 ext4_group_t first_not_zeroed); 2704 2705 static inline int ext4_has_metadata_csum(struct super_block *sb) 2706 { 2707 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2708 !EXT4_SB(sb)->s_chksum_driver); 2709 2710 return ext4_has_feature_metadata_csum(sb) && 2711 (EXT4_SB(sb)->s_chksum_driver != NULL); 2712 } 2713 2714 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2715 { 2716 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2717 } 2718 2719 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2720 { 2721 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2722 le32_to_cpu(es->s_blocks_count_lo); 2723 } 2724 2725 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2726 { 2727 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2728 le32_to_cpu(es->s_r_blocks_count_lo); 2729 } 2730 2731 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2732 { 2733 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2734 le32_to_cpu(es->s_free_blocks_count_lo); 2735 } 2736 2737 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2738 ext4_fsblk_t blk) 2739 { 2740 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2741 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2742 } 2743 2744 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2745 ext4_fsblk_t blk) 2746 { 2747 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2748 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2749 } 2750 2751 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2752 ext4_fsblk_t blk) 2753 { 2754 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2755 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2756 } 2757 2758 static inline loff_t ext4_isize(struct super_block *sb, 2759 struct ext4_inode *raw_inode) 2760 { 2761 if (ext4_has_feature_largedir(sb) || 2762 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2763 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2764 le32_to_cpu(raw_inode->i_size_lo); 2765 2766 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2767 } 2768 2769 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2770 { 2771 raw_inode->i_size_lo = cpu_to_le32(i_size); 2772 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2773 } 2774 2775 static inline 2776 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2777 ext4_group_t group) 2778 { 2779 struct ext4_group_info ***grp_info; 2780 long indexv, indexh; 2781 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2782 grp_info = EXT4_SB(sb)->s_group_info; 2783 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2784 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2785 return grp_info[indexv][indexh]; 2786 } 2787 2788 /* 2789 * Reading s_groups_count requires using smp_rmb() afterwards. See 2790 * the locking protocol documented in the comments of ext4_group_add() 2791 * in resize.c 2792 */ 2793 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2794 { 2795 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2796 2797 smp_rmb(); 2798 return ngroups; 2799 } 2800 2801 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2802 ext4_group_t block_group) 2803 { 2804 return block_group >> sbi->s_log_groups_per_flex; 2805 } 2806 2807 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2808 { 2809 return 1 << sbi->s_log_groups_per_flex; 2810 } 2811 2812 #define ext4_std_error(sb, errno) \ 2813 do { \ 2814 if ((errno)) \ 2815 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2816 } while (0) 2817 2818 #ifdef CONFIG_SMP 2819 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2820 * counters. So we need to make sure we have free clusters more 2821 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2822 */ 2823 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2824 #else 2825 #define EXT4_FREECLUSTERS_WATERMARK 0 2826 #endif 2827 2828 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2829 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2830 { 2831 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2832 !inode_is_locked(inode)); 2833 down_write(&EXT4_I(inode)->i_data_sem); 2834 if (newsize > EXT4_I(inode)->i_disksize) 2835 EXT4_I(inode)->i_disksize = newsize; 2836 up_write(&EXT4_I(inode)->i_data_sem); 2837 } 2838 2839 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2840 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2841 { 2842 int changed = 0; 2843 2844 if (newsize > inode->i_size) { 2845 i_size_write(inode, newsize); 2846 changed = 1; 2847 } 2848 if (newsize > EXT4_I(inode)->i_disksize) { 2849 ext4_update_i_disksize(inode, newsize); 2850 changed |= 2; 2851 } 2852 return changed; 2853 } 2854 2855 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2856 loff_t len); 2857 2858 struct ext4_group_info { 2859 unsigned long bb_state; 2860 struct rb_root bb_free_root; 2861 ext4_grpblk_t bb_first_free; /* first free block */ 2862 ext4_grpblk_t bb_free; /* total free blocks */ 2863 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2864 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2865 struct list_head bb_prealloc_list; 2866 #ifdef DOUBLE_CHECK 2867 void *bb_bitmap; 2868 #endif 2869 struct rw_semaphore alloc_sem; 2870 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2871 * regions, index is order. 2872 * bb_counters[3] = 5 means 2873 * 5 free 8-block regions. */ 2874 }; 2875 2876 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2877 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2878 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2879 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2880 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2881 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2882 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2883 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2884 2885 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2886 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2887 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2888 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2889 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2890 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2891 2892 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2893 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2894 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2895 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2896 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2897 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2898 2899 #define EXT4_MAX_CONTENTION 8 2900 #define EXT4_CONTENTION_THRESHOLD 2 2901 2902 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2903 ext4_group_t group) 2904 { 2905 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2906 } 2907 2908 /* 2909 * Returns true if the filesystem is busy enough that attempts to 2910 * access the block group locks has run into contention. 2911 */ 2912 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2913 { 2914 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2915 } 2916 2917 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2918 { 2919 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2920 if (spin_trylock(lock)) 2921 /* 2922 * We're able to grab the lock right away, so drop the 2923 * lock contention counter. 2924 */ 2925 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2926 else { 2927 /* 2928 * The lock is busy, so bump the contention counter, 2929 * and then wait on the spin lock. 2930 */ 2931 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2932 EXT4_MAX_CONTENTION); 2933 spin_lock(lock); 2934 } 2935 } 2936 2937 static inline void ext4_unlock_group(struct super_block *sb, 2938 ext4_group_t group) 2939 { 2940 spin_unlock(ext4_group_lock_ptr(sb, group)); 2941 } 2942 2943 /* 2944 * Block validity checking 2945 */ 2946 #define ext4_check_indirect_blockref(inode, bh) \ 2947 ext4_check_blockref(__func__, __LINE__, inode, \ 2948 (__le32 *)(bh)->b_data, \ 2949 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2950 2951 #define ext4_ind_check_inode(inode) \ 2952 ext4_check_blockref(__func__, __LINE__, inode, \ 2953 EXT4_I(inode)->i_data, \ 2954 EXT4_NDIR_BLOCKS) 2955 2956 /* 2957 * Inodes and files operations 2958 */ 2959 2960 /* dir.c */ 2961 extern const struct file_operations ext4_dir_operations; 2962 2963 /* file.c */ 2964 extern const struct inode_operations ext4_file_inode_operations; 2965 extern const struct file_operations ext4_file_operations; 2966 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2967 2968 /* inline.c */ 2969 extern int ext4_get_max_inline_size(struct inode *inode); 2970 extern int ext4_find_inline_data_nolock(struct inode *inode); 2971 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2972 unsigned int len); 2973 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2974 2975 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2976 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2977 struct inode *inode, 2978 loff_t pos, unsigned len, 2979 unsigned flags, 2980 struct page **pagep); 2981 extern int ext4_write_inline_data_end(struct inode *inode, 2982 loff_t pos, unsigned len, 2983 unsigned copied, 2984 struct page *page); 2985 extern struct buffer_head * 2986 ext4_journalled_write_inline_data(struct inode *inode, 2987 unsigned len, 2988 struct page *page); 2989 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2990 struct inode *inode, 2991 loff_t pos, unsigned len, 2992 unsigned flags, 2993 struct page **pagep, 2994 void **fsdata); 2995 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2996 unsigned len, unsigned copied, 2997 struct page *page); 2998 extern int ext4_try_add_inline_entry(handle_t *handle, 2999 struct ext4_filename *fname, 3000 struct inode *dir, struct inode *inode); 3001 extern int ext4_try_create_inline_dir(handle_t *handle, 3002 struct inode *parent, 3003 struct inode *inode); 3004 extern int ext4_read_inline_dir(struct file *filp, 3005 struct dir_context *ctx, 3006 int *has_inline_data); 3007 extern int htree_inlinedir_to_tree(struct file *dir_file, 3008 struct inode *dir, ext4_lblk_t block, 3009 struct dx_hash_info *hinfo, 3010 __u32 start_hash, __u32 start_minor_hash, 3011 int *has_inline_data); 3012 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3013 struct ext4_filename *fname, 3014 struct ext4_dir_entry_2 **res_dir, 3015 int *has_inline_data); 3016 extern int ext4_delete_inline_entry(handle_t *handle, 3017 struct inode *dir, 3018 struct ext4_dir_entry_2 *de_del, 3019 struct buffer_head *bh, 3020 int *has_inline_data); 3021 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3022 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3023 struct ext4_dir_entry_2 **parent_de, 3024 int *retval); 3025 extern int ext4_inline_data_fiemap(struct inode *inode, 3026 struct fiemap_extent_info *fieinfo, 3027 int *has_inline, __u64 start, __u64 len); 3028 3029 struct iomap; 3030 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3031 3032 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3033 3034 extern int ext4_convert_inline_data(struct inode *inode); 3035 3036 static inline int ext4_has_inline_data(struct inode *inode) 3037 { 3038 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3039 EXT4_I(inode)->i_inline_off; 3040 } 3041 3042 /* namei.c */ 3043 extern const struct inode_operations ext4_dir_inode_operations; 3044 extern const struct inode_operations ext4_special_inode_operations; 3045 extern struct dentry *ext4_get_parent(struct dentry *child); 3046 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3047 struct ext4_dir_entry_2 *de, 3048 int blocksize, int csum_size, 3049 unsigned int parent_ino, int dotdot_real_len); 3050 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3051 unsigned int blocksize); 3052 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3053 struct inode *inode, 3054 struct buffer_head *bh); 3055 #define S_SHIFT 12 3056 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3057 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3058 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3059 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3060 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3061 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3062 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3063 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3064 }; 3065 3066 static inline void ext4_set_de_type(struct super_block *sb, 3067 struct ext4_dir_entry_2 *de, 3068 umode_t mode) { 3069 if (ext4_has_feature_filetype(sb)) 3070 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3071 } 3072 3073 /* readpages.c */ 3074 extern int ext4_mpage_readpages(struct address_space *mapping, 3075 struct list_head *pages, struct page *page, 3076 unsigned nr_pages); 3077 3078 /* symlink.c */ 3079 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3080 extern const struct inode_operations ext4_symlink_inode_operations; 3081 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3082 3083 /* sysfs.c */ 3084 extern int ext4_register_sysfs(struct super_block *sb); 3085 extern void ext4_unregister_sysfs(struct super_block *sb); 3086 extern int __init ext4_init_sysfs(void); 3087 extern void ext4_exit_sysfs(void); 3088 3089 /* block_validity */ 3090 extern void ext4_release_system_zone(struct super_block *sb); 3091 extern int ext4_setup_system_zone(struct super_block *sb); 3092 extern int __init ext4_init_system_zone(void); 3093 extern void ext4_exit_system_zone(void); 3094 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3095 ext4_fsblk_t start_blk, 3096 unsigned int count); 3097 extern int ext4_check_blockref(const char *, unsigned int, 3098 struct inode *, __le32 *, unsigned int); 3099 3100 /* extents.c */ 3101 struct ext4_ext_path; 3102 struct ext4_extent; 3103 3104 /* 3105 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3106 * __le32. 3107 */ 3108 #define EXT_MAX_BLOCKS 0xffffffff 3109 3110 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3111 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3112 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3113 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3114 struct ext4_map_blocks *map, int flags); 3115 extern int ext4_ext_truncate(handle_t *, struct inode *); 3116 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3117 ext4_lblk_t end); 3118 extern void ext4_ext_init(struct super_block *); 3119 extern void ext4_ext_release(struct super_block *); 3120 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3121 loff_t len); 3122 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3123 loff_t offset, ssize_t len); 3124 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3125 struct ext4_map_blocks *map, int flags); 3126 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3127 ext4_lblk_t lblocks); 3128 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3129 int num, 3130 struct ext4_ext_path *path); 3131 extern int ext4_can_extents_be_merged(struct inode *inode, 3132 struct ext4_extent *ex1, 3133 struct ext4_extent *ex2); 3134 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3135 struct ext4_ext_path **, 3136 struct ext4_extent *, int); 3137 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3138 struct ext4_ext_path **, 3139 int flags); 3140 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3141 extern int ext4_ext_check_inode(struct inode *inode); 3142 extern int ext4_find_delalloc_range(struct inode *inode, 3143 ext4_lblk_t lblk_start, 3144 ext4_lblk_t lblk_end); 3145 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3146 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3147 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3148 __u64 start, __u64 len); 3149 extern int ext4_ext_precache(struct inode *inode); 3150 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3151 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3152 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3153 struct inode *inode2, ext4_lblk_t lblk1, 3154 ext4_lblk_t lblk2, ext4_lblk_t count, 3155 int mark_unwritten,int *err); 3156 3157 /* move_extent.c */ 3158 extern void ext4_double_down_write_data_sem(struct inode *first, 3159 struct inode *second); 3160 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3161 struct inode *donor_inode); 3162 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3163 __u64 start_orig, __u64 start_donor, 3164 __u64 len, __u64 *moved_len); 3165 3166 /* page-io.c */ 3167 extern int __init ext4_init_pageio(void); 3168 extern void ext4_exit_pageio(void); 3169 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3170 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3171 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3172 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3173 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3174 struct writeback_control *wbc); 3175 extern void ext4_end_io_rsv_work(struct work_struct *work); 3176 extern void ext4_io_submit(struct ext4_io_submit *io); 3177 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3178 struct page *page, 3179 int len, 3180 struct writeback_control *wbc, 3181 bool keep_towrite); 3182 3183 /* mmp.c */ 3184 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3185 3186 /* 3187 * Add new method to test whether block and inode bitmaps are properly 3188 * initialized. With uninit_bg reading the block from disk is not enough 3189 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3190 */ 3191 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3192 3193 static inline int bitmap_uptodate(struct buffer_head *bh) 3194 { 3195 return (buffer_uptodate(bh) && 3196 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3197 } 3198 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3199 { 3200 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3201 } 3202 3203 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3204 3205 /* For ioend & aio unwritten conversion wait queues */ 3206 #define EXT4_WQ_HASH_SZ 37 3207 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3208 EXT4_WQ_HASH_SZ]) 3209 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3210 3211 extern int ext4_resize_begin(struct super_block *sb); 3212 extern void ext4_resize_end(struct super_block *sb); 3213 3214 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3215 struct ext4_io_end *io_end) 3216 { 3217 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3218 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3219 atomic_inc(&EXT4_I(inode)->i_unwritten); 3220 } 3221 } 3222 3223 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3224 { 3225 struct inode *inode = io_end->inode; 3226 3227 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3228 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3229 /* Wake up anyone waiting on unwritten extent conversion */ 3230 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3231 wake_up_all(ext4_ioend_wq(inode)); 3232 } 3233 } 3234 3235 extern const struct iomap_ops ext4_iomap_ops; 3236 3237 #endif /* __KERNEL__ */ 3238 3239 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3240 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3241 3242 #endif /* _EXT4_H */ 3243