1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION) 44 #include <linux/fscrypt.h> 45 46 #include <linux/compiler.h> 47 48 /* 49 * The fourth extended filesystem constants/structures 50 */ 51 52 /* 53 * with AGGRESSIVE_CHECK allocator runs consistency checks over 54 * structures. these checks slow things down a lot 55 */ 56 #define AGGRESSIVE_CHECK__ 57 58 /* 59 * with DOUBLE_CHECK defined mballoc creates persistent in-core 60 * bitmaps, maintains and uses them to check for double allocations 61 */ 62 #define DOUBLE_CHECK__ 63 64 /* 65 * Define EXT4FS_DEBUG to produce debug messages 66 */ 67 #undef EXT4FS_DEBUG 68 69 /* 70 * Debug code 71 */ 72 #ifdef EXT4FS_DEBUG 73 #define ext4_debug(f, a...) \ 74 do { \ 75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 76 __FILE__, __LINE__, __func__); \ 77 printk(KERN_DEBUG f, ## a); \ 78 } while (0) 79 #else 80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 81 #endif 82 83 /* 84 * Turn on EXT_DEBUG to get lots of info about extents operations. 85 */ 86 #define EXT_DEBUG__ 87 #ifdef EXT_DEBUG 88 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 89 #else 90 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 91 #endif 92 93 /* data type for block offset of block group */ 94 typedef int ext4_grpblk_t; 95 96 /* data type for filesystem-wide blocks number */ 97 typedef unsigned long long ext4_fsblk_t; 98 99 /* data type for file logical block number */ 100 typedef __u32 ext4_lblk_t; 101 102 /* data type for block group number */ 103 typedef unsigned int ext4_group_t; 104 105 enum SHIFT_DIRECTION { 106 SHIFT_LEFT = 0, 107 SHIFT_RIGHT, 108 }; 109 110 /* 111 * Flags used in mballoc's allocation_context flags field. 112 * 113 * Also used to show what's going on for debugging purposes when the 114 * flag field is exported via the traceport interface 115 */ 116 117 /* prefer goal again. length */ 118 #define EXT4_MB_HINT_MERGE 0x0001 119 /* blocks already reserved */ 120 #define EXT4_MB_HINT_RESERVED 0x0002 121 /* metadata is being allocated */ 122 #define EXT4_MB_HINT_METADATA 0x0004 123 /* first blocks in the file */ 124 #define EXT4_MB_HINT_FIRST 0x0008 125 /* search for the best chunk */ 126 #define EXT4_MB_HINT_BEST 0x0010 127 /* data is being allocated */ 128 #define EXT4_MB_HINT_DATA 0x0020 129 /* don't preallocate (for tails) */ 130 #define EXT4_MB_HINT_NOPREALLOC 0x0040 131 /* allocate for locality group */ 132 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 133 /* allocate goal blocks or none */ 134 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 135 /* goal is meaningful */ 136 #define EXT4_MB_HINT_TRY_GOAL 0x0200 137 /* blocks already pre-reserved by delayed allocation */ 138 #define EXT4_MB_DELALLOC_RESERVED 0x0400 139 /* We are doing stream allocation */ 140 #define EXT4_MB_STREAM_ALLOC 0x0800 141 /* Use reserved root blocks if needed */ 142 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 143 /* Use blocks from reserved pool */ 144 #define EXT4_MB_USE_RESERVED 0x2000 145 146 struct ext4_allocation_request { 147 /* target inode for block we're allocating */ 148 struct inode *inode; 149 /* how many blocks we want to allocate */ 150 unsigned int len; 151 /* logical block in target inode */ 152 ext4_lblk_t logical; 153 /* the closest logical allocated block to the left */ 154 ext4_lblk_t lleft; 155 /* the closest logical allocated block to the right */ 156 ext4_lblk_t lright; 157 /* phys. target (a hint) */ 158 ext4_fsblk_t goal; 159 /* phys. block for the closest logical allocated block to the left */ 160 ext4_fsblk_t pleft; 161 /* phys. block for the closest logical allocated block to the right */ 162 ext4_fsblk_t pright; 163 /* flags. see above EXT4_MB_HINT_* */ 164 unsigned int flags; 165 }; 166 167 /* 168 * Logical to physical block mapping, used by ext4_map_blocks() 169 * 170 * This structure is used to pass requests into ext4_map_blocks() as 171 * well as to store the information returned by ext4_map_blocks(). It 172 * takes less room on the stack than a struct buffer_head. 173 */ 174 #define EXT4_MAP_NEW (1 << BH_New) 175 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 176 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 177 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 178 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 180 181 struct ext4_map_blocks { 182 ext4_fsblk_t m_pblk; 183 ext4_lblk_t m_lblk; 184 unsigned int m_len; 185 unsigned int m_flags; 186 }; 187 188 /* 189 * Flags for ext4_io_end->flags 190 */ 191 #define EXT4_IO_END_UNWRITTEN 0x0001 192 193 /* 194 * For converting unwritten extents on a work queue. 'handle' is used for 195 * buffered writeback. 196 */ 197 typedef struct ext4_io_end { 198 struct list_head list; /* per-file finished IO list */ 199 handle_t *handle; /* handle reserved for extent 200 * conversion */ 201 struct inode *inode; /* file being written to */ 202 struct bio *bio; /* Linked list of completed 203 * bios covering the extent */ 204 unsigned int flag; /* unwritten or not */ 205 atomic_t count; /* reference counter */ 206 loff_t offset; /* offset in the file */ 207 ssize_t size; /* size of the extent */ 208 } ext4_io_end_t; 209 210 struct ext4_io_submit { 211 struct writeback_control *io_wbc; 212 struct bio *io_bio; 213 ext4_io_end_t *io_end; 214 sector_t io_next_block; 215 }; 216 217 /* 218 * Special inodes numbers 219 */ 220 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 221 #define EXT4_ROOT_INO 2 /* Root inode */ 222 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 223 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 224 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 225 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 226 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 227 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 228 229 /* First non-reserved inode for old ext4 filesystems */ 230 #define EXT4_GOOD_OLD_FIRST_INO 11 231 232 /* 233 * Maximal count of links to a file 234 */ 235 #define EXT4_LINK_MAX 65000 236 237 /* 238 * Macro-instructions used to manage several block sizes 239 */ 240 #define EXT4_MIN_BLOCK_SIZE 1024 241 #define EXT4_MAX_BLOCK_SIZE 65536 242 #define EXT4_MIN_BLOCK_LOG_SIZE 10 243 #define EXT4_MAX_BLOCK_LOG_SIZE 16 244 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 245 #ifdef __KERNEL__ 246 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 247 #else 248 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 249 #endif 250 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 251 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 252 EXT4_SB(s)->s_cluster_bits) 253 #ifdef __KERNEL__ 254 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 255 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 256 #else 257 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 258 #endif 259 #ifdef __KERNEL__ 260 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 261 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 262 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 263 #else 264 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 265 EXT4_GOOD_OLD_INODE_SIZE : \ 266 (s)->s_inode_size) 267 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 268 EXT4_GOOD_OLD_FIRST_INO : \ 269 (s)->s_first_ino) 270 #endif 271 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 272 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 273 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 274 blkbits)) 275 276 /* Translate a block number to a cluster number */ 277 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 278 /* Translate a cluster number to a block number */ 279 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 280 /* Translate # of blks to # of clusters */ 281 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 282 (sbi)->s_cluster_bits) 283 /* Mask out the low bits to get the starting block of the cluster */ 284 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 285 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 286 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 287 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 288 /* Get the cluster offset */ 289 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 290 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 291 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 292 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 293 294 /* 295 * Structure of a blocks group descriptor 296 */ 297 struct ext4_group_desc 298 { 299 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 300 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 301 __le32 bg_inode_table_lo; /* Inodes table block */ 302 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 303 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 304 __le16 bg_used_dirs_count_lo; /* Directories count */ 305 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 306 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 307 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 308 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 309 __le16 bg_itable_unused_lo; /* Unused inodes count */ 310 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 311 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 312 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 313 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 314 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 315 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 316 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 317 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 318 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 319 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 320 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 321 __u32 bg_reserved; 322 }; 323 324 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 325 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 326 sizeof(__le16)) 327 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 328 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 329 sizeof(__le16)) 330 331 /* 332 * Structure of a flex block group info 333 */ 334 335 struct flex_groups { 336 atomic64_t free_clusters; 337 atomic_t free_inodes; 338 atomic_t used_dirs; 339 }; 340 341 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 342 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 343 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 344 345 /* 346 * Macro-instructions used to manage group descriptors 347 */ 348 #define EXT4_MIN_DESC_SIZE 32 349 #define EXT4_MIN_DESC_SIZE_64BIT 64 350 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 351 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 352 #ifdef __KERNEL__ 353 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 354 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 356 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 357 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 358 #else 359 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 361 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 362 #endif 363 364 /* 365 * Constants relative to the data blocks 366 */ 367 #define EXT4_NDIR_BLOCKS 12 368 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 369 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 370 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 371 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 372 373 /* 374 * Inode flags 375 */ 376 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 377 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 378 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 379 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 380 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 381 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 382 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 383 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 384 /* Reserved for compression usage... */ 385 #define EXT4_DIRTY_FL 0x00000100 386 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 387 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 388 /* nb: was previously EXT2_ECOMPR_FL */ 389 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 390 /* End compression flags --- maybe not all used */ 391 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 392 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 393 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 394 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 395 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 396 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 397 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 398 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 399 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 400 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 401 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 402 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 404 405 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 406 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 407 408 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 409 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 410 EXT4_IMMUTABLE_FL | \ 411 EXT4_APPEND_FL | \ 412 EXT4_NODUMP_FL | \ 413 EXT4_NOATIME_FL | \ 414 EXT4_PROJINHERIT_FL) 415 416 /* Flags that should be inherited by new inodes from their parent. */ 417 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 418 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 419 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 420 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 421 EXT4_PROJINHERIT_FL) 422 423 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 424 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 425 426 /* Flags that are appropriate for non-directories/regular files. */ 427 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 428 429 /* Mask out flags that are inappropriate for the given type of inode. */ 430 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 431 { 432 if (S_ISDIR(mode)) 433 return flags; 434 else if (S_ISREG(mode)) 435 return flags & EXT4_REG_FLMASK; 436 else 437 return flags & EXT4_OTHER_FLMASK; 438 } 439 440 /* 441 * Inode flags used for atomic set/get 442 */ 443 enum { 444 EXT4_INODE_SECRM = 0, /* Secure deletion */ 445 EXT4_INODE_UNRM = 1, /* Undelete */ 446 EXT4_INODE_COMPR = 2, /* Compress file */ 447 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 448 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 449 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 450 EXT4_INODE_NODUMP = 6, /* do not dump file */ 451 EXT4_INODE_NOATIME = 7, /* do not update atime */ 452 /* Reserved for compression usage... */ 453 EXT4_INODE_DIRTY = 8, 454 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 455 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 456 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 457 /* End compression flags --- maybe not all used */ 458 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 459 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 460 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 461 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 462 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 463 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 464 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 465 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 466 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 467 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 468 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 469 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 470 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 471 }; 472 473 /* 474 * Since it's pretty easy to mix up bit numbers and hex values, we use a 475 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 476 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 477 * any extra space in the compiled kernel image, otherwise, the build will fail. 478 * It's important that these values are the same, since we are using 479 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 480 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 481 * values found in ext2, ext3 and ext4 filesystems, and of course the values 482 * defined in e2fsprogs. 483 * 484 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 485 */ 486 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 487 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 488 489 static inline void ext4_check_flag_values(void) 490 { 491 CHECK_FLAG_VALUE(SECRM); 492 CHECK_FLAG_VALUE(UNRM); 493 CHECK_FLAG_VALUE(COMPR); 494 CHECK_FLAG_VALUE(SYNC); 495 CHECK_FLAG_VALUE(IMMUTABLE); 496 CHECK_FLAG_VALUE(APPEND); 497 CHECK_FLAG_VALUE(NODUMP); 498 CHECK_FLAG_VALUE(NOATIME); 499 CHECK_FLAG_VALUE(DIRTY); 500 CHECK_FLAG_VALUE(COMPRBLK); 501 CHECK_FLAG_VALUE(NOCOMPR); 502 CHECK_FLAG_VALUE(ENCRYPT); 503 CHECK_FLAG_VALUE(INDEX); 504 CHECK_FLAG_VALUE(IMAGIC); 505 CHECK_FLAG_VALUE(JOURNAL_DATA); 506 CHECK_FLAG_VALUE(NOTAIL); 507 CHECK_FLAG_VALUE(DIRSYNC); 508 CHECK_FLAG_VALUE(TOPDIR); 509 CHECK_FLAG_VALUE(HUGE_FILE); 510 CHECK_FLAG_VALUE(EXTENTS); 511 CHECK_FLAG_VALUE(EA_INODE); 512 CHECK_FLAG_VALUE(EOFBLOCKS); 513 CHECK_FLAG_VALUE(INLINE_DATA); 514 CHECK_FLAG_VALUE(PROJINHERIT); 515 CHECK_FLAG_VALUE(RESERVED); 516 } 517 518 /* Used to pass group descriptor data when online resize is done */ 519 struct ext4_new_group_input { 520 __u32 group; /* Group number for this data */ 521 __u64 block_bitmap; /* Absolute block number of block bitmap */ 522 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 523 __u64 inode_table; /* Absolute block number of inode table start */ 524 __u32 blocks_count; /* Total number of blocks in this group */ 525 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 526 __u16 unused; 527 }; 528 529 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 530 struct compat_ext4_new_group_input { 531 u32 group; 532 compat_u64 block_bitmap; 533 compat_u64 inode_bitmap; 534 compat_u64 inode_table; 535 u32 blocks_count; 536 u16 reserved_blocks; 537 u16 unused; 538 }; 539 #endif 540 541 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 542 struct ext4_new_group_data { 543 __u32 group; 544 __u64 block_bitmap; 545 __u64 inode_bitmap; 546 __u64 inode_table; 547 __u32 blocks_count; 548 __u16 reserved_blocks; 549 __u16 mdata_blocks; 550 __u32 free_clusters_count; 551 }; 552 553 /* Indexes used to index group tables in ext4_new_group_data */ 554 enum { 555 BLOCK_BITMAP = 0, /* block bitmap */ 556 INODE_BITMAP, /* inode bitmap */ 557 INODE_TABLE, /* inode tables */ 558 GROUP_TABLE_COUNT, 559 }; 560 561 /* 562 * Flags used by ext4_map_blocks() 563 */ 564 /* Allocate any needed blocks and/or convert an unwritten 565 extent to be an initialized ext4 */ 566 #define EXT4_GET_BLOCKS_CREATE 0x0001 567 /* Request the creation of an unwritten extent */ 568 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 569 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 570 EXT4_GET_BLOCKS_CREATE) 571 /* Caller is from the delayed allocation writeout path 572 * finally doing the actual allocation of delayed blocks */ 573 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 574 /* caller is from the direct IO path, request to creation of an 575 unwritten extents if not allocated, split the unwritten 576 extent if blocks has been preallocated already*/ 577 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 578 #define EXT4_GET_BLOCKS_CONVERT 0x0010 579 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 580 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 581 /* Convert extent to initialized after IO complete */ 582 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 583 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 584 /* Eventual metadata allocation (due to growing extent tree) 585 * should not fail, so try to use reserved blocks for that.*/ 586 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 587 /* Don't normalize allocation size (used for fallocate) */ 588 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 589 /* Request will not result in inode size update (user for fallocate) */ 590 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 591 /* Convert written extents to unwritten */ 592 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 593 /* Write zeros to newly created written extents */ 594 #define EXT4_GET_BLOCKS_ZERO 0x0200 595 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 596 EXT4_GET_BLOCKS_ZERO) 597 /* Caller will submit data before dropping transaction handle. This 598 * allows jbd2 to avoid submitting data before commit. */ 599 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 600 601 /* 602 * The bit position of these flags must not overlap with any of the 603 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 604 * read_extent_tree_block(), ext4_split_extent_at(), 605 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 606 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 607 * caching the extents when reading from the extent tree while a 608 * truncate or punch hole operation is in progress. 609 */ 610 #define EXT4_EX_NOCACHE 0x40000000 611 #define EXT4_EX_FORCE_CACHE 0x20000000 612 613 /* 614 * Flags used by ext4_free_blocks 615 */ 616 #define EXT4_FREE_BLOCKS_METADATA 0x0001 617 #define EXT4_FREE_BLOCKS_FORGET 0x0002 618 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 619 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 620 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 621 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 622 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 623 624 /* 625 * ioctl commands 626 */ 627 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 628 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 629 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 630 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 631 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 632 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 633 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 634 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 635 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 636 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 637 #define EXT4_IOC_MIGRATE _IO('f', 9) 638 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 639 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 640 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 641 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 642 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 643 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 644 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 645 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 646 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 647 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 648 649 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 650 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 651 652 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 653 654 /* 655 * Flags for going down operation 656 */ 657 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 658 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 659 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 660 661 662 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 663 /* 664 * ioctl commands in 32 bit emulation 665 */ 666 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 667 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 668 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 669 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 670 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 671 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 672 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 673 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 674 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 675 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 676 #endif 677 678 /* Max physical block we can address w/o extents */ 679 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 680 681 /* Max logical block we can support */ 682 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF 683 684 /* 685 * Structure of an inode on the disk 686 */ 687 struct ext4_inode { 688 __le16 i_mode; /* File mode */ 689 __le16 i_uid; /* Low 16 bits of Owner Uid */ 690 __le32 i_size_lo; /* Size in bytes */ 691 __le32 i_atime; /* Access time */ 692 __le32 i_ctime; /* Inode Change time */ 693 __le32 i_mtime; /* Modification time */ 694 __le32 i_dtime; /* Deletion Time */ 695 __le16 i_gid; /* Low 16 bits of Group Id */ 696 __le16 i_links_count; /* Links count */ 697 __le32 i_blocks_lo; /* Blocks count */ 698 __le32 i_flags; /* File flags */ 699 union { 700 struct { 701 __le32 l_i_version; 702 } linux1; 703 struct { 704 __u32 h_i_translator; 705 } hurd1; 706 struct { 707 __u32 m_i_reserved1; 708 } masix1; 709 } osd1; /* OS dependent 1 */ 710 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 711 __le32 i_generation; /* File version (for NFS) */ 712 __le32 i_file_acl_lo; /* File ACL */ 713 __le32 i_size_high; 714 __le32 i_obso_faddr; /* Obsoleted fragment address */ 715 union { 716 struct { 717 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 718 __le16 l_i_file_acl_high; 719 __le16 l_i_uid_high; /* these 2 fields */ 720 __le16 l_i_gid_high; /* were reserved2[0] */ 721 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 722 __le16 l_i_reserved; 723 } linux2; 724 struct { 725 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 726 __u16 h_i_mode_high; 727 __u16 h_i_uid_high; 728 __u16 h_i_gid_high; 729 __u32 h_i_author; 730 } hurd2; 731 struct { 732 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 733 __le16 m_i_file_acl_high; 734 __u32 m_i_reserved2[2]; 735 } masix2; 736 } osd2; /* OS dependent 2 */ 737 __le16 i_extra_isize; 738 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 739 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 740 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 741 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 742 __le32 i_crtime; /* File Creation time */ 743 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 744 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 745 __le32 i_projid; /* Project ID */ 746 }; 747 748 struct move_extent { 749 __u32 reserved; /* should be zero */ 750 __u32 donor_fd; /* donor file descriptor */ 751 __u64 orig_start; /* logical start offset in block for orig */ 752 __u64 donor_start; /* logical start offset in block for donor */ 753 __u64 len; /* block length to be moved */ 754 __u64 moved_len; /* moved block length */ 755 }; 756 757 #define EXT4_EPOCH_BITS 2 758 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 759 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 760 761 /* 762 * Extended fields will fit into an inode if the filesystem was formatted 763 * with large inodes (-I 256 or larger) and there are not currently any EAs 764 * consuming all of the available space. For new inodes we always reserve 765 * enough space for the kernel's known extended fields, but for inodes 766 * created with an old kernel this might not have been the case. None of 767 * the extended inode fields is critical for correct filesystem operation. 768 * This macro checks if a certain field fits in the inode. Note that 769 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 770 */ 771 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 772 ((offsetof(typeof(*ext4_inode), field) + \ 773 sizeof((ext4_inode)->field)) \ 774 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 775 (einode)->i_extra_isize)) \ 776 777 /* 778 * We use an encoding that preserves the times for extra epoch "00": 779 * 780 * extra msb of adjust for signed 781 * epoch 32-bit 32-bit tv_sec to 782 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 783 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 784 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 785 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 786 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 787 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 788 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 789 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 790 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 791 * 792 * Note that previous versions of the kernel on 64-bit systems would 793 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 794 * 1970. e2fsck will correct this, assuming that it is run on the 795 * affected filesystem before 2242. 796 */ 797 798 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 799 { 800 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 801 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 802 } 803 804 static inline void ext4_decode_extra_time(struct timespec64 *time, 805 __le32 extra) 806 { 807 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 808 809 #if 1 810 /* Handle legacy encoding of pre-1970 dates with epoch 811 * bits 1,1. (This backwards compatibility may be removed 812 * at the discretion of the ext4 developers.) 813 */ 814 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 815 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 816 extra_bits = 0; 817 time->tv_sec += extra_bits << 32; 818 #else 819 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 820 #endif 821 } 822 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 823 } 824 825 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 826 do { \ 827 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 828 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 829 (raw_inode)->xtime ## _extra = \ 830 ext4_encode_extra_time(&(inode)->xtime); \ 831 } \ 832 } while (0) 833 834 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 835 do { \ 836 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 837 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 838 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 839 (raw_inode)->xtime ## _extra = \ 840 ext4_encode_extra_time(&(einode)->xtime); \ 841 } while (0) 842 843 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 844 do { \ 845 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 846 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 847 ext4_decode_extra_time(&(inode)->xtime, \ 848 raw_inode->xtime ## _extra); \ 849 } \ 850 else \ 851 (inode)->xtime.tv_nsec = 0; \ 852 } while (0) 853 854 855 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 856 do { \ 857 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 858 (einode)->xtime.tv_sec = \ 859 (signed)le32_to_cpu((raw_inode)->xtime); \ 860 else \ 861 (einode)->xtime.tv_sec = 0; \ 862 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 863 ext4_decode_extra_time(&(einode)->xtime, \ 864 raw_inode->xtime ## _extra); \ 865 else \ 866 (einode)->xtime.tv_nsec = 0; \ 867 } while (0) 868 869 #define i_disk_version osd1.linux1.l_i_version 870 871 #if defined(__KERNEL__) || defined(__linux__) 872 #define i_reserved1 osd1.linux1.l_i_reserved1 873 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 874 #define i_blocks_high osd2.linux2.l_i_blocks_high 875 #define i_uid_low i_uid 876 #define i_gid_low i_gid 877 #define i_uid_high osd2.linux2.l_i_uid_high 878 #define i_gid_high osd2.linux2.l_i_gid_high 879 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 880 881 #elif defined(__GNU__) 882 883 #define i_translator osd1.hurd1.h_i_translator 884 #define i_uid_high osd2.hurd2.h_i_uid_high 885 #define i_gid_high osd2.hurd2.h_i_gid_high 886 #define i_author osd2.hurd2.h_i_author 887 888 #elif defined(__masix__) 889 890 #define i_reserved1 osd1.masix1.m_i_reserved1 891 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 892 #define i_reserved2 osd2.masix2.m_i_reserved2 893 894 #endif /* defined(__KERNEL__) || defined(__linux__) */ 895 896 #include "extents_status.h" 897 898 /* 899 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 900 * 901 * These are needed to avoid lockdep false positives when we need to 902 * allocate blocks to the quota inode during ext4_map_blocks(), while 903 * holding i_data_sem for a normal (non-quota) inode. Since we don't 904 * do quota tracking for the quota inode, this avoids deadlock (as 905 * well as infinite recursion, since it isn't turtles all the way 906 * down...) 907 * 908 * I_DATA_SEM_NORMAL - Used for most inodes 909 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 910 * where the second inode has larger inode number 911 * than the first 912 * I_DATA_SEM_QUOTA - Used for quota inodes only 913 */ 914 enum { 915 I_DATA_SEM_NORMAL = 0, 916 I_DATA_SEM_OTHER, 917 I_DATA_SEM_QUOTA, 918 }; 919 920 921 /* 922 * fourth extended file system inode data in memory 923 */ 924 struct ext4_inode_info { 925 __le32 i_data[15]; /* unconverted */ 926 __u32 i_dtime; 927 ext4_fsblk_t i_file_acl; 928 929 /* 930 * i_block_group is the number of the block group which contains 931 * this file's inode. Constant across the lifetime of the inode, 932 * it is used for making block allocation decisions - we try to 933 * place a file's data blocks near its inode block, and new inodes 934 * near to their parent directory's inode. 935 */ 936 ext4_group_t i_block_group; 937 ext4_lblk_t i_dir_start_lookup; 938 #if (BITS_PER_LONG < 64) 939 unsigned long i_state_flags; /* Dynamic state flags */ 940 #endif 941 unsigned long i_flags; 942 943 /* 944 * Extended attributes can be read independently of the main file 945 * data. Taking i_mutex even when reading would cause contention 946 * between readers of EAs and writers of regular file data, so 947 * instead we synchronize on xattr_sem when reading or changing 948 * EAs. 949 */ 950 struct rw_semaphore xattr_sem; 951 952 struct list_head i_orphan; /* unlinked but open inodes */ 953 954 /* 955 * i_disksize keeps track of what the inode size is ON DISK, not 956 * in memory. During truncate, i_size is set to the new size by 957 * the VFS prior to calling ext4_truncate(), but the filesystem won't 958 * set i_disksize to 0 until the truncate is actually under way. 959 * 960 * The intent is that i_disksize always represents the blocks which 961 * are used by this file. This allows recovery to restart truncate 962 * on orphans if we crash during truncate. We actually write i_disksize 963 * into the on-disk inode when writing inodes out, instead of i_size. 964 * 965 * The only time when i_disksize and i_size may be different is when 966 * a truncate is in progress. The only things which change i_disksize 967 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 968 */ 969 loff_t i_disksize; 970 971 /* 972 * i_data_sem is for serialising ext4_truncate() against 973 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 974 * data tree are chopped off during truncate. We can't do that in 975 * ext4 because whenever we perform intermediate commits during 976 * truncate, the inode and all the metadata blocks *must* be in a 977 * consistent state which allows truncation of the orphans to restart 978 * during recovery. Hence we must fix the get_block-vs-truncate race 979 * by other means, so we have i_data_sem. 980 */ 981 struct rw_semaphore i_data_sem; 982 /* 983 * i_mmap_sem is for serializing page faults with truncate / punch hole 984 * operations. We have to make sure that new page cannot be faulted in 985 * a section of the inode that is being punched. We cannot easily use 986 * i_data_sem for this since we need protection for the whole punch 987 * operation and i_data_sem ranks below transaction start so we have 988 * to occasionally drop it. 989 */ 990 struct rw_semaphore i_mmap_sem; 991 struct inode vfs_inode; 992 struct jbd2_inode *jinode; 993 994 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 995 996 /* 997 * File creation time. Its function is same as that of 998 * struct timespec64 i_{a,c,m}time in the generic inode. 999 */ 1000 struct timespec64 i_crtime; 1001 1002 /* mballoc */ 1003 struct list_head i_prealloc_list; 1004 spinlock_t i_prealloc_lock; 1005 1006 /* extents status tree */ 1007 struct ext4_es_tree i_es_tree; 1008 rwlock_t i_es_lock; 1009 struct list_head i_es_list; 1010 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1011 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1012 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1013 extents to shrink. Protected by 1014 i_es_lock */ 1015 1016 /* ialloc */ 1017 ext4_group_t i_last_alloc_group; 1018 1019 /* allocation reservation info for delalloc */ 1020 /* In case of bigalloc, this refer to clusters rather than blocks */ 1021 unsigned int i_reserved_data_blocks; 1022 ext4_lblk_t i_da_metadata_calc_last_lblock; 1023 int i_da_metadata_calc_len; 1024 1025 /* pending cluster reservations for bigalloc file systems */ 1026 struct ext4_pending_tree i_pending_tree; 1027 1028 /* on-disk additional length */ 1029 __u16 i_extra_isize; 1030 1031 /* Indicate the inline data space. */ 1032 u16 i_inline_off; 1033 u16 i_inline_size; 1034 1035 #ifdef CONFIG_QUOTA 1036 /* quota space reservation, managed internally by quota code */ 1037 qsize_t i_reserved_quota; 1038 #endif 1039 1040 /* Lock protecting lists below */ 1041 spinlock_t i_completed_io_lock; 1042 /* 1043 * Completed IOs that need unwritten extents handling and have 1044 * transaction reserved 1045 */ 1046 struct list_head i_rsv_conversion_list; 1047 struct work_struct i_rsv_conversion_work; 1048 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1049 1050 spinlock_t i_block_reservation_lock; 1051 1052 /* 1053 * Transactions that contain inode's metadata needed to complete 1054 * fsync and fdatasync, respectively. 1055 */ 1056 tid_t i_sync_tid; 1057 tid_t i_datasync_tid; 1058 1059 #ifdef CONFIG_QUOTA 1060 struct dquot *i_dquot[MAXQUOTAS]; 1061 #endif 1062 1063 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1064 __u32 i_csum_seed; 1065 1066 kprojid_t i_projid; 1067 }; 1068 1069 /* 1070 * File system states 1071 */ 1072 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1073 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1074 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1075 1076 /* 1077 * Misc. filesystem flags 1078 */ 1079 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1080 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1081 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1082 1083 /* 1084 * Mount flags set via mount options or defaults 1085 */ 1086 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1087 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1088 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1089 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1090 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1091 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1092 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1093 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1094 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1095 #ifdef CONFIG_FS_DAX 1096 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1097 #else 1098 #define EXT4_MOUNT_DAX 0 1099 #endif 1100 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1101 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1102 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1103 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1104 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1105 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1106 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1107 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1108 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1109 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1110 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1111 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1112 * enable enforcement for hidden 1113 * quota files */ 1114 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1115 * enforcement for hidden quota 1116 * files */ 1117 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1118 * enforcement */ 1119 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1120 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1121 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1122 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1123 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1124 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1125 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1126 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1127 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1128 1129 /* 1130 * Mount flags set either automatically (could not be set by mount option) 1131 * based on per file system feature or property or in special cases such as 1132 * distinguishing between explicit mount option definition and default. 1133 */ 1134 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1135 specified delalloc */ 1136 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1137 size of blocksize * 8 1138 blocks */ 1139 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1140 file systems */ 1141 1142 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1143 specified journal checksum */ 1144 1145 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1146 ~EXT4_MOUNT_##opt 1147 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1148 EXT4_MOUNT_##opt 1149 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1150 EXT4_MOUNT_##opt) 1151 1152 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1153 ~EXT4_MOUNT2_##opt 1154 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1155 EXT4_MOUNT2_##opt 1156 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1157 EXT4_MOUNT2_##opt) 1158 1159 #define ext4_test_and_set_bit __test_and_set_bit_le 1160 #define ext4_set_bit __set_bit_le 1161 #define ext4_set_bit_atomic ext2_set_bit_atomic 1162 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1163 #define ext4_clear_bit __clear_bit_le 1164 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1165 #define ext4_test_bit test_bit_le 1166 #define ext4_find_next_zero_bit find_next_zero_bit_le 1167 #define ext4_find_next_bit find_next_bit_le 1168 1169 extern void ext4_set_bits(void *bm, int cur, int len); 1170 1171 /* 1172 * Maximal mount counts between two filesystem checks 1173 */ 1174 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1175 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1176 1177 /* 1178 * Behaviour when detecting errors 1179 */ 1180 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1181 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1182 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1183 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1184 1185 /* Metadata checksum algorithm codes */ 1186 #define EXT4_CRC32C_CHKSUM 1 1187 1188 /* 1189 * Structure of the super block 1190 */ 1191 struct ext4_super_block { 1192 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1193 __le32 s_blocks_count_lo; /* Blocks count */ 1194 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1195 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1196 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1197 __le32 s_first_data_block; /* First Data Block */ 1198 __le32 s_log_block_size; /* Block size */ 1199 __le32 s_log_cluster_size; /* Allocation cluster size */ 1200 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1201 __le32 s_clusters_per_group; /* # Clusters per group */ 1202 __le32 s_inodes_per_group; /* # Inodes per group */ 1203 __le32 s_mtime; /* Mount time */ 1204 /*30*/ __le32 s_wtime; /* Write time */ 1205 __le16 s_mnt_count; /* Mount count */ 1206 __le16 s_max_mnt_count; /* Maximal mount count */ 1207 __le16 s_magic; /* Magic signature */ 1208 __le16 s_state; /* File system state */ 1209 __le16 s_errors; /* Behaviour when detecting errors */ 1210 __le16 s_minor_rev_level; /* minor revision level */ 1211 /*40*/ __le32 s_lastcheck; /* time of last check */ 1212 __le32 s_checkinterval; /* max. time between checks */ 1213 __le32 s_creator_os; /* OS */ 1214 __le32 s_rev_level; /* Revision level */ 1215 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1216 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1217 /* 1218 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1219 * 1220 * Note: the difference between the compatible feature set and 1221 * the incompatible feature set is that if there is a bit set 1222 * in the incompatible feature set that the kernel doesn't 1223 * know about, it should refuse to mount the filesystem. 1224 * 1225 * e2fsck's requirements are more strict; if it doesn't know 1226 * about a feature in either the compatible or incompatible 1227 * feature set, it must abort and not try to meddle with 1228 * things it doesn't understand... 1229 */ 1230 __le32 s_first_ino; /* First non-reserved inode */ 1231 __le16 s_inode_size; /* size of inode structure */ 1232 __le16 s_block_group_nr; /* block group # of this superblock */ 1233 __le32 s_feature_compat; /* compatible feature set */ 1234 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1235 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1236 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1237 /*78*/ char s_volume_name[16]; /* volume name */ 1238 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1239 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1240 /* 1241 * Performance hints. Directory preallocation should only 1242 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1243 */ 1244 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1245 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1246 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1247 /* 1248 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1249 */ 1250 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1251 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1252 __le32 s_journal_dev; /* device number of journal file */ 1253 __le32 s_last_orphan; /* start of list of inodes to delete */ 1254 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1255 __u8 s_def_hash_version; /* Default hash version to use */ 1256 __u8 s_jnl_backup_type; 1257 __le16 s_desc_size; /* size of group descriptor */ 1258 /*100*/ __le32 s_default_mount_opts; 1259 __le32 s_first_meta_bg; /* First metablock block group */ 1260 __le32 s_mkfs_time; /* When the filesystem was created */ 1261 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1262 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1263 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1264 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1265 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1266 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1267 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1268 __le32 s_flags; /* Miscellaneous flags */ 1269 __le16 s_raid_stride; /* RAID stride */ 1270 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1271 __le64 s_mmp_block; /* Block for multi-mount protection */ 1272 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1273 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1274 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1275 __u8 s_encryption_level; /* versioning level for encryption */ 1276 __u8 s_reserved_pad; /* Padding to next 32bits */ 1277 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1278 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1279 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1280 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1281 snapshot's future use */ 1282 __le32 s_snapshot_list; /* inode number of the head of the 1283 on-disk snapshot list */ 1284 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1285 __le32 s_error_count; /* number of fs errors */ 1286 __le32 s_first_error_time; /* first time an error happened */ 1287 __le32 s_first_error_ino; /* inode involved in first error */ 1288 __le64 s_first_error_block; /* block involved of first error */ 1289 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1290 __le32 s_first_error_line; /* line number where error happened */ 1291 __le32 s_last_error_time; /* most recent time of an error */ 1292 __le32 s_last_error_ino; /* inode involved in last error */ 1293 __le32 s_last_error_line; /* line number where error happened */ 1294 __le64 s_last_error_block; /* block involved of last error */ 1295 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1296 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1297 __u8 s_mount_opts[64]; 1298 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1299 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1300 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1301 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1302 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1303 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1304 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1305 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1306 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1307 __u8 s_wtime_hi; 1308 __u8 s_mtime_hi; 1309 __u8 s_mkfs_time_hi; 1310 __u8 s_lastcheck_hi; 1311 __u8 s_first_error_time_hi; 1312 __u8 s_last_error_time_hi; 1313 __u8 s_pad[2]; 1314 __le32 s_reserved[96]; /* Padding to the end of the block */ 1315 __le32 s_checksum; /* crc32c(superblock) */ 1316 }; 1317 1318 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1319 1320 #ifdef __KERNEL__ 1321 1322 /* 1323 * run-time mount flags 1324 */ 1325 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1326 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1327 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1328 1329 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1330 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1331 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1332 #else 1333 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1334 #endif 1335 1336 /* Number of quota types we support */ 1337 #define EXT4_MAXQUOTAS 3 1338 1339 /* 1340 * fourth extended-fs super-block data in memory 1341 */ 1342 struct ext4_sb_info { 1343 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1344 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1345 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1346 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1347 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1348 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1349 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1350 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1351 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1352 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1353 unsigned long s_overhead; /* # of fs overhead clusters */ 1354 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1355 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1356 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1357 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1358 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1359 struct buffer_head **s_group_desc; 1360 unsigned int s_mount_opt; 1361 unsigned int s_mount_opt2; 1362 unsigned int s_mount_flags; 1363 unsigned int s_def_mount_opt; 1364 ext4_fsblk_t s_sb_block; 1365 atomic64_t s_resv_clusters; 1366 kuid_t s_resuid; 1367 kgid_t s_resgid; 1368 unsigned short s_mount_state; 1369 unsigned short s_pad; 1370 int s_addr_per_block_bits; 1371 int s_desc_per_block_bits; 1372 int s_inode_size; 1373 int s_first_ino; 1374 unsigned int s_inode_readahead_blks; 1375 unsigned int s_inode_goal; 1376 u32 s_hash_seed[4]; 1377 int s_def_hash_version; 1378 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1379 struct percpu_counter s_freeclusters_counter; 1380 struct percpu_counter s_freeinodes_counter; 1381 struct percpu_counter s_dirs_counter; 1382 struct percpu_counter s_dirtyclusters_counter; 1383 struct blockgroup_lock *s_blockgroup_lock; 1384 struct proc_dir_entry *s_proc; 1385 struct kobject s_kobj; 1386 struct completion s_kobj_unregister; 1387 struct super_block *s_sb; 1388 1389 /* Journaling */ 1390 struct journal_s *s_journal; 1391 struct list_head s_orphan; 1392 struct mutex s_orphan_lock; 1393 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1394 unsigned long s_commit_interval; 1395 u32 s_max_batch_time; 1396 u32 s_min_batch_time; 1397 struct block_device *journal_bdev; 1398 #ifdef CONFIG_QUOTA 1399 /* Names of quota files with journalled quota */ 1400 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1401 int s_jquota_fmt; /* Format of quota to use */ 1402 #endif 1403 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1404 struct rb_root system_blks; 1405 1406 #ifdef EXTENTS_STATS 1407 /* ext4 extents stats */ 1408 unsigned long s_ext_min; 1409 unsigned long s_ext_max; 1410 unsigned long s_depth_max; 1411 spinlock_t s_ext_stats_lock; 1412 unsigned long s_ext_blocks; 1413 unsigned long s_ext_extents; 1414 #endif 1415 1416 /* for buddy allocator */ 1417 struct ext4_group_info ***s_group_info; 1418 struct inode *s_buddy_cache; 1419 spinlock_t s_md_lock; 1420 unsigned short *s_mb_offsets; 1421 unsigned int *s_mb_maxs; 1422 unsigned int s_group_info_size; 1423 unsigned int s_mb_free_pending; 1424 struct list_head s_freed_data_list; /* List of blocks to be freed 1425 after commit completed */ 1426 1427 /* tunables */ 1428 unsigned long s_stripe; 1429 unsigned int s_mb_stream_request; 1430 unsigned int s_mb_max_to_scan; 1431 unsigned int s_mb_min_to_scan; 1432 unsigned int s_mb_stats; 1433 unsigned int s_mb_order2_reqs; 1434 unsigned int s_mb_group_prealloc; 1435 unsigned int s_max_dir_size_kb; 1436 /* where last allocation was done - for stream allocation */ 1437 unsigned long s_mb_last_group; 1438 unsigned long s_mb_last_start; 1439 1440 /* stats for buddy allocator */ 1441 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1442 atomic_t s_bal_success; /* we found long enough chunks */ 1443 atomic_t s_bal_allocated; /* in blocks */ 1444 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1445 atomic_t s_bal_goals; /* goal hits */ 1446 atomic_t s_bal_breaks; /* too long searches */ 1447 atomic_t s_bal_2orders; /* 2^order hits */ 1448 spinlock_t s_bal_lock; 1449 unsigned long s_mb_buddies_generated; 1450 unsigned long long s_mb_generation_time; 1451 atomic_t s_mb_lost_chunks; 1452 atomic_t s_mb_preallocated; 1453 atomic_t s_mb_discarded; 1454 atomic_t s_lock_busy; 1455 1456 /* locality groups */ 1457 struct ext4_locality_group __percpu *s_locality_groups; 1458 1459 /* for write statistics */ 1460 unsigned long s_sectors_written_start; 1461 u64 s_kbytes_written; 1462 1463 /* the size of zero-out chunk */ 1464 unsigned int s_extent_max_zeroout_kb; 1465 1466 unsigned int s_log_groups_per_flex; 1467 struct flex_groups *s_flex_groups; 1468 ext4_group_t s_flex_groups_allocated; 1469 1470 /* workqueue for reserved extent conversions (buffered io) */ 1471 struct workqueue_struct *rsv_conversion_wq; 1472 1473 /* timer for periodic error stats printing */ 1474 struct timer_list s_err_report; 1475 1476 /* Lazy inode table initialization info */ 1477 struct ext4_li_request *s_li_request; 1478 /* Wait multiplier for lazy initialization thread */ 1479 unsigned int s_li_wait_mult; 1480 1481 /* Kernel thread for multiple mount protection */ 1482 struct task_struct *s_mmp_tsk; 1483 1484 /* record the last minlen when FITRIM is called. */ 1485 atomic_t s_last_trim_minblks; 1486 1487 /* Reference to checksum algorithm driver via cryptoapi */ 1488 struct crypto_shash *s_chksum_driver; 1489 1490 /* Precomputed FS UUID checksum for seeding other checksums */ 1491 __u32 s_csum_seed; 1492 1493 /* Reclaim extents from extent status tree */ 1494 struct shrinker s_es_shrinker; 1495 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1496 long s_es_nr_inode; 1497 struct ext4_es_stats s_es_stats; 1498 struct mb_cache *s_ea_block_cache; 1499 struct mb_cache *s_ea_inode_cache; 1500 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1501 1502 /* Ratelimit ext4 messages. */ 1503 struct ratelimit_state s_err_ratelimit_state; 1504 struct ratelimit_state s_warning_ratelimit_state; 1505 struct ratelimit_state s_msg_ratelimit_state; 1506 1507 /* Barrier between changing inodes' journal flags and writepages ops. */ 1508 struct percpu_rw_semaphore s_journal_flag_rwsem; 1509 struct dax_device *s_daxdev; 1510 }; 1511 1512 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1513 { 1514 return sb->s_fs_info; 1515 } 1516 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1517 { 1518 return container_of(inode, struct ext4_inode_info, vfs_inode); 1519 } 1520 1521 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1522 { 1523 return ino == EXT4_ROOT_INO || 1524 (ino >= EXT4_FIRST_INO(sb) && 1525 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1526 } 1527 1528 /* 1529 * Inode dynamic state flags 1530 */ 1531 enum { 1532 EXT4_STATE_JDATA, /* journaled data exists */ 1533 EXT4_STATE_NEW, /* inode is newly created */ 1534 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1535 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1536 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1537 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1538 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1539 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1540 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1541 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1542 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1543 }; 1544 1545 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1546 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1547 { \ 1548 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1549 } \ 1550 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1551 { \ 1552 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1553 } \ 1554 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1555 { \ 1556 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1557 } 1558 1559 /* Add these declarations here only so that these functions can be 1560 * found by name. Otherwise, they are very hard to locate. */ 1561 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1562 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1563 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1564 EXT4_INODE_BIT_FNS(flag, flags, 0) 1565 1566 /* Add these declarations here only so that these functions can be 1567 * found by name. Otherwise, they are very hard to locate. */ 1568 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1569 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1570 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1571 #if (BITS_PER_LONG < 64) 1572 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1573 1574 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1575 { 1576 (ei)->i_state_flags = 0; 1577 } 1578 #else 1579 EXT4_INODE_BIT_FNS(state, flags, 32) 1580 1581 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1582 { 1583 /* We depend on the fact that callers will set i_flags */ 1584 } 1585 #endif 1586 #else 1587 /* Assume that user mode programs are passing in an ext4fs superblock, not 1588 * a kernel struct super_block. This will allow us to call the feature-test 1589 * macros from user land. */ 1590 #define EXT4_SB(sb) (sb) 1591 #endif 1592 1593 /* 1594 * Returns true if the inode is inode is encrypted 1595 */ 1596 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1597 1598 /* 1599 * Codes for operating systems 1600 */ 1601 #define EXT4_OS_LINUX 0 1602 #define EXT4_OS_HURD 1 1603 #define EXT4_OS_MASIX 2 1604 #define EXT4_OS_FREEBSD 3 1605 #define EXT4_OS_LITES 4 1606 1607 /* 1608 * Revision levels 1609 */ 1610 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1611 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1612 1613 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1614 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1615 1616 #define EXT4_GOOD_OLD_INODE_SIZE 128 1617 1618 /* 1619 * Feature set definitions 1620 */ 1621 1622 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1623 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1624 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1625 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1626 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1627 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1628 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1629 1630 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1631 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1632 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1633 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1634 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1635 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1636 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1637 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1638 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1639 /* 1640 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1641 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1642 * all other data structures' checksums. However, the METADATA_CSUM and 1643 * GDT_CSUM bits are mutually exclusive. 1644 */ 1645 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1646 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1647 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1648 1649 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1650 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1651 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1652 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1653 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1654 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1655 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1656 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1657 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1658 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1659 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1660 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1661 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1662 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1663 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1664 1665 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1666 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1667 { \ 1668 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1669 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1670 } \ 1671 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1672 { \ 1673 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1674 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1675 } \ 1676 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1677 { \ 1678 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1679 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1680 } 1681 1682 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1683 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1684 { \ 1685 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1686 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1687 } \ 1688 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1689 { \ 1690 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1691 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1692 } \ 1693 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1694 { \ 1695 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1696 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1697 } 1698 1699 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1700 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1701 { \ 1702 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1703 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1704 } \ 1705 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1706 { \ 1707 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1708 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1709 } \ 1710 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1711 { \ 1712 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1713 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1714 } 1715 1716 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1717 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1718 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1719 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1720 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1721 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1722 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1723 1724 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1725 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1726 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1727 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1728 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1729 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1730 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1731 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1732 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1733 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1734 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1735 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1736 1737 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1738 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1739 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1740 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1741 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1742 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1743 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1744 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1745 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1746 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1747 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1748 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1749 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1750 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1751 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1752 1753 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1754 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1755 EXT4_FEATURE_INCOMPAT_META_BG) 1756 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1757 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1758 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1759 1760 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1761 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1762 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1763 EXT4_FEATURE_INCOMPAT_META_BG) 1764 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1765 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1766 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1767 1768 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1769 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1770 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1771 EXT4_FEATURE_INCOMPAT_META_BG| \ 1772 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1773 EXT4_FEATURE_INCOMPAT_64BIT| \ 1774 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1775 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1776 EXT4_FEATURE_INCOMPAT_MMP | \ 1777 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1778 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1779 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1780 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1781 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1782 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1783 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1784 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1785 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1786 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1787 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1788 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1789 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1790 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1791 EXT4_FEATURE_RO_COMPAT_PROJECT) 1792 1793 #define EXTN_FEATURE_FUNCS(ver) \ 1794 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1795 { \ 1796 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1797 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1798 } \ 1799 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1800 { \ 1801 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1802 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1803 } \ 1804 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1805 { \ 1806 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1807 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1808 } 1809 1810 EXTN_FEATURE_FUNCS(2) 1811 EXTN_FEATURE_FUNCS(3) 1812 EXTN_FEATURE_FUNCS(4) 1813 1814 static inline bool ext4_has_compat_features(struct super_block *sb) 1815 { 1816 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1817 } 1818 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1819 { 1820 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1821 } 1822 static inline bool ext4_has_incompat_features(struct super_block *sb) 1823 { 1824 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1825 } 1826 1827 /* 1828 * Superblock flags 1829 */ 1830 #define EXT4_FLAGS_RESIZING 0 1831 #define EXT4_FLAGS_SHUTDOWN 1 1832 1833 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1834 { 1835 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1836 } 1837 1838 1839 /* 1840 * Default values for user and/or group using reserved blocks 1841 */ 1842 #define EXT4_DEF_RESUID 0 1843 #define EXT4_DEF_RESGID 0 1844 1845 /* 1846 * Default project ID 1847 */ 1848 #define EXT4_DEF_PROJID 0 1849 1850 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1851 1852 /* 1853 * Default mount options 1854 */ 1855 #define EXT4_DEFM_DEBUG 0x0001 1856 #define EXT4_DEFM_BSDGROUPS 0x0002 1857 #define EXT4_DEFM_XATTR_USER 0x0004 1858 #define EXT4_DEFM_ACL 0x0008 1859 #define EXT4_DEFM_UID16 0x0010 1860 #define EXT4_DEFM_JMODE 0x0060 1861 #define EXT4_DEFM_JMODE_DATA 0x0020 1862 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1863 #define EXT4_DEFM_JMODE_WBACK 0x0060 1864 #define EXT4_DEFM_NOBARRIER 0x0100 1865 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1866 #define EXT4_DEFM_DISCARD 0x0400 1867 #define EXT4_DEFM_NODELALLOC 0x0800 1868 1869 /* 1870 * Default journal batch times 1871 */ 1872 #define EXT4_DEF_MIN_BATCH_TIME 0 1873 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1874 1875 /* 1876 * Minimum number of groups in a flexgroup before we separate out 1877 * directories into the first block group of a flexgroup 1878 */ 1879 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1880 1881 /* 1882 * Structure of a directory entry 1883 */ 1884 #define EXT4_NAME_LEN 255 1885 1886 struct ext4_dir_entry { 1887 __le32 inode; /* Inode number */ 1888 __le16 rec_len; /* Directory entry length */ 1889 __le16 name_len; /* Name length */ 1890 char name[EXT4_NAME_LEN]; /* File name */ 1891 }; 1892 1893 /* 1894 * The new version of the directory entry. Since EXT4 structures are 1895 * stored in intel byte order, and the name_len field could never be 1896 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1897 * file_type field. 1898 */ 1899 struct ext4_dir_entry_2 { 1900 __le32 inode; /* Inode number */ 1901 __le16 rec_len; /* Directory entry length */ 1902 __u8 name_len; /* Name length */ 1903 __u8 file_type; 1904 char name[EXT4_NAME_LEN]; /* File name */ 1905 }; 1906 1907 /* 1908 * This is a bogus directory entry at the end of each leaf block that 1909 * records checksums. 1910 */ 1911 struct ext4_dir_entry_tail { 1912 __le32 det_reserved_zero1; /* Pretend to be unused */ 1913 __le16 det_rec_len; /* 12 */ 1914 __u8 det_reserved_zero2; /* Zero name length */ 1915 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1916 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1917 }; 1918 1919 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1920 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1921 ((blocksize) - \ 1922 sizeof(struct ext4_dir_entry_tail)))) 1923 1924 /* 1925 * Ext4 directory file types. Only the low 3 bits are used. The 1926 * other bits are reserved for now. 1927 */ 1928 #define EXT4_FT_UNKNOWN 0 1929 #define EXT4_FT_REG_FILE 1 1930 #define EXT4_FT_DIR 2 1931 #define EXT4_FT_CHRDEV 3 1932 #define EXT4_FT_BLKDEV 4 1933 #define EXT4_FT_FIFO 5 1934 #define EXT4_FT_SOCK 6 1935 #define EXT4_FT_SYMLINK 7 1936 1937 #define EXT4_FT_MAX 8 1938 1939 #define EXT4_FT_DIR_CSUM 0xDE 1940 1941 /* 1942 * EXT4_DIR_PAD defines the directory entries boundaries 1943 * 1944 * NOTE: It must be a multiple of 4 1945 */ 1946 #define EXT4_DIR_PAD 4 1947 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1948 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1949 ~EXT4_DIR_ROUND) 1950 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1951 1952 /* 1953 * If we ever get support for fs block sizes > page_size, we'll need 1954 * to remove the #if statements in the next two functions... 1955 */ 1956 static inline unsigned int 1957 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1958 { 1959 unsigned len = le16_to_cpu(dlen); 1960 1961 #if (PAGE_SIZE >= 65536) 1962 if (len == EXT4_MAX_REC_LEN || len == 0) 1963 return blocksize; 1964 return (len & 65532) | ((len & 3) << 16); 1965 #else 1966 return len; 1967 #endif 1968 } 1969 1970 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1971 { 1972 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1973 BUG(); 1974 #if (PAGE_SIZE >= 65536) 1975 if (len < 65536) 1976 return cpu_to_le16(len); 1977 if (len == blocksize) { 1978 if (blocksize == 65536) 1979 return cpu_to_le16(EXT4_MAX_REC_LEN); 1980 else 1981 return cpu_to_le16(0); 1982 } 1983 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1984 #else 1985 return cpu_to_le16(len); 1986 #endif 1987 } 1988 1989 /* 1990 * Hash Tree Directory indexing 1991 * (c) Daniel Phillips, 2001 1992 */ 1993 1994 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1995 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1996 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 1997 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 1998 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1999 2000 /* Legal values for the dx_root hash_version field: */ 2001 2002 #define DX_HASH_LEGACY 0 2003 #define DX_HASH_HALF_MD4 1 2004 #define DX_HASH_TEA 2 2005 #define DX_HASH_LEGACY_UNSIGNED 3 2006 #define DX_HASH_HALF_MD4_UNSIGNED 4 2007 #define DX_HASH_TEA_UNSIGNED 5 2008 2009 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2010 const void *address, unsigned int length) 2011 { 2012 struct { 2013 struct shash_desc shash; 2014 char ctx[4]; 2015 } desc; 2016 2017 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2018 2019 desc.shash.tfm = sbi->s_chksum_driver; 2020 desc.shash.flags = 0; 2021 *(u32 *)desc.ctx = crc; 2022 2023 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2024 2025 return *(u32 *)desc.ctx; 2026 } 2027 2028 #ifdef __KERNEL__ 2029 2030 /* hash info structure used by the directory hash */ 2031 struct dx_hash_info 2032 { 2033 u32 hash; 2034 u32 minor_hash; 2035 int hash_version; 2036 u32 *seed; 2037 }; 2038 2039 2040 /* 32 and 64 bit signed EOF for dx directories */ 2041 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2042 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2043 2044 2045 /* 2046 * Control parameters used by ext4_htree_next_block 2047 */ 2048 #define HASH_NB_ALWAYS 1 2049 2050 struct ext4_filename { 2051 const struct qstr *usr_fname; 2052 struct fscrypt_str disk_name; 2053 struct dx_hash_info hinfo; 2054 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2055 struct fscrypt_str crypto_buf; 2056 #endif 2057 }; 2058 2059 #define fname_name(p) ((p)->disk_name.name) 2060 #define fname_len(p) ((p)->disk_name.len) 2061 2062 /* 2063 * Describe an inode's exact location on disk and in memory 2064 */ 2065 struct ext4_iloc 2066 { 2067 struct buffer_head *bh; 2068 unsigned long offset; 2069 ext4_group_t block_group; 2070 }; 2071 2072 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2073 { 2074 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2075 } 2076 2077 static inline bool ext4_is_quota_file(struct inode *inode) 2078 { 2079 return IS_NOQUOTA(inode) && 2080 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2081 } 2082 2083 /* 2084 * This structure is stuffed into the struct file's private_data field 2085 * for directories. It is where we put information so that we can do 2086 * readdir operations in hash tree order. 2087 */ 2088 struct dir_private_info { 2089 struct rb_root root; 2090 struct rb_node *curr_node; 2091 struct fname *extra_fname; 2092 loff_t last_pos; 2093 __u32 curr_hash; 2094 __u32 curr_minor_hash; 2095 __u32 next_hash; 2096 }; 2097 2098 /* calculate the first block number of the group */ 2099 static inline ext4_fsblk_t 2100 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2101 { 2102 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2103 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2104 } 2105 2106 /* 2107 * Special error return code only used by dx_probe() and its callers. 2108 */ 2109 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2110 2111 /* htree levels for ext4 */ 2112 #define EXT4_HTREE_LEVEL_COMPAT 2 2113 #define EXT4_HTREE_LEVEL 3 2114 2115 static inline int ext4_dir_htree_level(struct super_block *sb) 2116 { 2117 return ext4_has_feature_largedir(sb) ? 2118 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2119 } 2120 2121 /* 2122 * Timeout and state flag for lazy initialization inode thread. 2123 */ 2124 #define EXT4_DEF_LI_WAIT_MULT 10 2125 #define EXT4_DEF_LI_MAX_START_DELAY 5 2126 #define EXT4_LAZYINIT_QUIT 0x0001 2127 #define EXT4_LAZYINIT_RUNNING 0x0002 2128 2129 /* 2130 * Lazy inode table initialization info 2131 */ 2132 struct ext4_lazy_init { 2133 unsigned long li_state; 2134 struct list_head li_request_list; 2135 struct mutex li_list_mtx; 2136 }; 2137 2138 struct ext4_li_request { 2139 struct super_block *lr_super; 2140 struct ext4_sb_info *lr_sbi; 2141 ext4_group_t lr_next_group; 2142 struct list_head lr_request; 2143 unsigned long lr_next_sched; 2144 unsigned long lr_timeout; 2145 }; 2146 2147 struct ext4_features { 2148 struct kobject f_kobj; 2149 struct completion f_kobj_unregister; 2150 }; 2151 2152 /* 2153 * This structure will be used for multiple mount protection. It will be 2154 * written into the block number saved in the s_mmp_block field in the 2155 * superblock. Programs that check MMP should assume that if 2156 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2157 * to use the filesystem, regardless of how old the timestamp is. 2158 */ 2159 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2160 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2161 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2162 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2163 2164 struct mmp_struct { 2165 __le32 mmp_magic; /* Magic number for MMP */ 2166 __le32 mmp_seq; /* Sequence no. updated periodically */ 2167 2168 /* 2169 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2170 * purposes and do not affect the correctness of the algorithm 2171 */ 2172 __le64 mmp_time; /* Time last updated */ 2173 char mmp_nodename[64]; /* Node which last updated MMP block */ 2174 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2175 2176 /* 2177 * mmp_check_interval is used to verify if the MMP block has been 2178 * updated on the block device. The value is updated based on the 2179 * maximum time to write the MMP block during an update cycle. 2180 */ 2181 __le16 mmp_check_interval; 2182 2183 __le16 mmp_pad1; 2184 __le32 mmp_pad2[226]; 2185 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2186 }; 2187 2188 /* arguments passed to the mmp thread */ 2189 struct mmpd_data { 2190 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2191 struct super_block *sb; /* super block of the fs */ 2192 }; 2193 2194 /* 2195 * Check interval multiplier 2196 * The MMP block is written every update interval and initially checked every 2197 * update interval x the multiplier (the value is then adapted based on the 2198 * write latency). The reason is that writes can be delayed under load and we 2199 * don't want readers to incorrectly assume that the filesystem is no longer 2200 * in use. 2201 */ 2202 #define EXT4_MMP_CHECK_MULT 2UL 2203 2204 /* 2205 * Minimum interval for MMP checking in seconds. 2206 */ 2207 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2208 2209 /* 2210 * Maximum interval for MMP checking in seconds. 2211 */ 2212 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2213 2214 /* 2215 * Function prototypes 2216 */ 2217 2218 /* 2219 * Ok, these declarations are also in <linux/kernel.h> but none of the 2220 * ext4 source programs needs to include it so they are duplicated here. 2221 */ 2222 # define NORET_TYPE /**/ 2223 # define ATTRIB_NORET __attribute__((noreturn)) 2224 # define NORET_AND noreturn, 2225 2226 /* bitmap.c */ 2227 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2228 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2229 struct ext4_group_desc *gdp, 2230 struct buffer_head *bh, int sz); 2231 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2232 struct ext4_group_desc *gdp, 2233 struct buffer_head *bh, int sz); 2234 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2235 struct ext4_group_desc *gdp, 2236 struct buffer_head *bh); 2237 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2238 struct ext4_group_desc *gdp, 2239 struct buffer_head *bh); 2240 2241 /* balloc.c */ 2242 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2243 ext4_fsblk_t blocknr, 2244 ext4_group_t *blockgrpp, 2245 ext4_grpblk_t *offsetp); 2246 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2247 ext4_fsblk_t block); 2248 2249 extern unsigned int ext4_block_group(struct super_block *sb, 2250 ext4_fsblk_t blocknr); 2251 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2252 ext4_fsblk_t blocknr); 2253 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2254 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2255 ext4_group_t group); 2256 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2257 ext4_fsblk_t goal, 2258 unsigned int flags, 2259 unsigned long *count, 2260 int *errp); 2261 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2262 s64 nclusters, unsigned int flags); 2263 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2264 extern void ext4_check_blocks_bitmap(struct super_block *); 2265 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2266 ext4_group_t block_group, 2267 struct buffer_head ** bh); 2268 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2269 2270 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2271 ext4_group_t block_group); 2272 extern int ext4_wait_block_bitmap(struct super_block *sb, 2273 ext4_group_t block_group, 2274 struct buffer_head *bh); 2275 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2276 ext4_group_t block_group); 2277 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2278 ext4_group_t block_group, 2279 struct ext4_group_desc *gdp); 2280 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2281 2282 static inline bool ext4_encrypted_inode(struct inode *inode) 2283 { 2284 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2285 } 2286 2287 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2288 static inline int ext4_fname_setup_filename(struct inode *dir, 2289 const struct qstr *iname, 2290 int lookup, struct ext4_filename *fname) 2291 { 2292 struct fscrypt_name name; 2293 int err; 2294 2295 memset(fname, 0, sizeof(struct ext4_filename)); 2296 2297 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2298 2299 fname->usr_fname = name.usr_fname; 2300 fname->disk_name = name.disk_name; 2301 fname->hinfo.hash = name.hash; 2302 fname->hinfo.minor_hash = name.minor_hash; 2303 fname->crypto_buf = name.crypto_buf; 2304 return err; 2305 } 2306 2307 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2308 { 2309 struct fscrypt_name name; 2310 2311 name.crypto_buf = fname->crypto_buf; 2312 fscrypt_free_filename(&name); 2313 2314 fname->crypto_buf.name = NULL; 2315 fname->usr_fname = NULL; 2316 fname->disk_name.name = NULL; 2317 } 2318 #else 2319 static inline int ext4_fname_setup_filename(struct inode *dir, 2320 const struct qstr *iname, 2321 int lookup, struct ext4_filename *fname) 2322 { 2323 fname->usr_fname = iname; 2324 fname->disk_name.name = (unsigned char *) iname->name; 2325 fname->disk_name.len = iname->len; 2326 return 0; 2327 } 2328 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2329 2330 #endif 2331 2332 /* dir.c */ 2333 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2334 struct file *, 2335 struct ext4_dir_entry_2 *, 2336 struct buffer_head *, char *, int, 2337 unsigned int); 2338 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2339 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2340 (de), (bh), (buf), (size), (offset))) 2341 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2342 __u32 minor_hash, 2343 struct ext4_dir_entry_2 *dirent, 2344 struct fscrypt_str *ent_name); 2345 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2346 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2347 struct buffer_head *bh, 2348 void *buf, int buf_size, 2349 struct ext4_filename *fname, 2350 struct ext4_dir_entry_2 **dest_de); 2351 void ext4_insert_dentry(struct inode *inode, 2352 struct ext4_dir_entry_2 *de, 2353 int buf_size, 2354 struct ext4_filename *fname); 2355 static inline void ext4_update_dx_flag(struct inode *inode) 2356 { 2357 if (!ext4_has_feature_dir_index(inode->i_sb)) 2358 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2359 } 2360 static const unsigned char ext4_filetype_table[] = { 2361 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2362 }; 2363 2364 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2365 { 2366 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2367 return DT_UNKNOWN; 2368 2369 return ext4_filetype_table[filetype]; 2370 } 2371 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2372 void *buf, int buf_size); 2373 2374 /* fsync.c */ 2375 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2376 2377 /* hash.c */ 2378 extern int ext4fs_dirhash(const char *name, int len, struct 2379 dx_hash_info *hinfo); 2380 2381 /* ialloc.c */ 2382 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2383 const struct qstr *qstr, __u32 goal, 2384 uid_t *owner, __u32 i_flags, 2385 int handle_type, unsigned int line_no, 2386 int nblocks); 2387 2388 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2389 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2390 i_flags, 0, 0, 0) 2391 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2392 type, nblocks) \ 2393 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2394 0, (type), __LINE__, (nblocks)) 2395 2396 2397 extern void ext4_free_inode(handle_t *, struct inode *); 2398 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2399 extern unsigned long ext4_count_free_inodes(struct super_block *); 2400 extern unsigned long ext4_count_dirs(struct super_block *); 2401 extern void ext4_check_inodes_bitmap(struct super_block *); 2402 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2403 extern int ext4_init_inode_table(struct super_block *sb, 2404 ext4_group_t group, int barrier); 2405 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2406 2407 /* mballoc.c */ 2408 extern const struct seq_operations ext4_mb_seq_groups_ops; 2409 extern long ext4_mb_stats; 2410 extern long ext4_mb_max_to_scan; 2411 extern int ext4_mb_init(struct super_block *); 2412 extern int ext4_mb_release(struct super_block *); 2413 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2414 struct ext4_allocation_request *, int *); 2415 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2416 extern void ext4_discard_preallocations(struct inode *); 2417 extern int __init ext4_init_mballoc(void); 2418 extern void ext4_exit_mballoc(void); 2419 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2420 struct buffer_head *bh, ext4_fsblk_t block, 2421 unsigned long count, int flags); 2422 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2423 ext4_group_t ngroups); 2424 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2425 ext4_group_t i, struct ext4_group_desc *desc); 2426 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2427 ext4_fsblk_t block, unsigned long count); 2428 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2429 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2430 2431 /* inode.c */ 2432 int ext4_inode_is_fast_symlink(struct inode *inode); 2433 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2434 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2435 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2436 bool wait, struct buffer_head **bhs); 2437 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2438 struct buffer_head *bh_result, int create); 2439 int ext4_get_block(struct inode *inode, sector_t iblock, 2440 struct buffer_head *bh_result, int create); 2441 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2442 struct buffer_head *bh_result, int create); 2443 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2444 struct buffer_head *bh, int create); 2445 int ext4_walk_page_buffers(handle_t *handle, 2446 struct buffer_head *head, 2447 unsigned from, 2448 unsigned to, 2449 int *partial, 2450 int (*fn)(handle_t *handle, 2451 struct buffer_head *bh)); 2452 int do_journal_get_write_access(handle_t *handle, 2453 struct buffer_head *bh); 2454 #define FALL_BACK_TO_NONDELALLOC 1 2455 #define CONVERT_INLINE_DATA 2 2456 2457 typedef enum { 2458 EXT4_IGET_NORMAL = 0, 2459 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 2460 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */ 2461 } ext4_iget_flags; 2462 2463 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 2464 ext4_iget_flags flags, const char *function, 2465 unsigned int line); 2466 2467 #define ext4_iget(sb, ino, flags) \ 2468 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 2469 2470 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2471 extern int ext4_setattr(struct dentry *, struct iattr *); 2472 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2473 extern void ext4_evict_inode(struct inode *); 2474 extern void ext4_clear_inode(struct inode *); 2475 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2476 extern int ext4_sync_inode(handle_t *, struct inode *); 2477 extern void ext4_dirty_inode(struct inode *, int); 2478 extern int ext4_change_inode_journal_flag(struct inode *, int); 2479 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2480 extern int ext4_inode_attach_jinode(struct inode *inode); 2481 extern int ext4_can_truncate(struct inode *inode); 2482 extern int ext4_truncate(struct inode *); 2483 extern int ext4_break_layouts(struct inode *); 2484 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2485 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2486 extern void ext4_set_inode_flags(struct inode *); 2487 extern int ext4_alloc_da_blocks(struct inode *inode); 2488 extern void ext4_set_aops(struct inode *inode); 2489 extern int ext4_writepage_trans_blocks(struct inode *); 2490 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2491 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2492 loff_t lstart, loff_t lend); 2493 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 2494 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf); 2495 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2496 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2497 extern void ext4_da_release_space(struct inode *inode, int to_free); 2498 extern void ext4_da_update_reserve_space(struct inode *inode, 2499 int used, int quota_claim); 2500 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2501 ext4_fsblk_t pblk, ext4_lblk_t len); 2502 2503 /* indirect.c */ 2504 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2505 struct ext4_map_blocks *map, int flags); 2506 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2507 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2508 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2509 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2510 ext4_lblk_t start, ext4_lblk_t end); 2511 2512 /* ioctl.c */ 2513 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2514 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2515 2516 /* migrate.c */ 2517 extern int ext4_ext_migrate(struct inode *); 2518 extern int ext4_ind_migrate(struct inode *inode); 2519 2520 /* namei.c */ 2521 extern int ext4_dirent_csum_verify(struct inode *inode, 2522 struct ext4_dir_entry *dirent); 2523 extern int ext4_orphan_add(handle_t *, struct inode *); 2524 extern int ext4_orphan_del(handle_t *, struct inode *); 2525 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2526 __u32 start_minor_hash, __u32 *next_hash); 2527 extern int ext4_search_dir(struct buffer_head *bh, 2528 char *search_buf, 2529 int buf_size, 2530 struct inode *dir, 2531 struct ext4_filename *fname, 2532 unsigned int offset, 2533 struct ext4_dir_entry_2 **res_dir); 2534 extern int ext4_generic_delete_entry(handle_t *handle, 2535 struct inode *dir, 2536 struct ext4_dir_entry_2 *de_del, 2537 struct buffer_head *bh, 2538 void *entry_buf, 2539 int buf_size, 2540 int csum_size); 2541 extern bool ext4_empty_dir(struct inode *inode); 2542 2543 /* resize.c */ 2544 extern int ext4_group_add(struct super_block *sb, 2545 struct ext4_new_group_data *input); 2546 extern int ext4_group_extend(struct super_block *sb, 2547 struct ext4_super_block *es, 2548 ext4_fsblk_t n_blocks_count); 2549 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2550 2551 /* super.c */ 2552 extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 2553 sector_t block, int op_flags); 2554 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2555 extern int ext4_calculate_overhead(struct super_block *sb); 2556 extern void ext4_superblock_csum_set(struct super_block *sb); 2557 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2558 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2559 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2560 ext4_group_t ngroup); 2561 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2562 char nbuf[16]); 2563 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2564 ext4_group_t block_group, 2565 unsigned int flags); 2566 2567 extern __printf(4, 5) 2568 void __ext4_error(struct super_block *, const char *, unsigned int, 2569 const char *, ...); 2570 extern __printf(5, 6) 2571 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2572 const char *, ...); 2573 extern __printf(5, 6) 2574 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2575 const char *, ...); 2576 extern void __ext4_std_error(struct super_block *, const char *, 2577 unsigned int, int); 2578 extern __printf(4, 5) 2579 void __ext4_abort(struct super_block *, const char *, unsigned int, 2580 const char *, ...); 2581 extern __printf(4, 5) 2582 void __ext4_warning(struct super_block *, const char *, unsigned int, 2583 const char *, ...); 2584 extern __printf(4, 5) 2585 void __ext4_warning_inode(const struct inode *inode, const char *function, 2586 unsigned int line, const char *fmt, ...); 2587 extern __printf(3, 4) 2588 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2589 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2590 const char *, unsigned int, const char *); 2591 extern __printf(7, 8) 2592 void __ext4_grp_locked_error(const char *, unsigned int, 2593 struct super_block *, ext4_group_t, 2594 unsigned long, ext4_fsblk_t, 2595 const char *, ...); 2596 2597 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2598 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2599 2600 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2601 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2602 2603 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2604 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2605 2606 #ifdef CONFIG_PRINTK 2607 2608 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2609 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2610 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2611 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2612 #define ext4_error(sb, fmt, ...) \ 2613 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2614 #define ext4_abort(sb, fmt, ...) \ 2615 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2616 #define ext4_warning(sb, fmt, ...) \ 2617 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2618 #define ext4_warning_inode(inode, fmt, ...) \ 2619 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2620 #define ext4_msg(sb, level, fmt, ...) \ 2621 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2622 #define dump_mmp_msg(sb, mmp, msg) \ 2623 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2624 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2625 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2626 fmt, ##__VA_ARGS__) 2627 2628 #else 2629 2630 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2631 do { \ 2632 no_printk(fmt, ##__VA_ARGS__); \ 2633 __ext4_error_inode(inode, "", 0, block, " "); \ 2634 } while (0) 2635 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2636 do { \ 2637 no_printk(fmt, ##__VA_ARGS__); \ 2638 __ext4_error_file(file, "", 0, block, " "); \ 2639 } while (0) 2640 #define ext4_error(sb, fmt, ...) \ 2641 do { \ 2642 no_printk(fmt, ##__VA_ARGS__); \ 2643 __ext4_error(sb, "", 0, " "); \ 2644 } while (0) 2645 #define ext4_abort(sb, fmt, ...) \ 2646 do { \ 2647 no_printk(fmt, ##__VA_ARGS__); \ 2648 __ext4_abort(sb, "", 0, " "); \ 2649 } while (0) 2650 #define ext4_warning(sb, fmt, ...) \ 2651 do { \ 2652 no_printk(fmt, ##__VA_ARGS__); \ 2653 __ext4_warning(sb, "", 0, " "); \ 2654 } while (0) 2655 #define ext4_warning_inode(inode, fmt, ...) \ 2656 do { \ 2657 no_printk(fmt, ##__VA_ARGS__); \ 2658 __ext4_warning_inode(inode, "", 0, " "); \ 2659 } while (0) 2660 #define ext4_msg(sb, level, fmt, ...) \ 2661 do { \ 2662 no_printk(fmt, ##__VA_ARGS__); \ 2663 __ext4_msg(sb, "", " "); \ 2664 } while (0) 2665 #define dump_mmp_msg(sb, mmp, msg) \ 2666 __dump_mmp_msg(sb, mmp, "", 0, "") 2667 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2668 do { \ 2669 no_printk(fmt, ##__VA_ARGS__); \ 2670 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2671 } while (0) 2672 2673 #endif 2674 2675 extern void ext4_update_dynamic_rev(struct super_block *sb); 2676 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2677 __u32 compat); 2678 extern int ext4_update_rocompat_feature(handle_t *handle, 2679 struct super_block *sb, __u32 rocompat); 2680 extern int ext4_update_incompat_feature(handle_t *handle, 2681 struct super_block *sb, __u32 incompat); 2682 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2683 struct ext4_group_desc *bg); 2684 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2685 struct ext4_group_desc *bg); 2686 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2687 struct ext4_group_desc *bg); 2688 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2689 struct ext4_group_desc *bg); 2690 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2691 struct ext4_group_desc *bg); 2692 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2693 struct ext4_group_desc *bg); 2694 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2695 struct ext4_group_desc *bg); 2696 extern void ext4_block_bitmap_set(struct super_block *sb, 2697 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2698 extern void ext4_inode_bitmap_set(struct super_block *sb, 2699 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2700 extern void ext4_inode_table_set(struct super_block *sb, 2701 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2702 extern void ext4_free_group_clusters_set(struct super_block *sb, 2703 struct ext4_group_desc *bg, 2704 __u32 count); 2705 extern void ext4_free_inodes_set(struct super_block *sb, 2706 struct ext4_group_desc *bg, __u32 count); 2707 extern void ext4_used_dirs_set(struct super_block *sb, 2708 struct ext4_group_desc *bg, __u32 count); 2709 extern void ext4_itable_unused_set(struct super_block *sb, 2710 struct ext4_group_desc *bg, __u32 count); 2711 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2712 struct ext4_group_desc *gdp); 2713 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2714 struct ext4_group_desc *gdp); 2715 extern int ext4_register_li_request(struct super_block *sb, 2716 ext4_group_t first_not_zeroed); 2717 2718 static inline int ext4_has_metadata_csum(struct super_block *sb) 2719 { 2720 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2721 !EXT4_SB(sb)->s_chksum_driver); 2722 2723 return ext4_has_feature_metadata_csum(sb) && 2724 (EXT4_SB(sb)->s_chksum_driver != NULL); 2725 } 2726 2727 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2728 { 2729 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2730 } 2731 2732 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2733 { 2734 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2735 le32_to_cpu(es->s_blocks_count_lo); 2736 } 2737 2738 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2739 { 2740 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2741 le32_to_cpu(es->s_r_blocks_count_lo); 2742 } 2743 2744 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2745 { 2746 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2747 le32_to_cpu(es->s_free_blocks_count_lo); 2748 } 2749 2750 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2751 ext4_fsblk_t blk) 2752 { 2753 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2754 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2755 } 2756 2757 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2758 ext4_fsblk_t blk) 2759 { 2760 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2761 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2762 } 2763 2764 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2765 ext4_fsblk_t blk) 2766 { 2767 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2768 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2769 } 2770 2771 static inline loff_t ext4_isize(struct super_block *sb, 2772 struct ext4_inode *raw_inode) 2773 { 2774 if (ext4_has_feature_largedir(sb) || 2775 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2776 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2777 le32_to_cpu(raw_inode->i_size_lo); 2778 2779 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2780 } 2781 2782 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2783 { 2784 raw_inode->i_size_lo = cpu_to_le32(i_size); 2785 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2786 } 2787 2788 static inline 2789 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2790 ext4_group_t group) 2791 { 2792 struct ext4_group_info ***grp_info; 2793 long indexv, indexh; 2794 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2795 grp_info = EXT4_SB(sb)->s_group_info; 2796 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2797 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2798 return grp_info[indexv][indexh]; 2799 } 2800 2801 /* 2802 * Reading s_groups_count requires using smp_rmb() afterwards. See 2803 * the locking protocol documented in the comments of ext4_group_add() 2804 * in resize.c 2805 */ 2806 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2807 { 2808 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2809 2810 smp_rmb(); 2811 return ngroups; 2812 } 2813 2814 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2815 ext4_group_t block_group) 2816 { 2817 return block_group >> sbi->s_log_groups_per_flex; 2818 } 2819 2820 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2821 { 2822 return 1 << sbi->s_log_groups_per_flex; 2823 } 2824 2825 #define ext4_std_error(sb, errno) \ 2826 do { \ 2827 if ((errno)) \ 2828 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2829 } while (0) 2830 2831 #ifdef CONFIG_SMP 2832 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2833 * counters. So we need to make sure we have free clusters more 2834 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2835 */ 2836 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2837 #else 2838 #define EXT4_FREECLUSTERS_WATERMARK 0 2839 #endif 2840 2841 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2842 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2843 { 2844 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2845 !inode_is_locked(inode)); 2846 down_write(&EXT4_I(inode)->i_data_sem); 2847 if (newsize > EXT4_I(inode)->i_disksize) 2848 EXT4_I(inode)->i_disksize = newsize; 2849 up_write(&EXT4_I(inode)->i_data_sem); 2850 } 2851 2852 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2853 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2854 { 2855 int changed = 0; 2856 2857 if (newsize > inode->i_size) { 2858 i_size_write(inode, newsize); 2859 changed = 1; 2860 } 2861 if (newsize > EXT4_I(inode)->i_disksize) { 2862 ext4_update_i_disksize(inode, newsize); 2863 changed |= 2; 2864 } 2865 return changed; 2866 } 2867 2868 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2869 loff_t len); 2870 2871 struct ext4_group_info { 2872 unsigned long bb_state; 2873 struct rb_root bb_free_root; 2874 ext4_grpblk_t bb_first_free; /* first free block */ 2875 ext4_grpblk_t bb_free; /* total free blocks */ 2876 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2877 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2878 struct list_head bb_prealloc_list; 2879 #ifdef DOUBLE_CHECK 2880 void *bb_bitmap; 2881 #endif 2882 struct rw_semaphore alloc_sem; 2883 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2884 * regions, index is order. 2885 * bb_counters[3] = 5 means 2886 * 5 free 8-block regions. */ 2887 }; 2888 2889 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2890 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2891 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2892 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2893 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2894 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2895 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2896 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2897 2898 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2899 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2900 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2901 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2902 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2903 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2904 2905 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2906 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2907 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2908 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2909 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2910 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2911 2912 #define EXT4_MAX_CONTENTION 8 2913 #define EXT4_CONTENTION_THRESHOLD 2 2914 2915 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2916 ext4_group_t group) 2917 { 2918 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2919 } 2920 2921 /* 2922 * Returns true if the filesystem is busy enough that attempts to 2923 * access the block group locks has run into contention. 2924 */ 2925 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2926 { 2927 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2928 } 2929 2930 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2931 { 2932 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2933 if (spin_trylock(lock)) 2934 /* 2935 * We're able to grab the lock right away, so drop the 2936 * lock contention counter. 2937 */ 2938 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2939 else { 2940 /* 2941 * The lock is busy, so bump the contention counter, 2942 * and then wait on the spin lock. 2943 */ 2944 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2945 EXT4_MAX_CONTENTION); 2946 spin_lock(lock); 2947 } 2948 } 2949 2950 static inline void ext4_unlock_group(struct super_block *sb, 2951 ext4_group_t group) 2952 { 2953 spin_unlock(ext4_group_lock_ptr(sb, group)); 2954 } 2955 2956 /* 2957 * Block validity checking 2958 */ 2959 #define ext4_check_indirect_blockref(inode, bh) \ 2960 ext4_check_blockref(__func__, __LINE__, inode, \ 2961 (__le32 *)(bh)->b_data, \ 2962 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2963 2964 #define ext4_ind_check_inode(inode) \ 2965 ext4_check_blockref(__func__, __LINE__, inode, \ 2966 EXT4_I(inode)->i_data, \ 2967 EXT4_NDIR_BLOCKS) 2968 2969 /* 2970 * Inodes and files operations 2971 */ 2972 2973 /* dir.c */ 2974 extern const struct file_operations ext4_dir_operations; 2975 2976 /* file.c */ 2977 extern const struct inode_operations ext4_file_inode_operations; 2978 extern const struct file_operations ext4_file_operations; 2979 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2980 2981 /* inline.c */ 2982 extern int ext4_get_max_inline_size(struct inode *inode); 2983 extern int ext4_find_inline_data_nolock(struct inode *inode); 2984 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2985 unsigned int len); 2986 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2987 2988 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2989 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2990 struct inode *inode, 2991 loff_t pos, unsigned len, 2992 unsigned flags, 2993 struct page **pagep); 2994 extern int ext4_write_inline_data_end(struct inode *inode, 2995 loff_t pos, unsigned len, 2996 unsigned copied, 2997 struct page *page); 2998 extern struct buffer_head * 2999 ext4_journalled_write_inline_data(struct inode *inode, 3000 unsigned len, 3001 struct page *page); 3002 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3003 struct inode *inode, 3004 loff_t pos, unsigned len, 3005 unsigned flags, 3006 struct page **pagep, 3007 void **fsdata); 3008 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3009 unsigned len, unsigned copied, 3010 struct page *page); 3011 extern int ext4_try_add_inline_entry(handle_t *handle, 3012 struct ext4_filename *fname, 3013 struct inode *dir, struct inode *inode); 3014 extern int ext4_try_create_inline_dir(handle_t *handle, 3015 struct inode *parent, 3016 struct inode *inode); 3017 extern int ext4_read_inline_dir(struct file *filp, 3018 struct dir_context *ctx, 3019 int *has_inline_data); 3020 extern int htree_inlinedir_to_tree(struct file *dir_file, 3021 struct inode *dir, ext4_lblk_t block, 3022 struct dx_hash_info *hinfo, 3023 __u32 start_hash, __u32 start_minor_hash, 3024 int *has_inline_data); 3025 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3026 struct ext4_filename *fname, 3027 struct ext4_dir_entry_2 **res_dir, 3028 int *has_inline_data); 3029 extern int ext4_delete_inline_entry(handle_t *handle, 3030 struct inode *dir, 3031 struct ext4_dir_entry_2 *de_del, 3032 struct buffer_head *bh, 3033 int *has_inline_data); 3034 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3035 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3036 struct ext4_dir_entry_2 **parent_de, 3037 int *retval); 3038 extern int ext4_inline_data_fiemap(struct inode *inode, 3039 struct fiemap_extent_info *fieinfo, 3040 int *has_inline, __u64 start, __u64 len); 3041 3042 struct iomap; 3043 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3044 3045 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3046 3047 extern int ext4_convert_inline_data(struct inode *inode); 3048 3049 static inline int ext4_has_inline_data(struct inode *inode) 3050 { 3051 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3052 EXT4_I(inode)->i_inline_off; 3053 } 3054 3055 /* namei.c */ 3056 extern const struct inode_operations ext4_dir_inode_operations; 3057 extern const struct inode_operations ext4_special_inode_operations; 3058 extern struct dentry *ext4_get_parent(struct dentry *child); 3059 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3060 struct ext4_dir_entry_2 *de, 3061 int blocksize, int csum_size, 3062 unsigned int parent_ino, int dotdot_real_len); 3063 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3064 unsigned int blocksize); 3065 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3066 struct inode *inode, 3067 struct buffer_head *bh); 3068 #define S_SHIFT 12 3069 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3070 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3071 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3072 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3073 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3074 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3075 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3076 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3077 }; 3078 3079 static inline void ext4_set_de_type(struct super_block *sb, 3080 struct ext4_dir_entry_2 *de, 3081 umode_t mode) { 3082 if (ext4_has_feature_filetype(sb)) 3083 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3084 } 3085 3086 /* readpages.c */ 3087 extern int ext4_mpage_readpages(struct address_space *mapping, 3088 struct list_head *pages, struct page *page, 3089 unsigned nr_pages, bool is_readahead); 3090 3091 /* symlink.c */ 3092 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3093 extern const struct inode_operations ext4_symlink_inode_operations; 3094 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3095 3096 /* sysfs.c */ 3097 extern int ext4_register_sysfs(struct super_block *sb); 3098 extern void ext4_unregister_sysfs(struct super_block *sb); 3099 extern int __init ext4_init_sysfs(void); 3100 extern void ext4_exit_sysfs(void); 3101 3102 /* block_validity */ 3103 extern void ext4_release_system_zone(struct super_block *sb); 3104 extern int ext4_setup_system_zone(struct super_block *sb); 3105 extern int __init ext4_init_system_zone(void); 3106 extern void ext4_exit_system_zone(void); 3107 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3108 ext4_fsblk_t start_blk, 3109 unsigned int count); 3110 extern int ext4_check_blockref(const char *, unsigned int, 3111 struct inode *, __le32 *, unsigned int); 3112 3113 /* extents.c */ 3114 struct ext4_ext_path; 3115 struct ext4_extent; 3116 3117 /* 3118 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3119 * __le32. 3120 */ 3121 #define EXT_MAX_BLOCKS 0xffffffff 3122 3123 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3124 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3125 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3126 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3127 struct ext4_map_blocks *map, int flags); 3128 extern int ext4_ext_truncate(handle_t *, struct inode *); 3129 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3130 ext4_lblk_t end); 3131 extern void ext4_ext_init(struct super_block *); 3132 extern void ext4_ext_release(struct super_block *); 3133 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3134 loff_t len); 3135 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3136 loff_t offset, ssize_t len); 3137 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3138 struct ext4_map_blocks *map, int flags); 3139 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3140 ext4_lblk_t lblocks); 3141 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3142 int num, 3143 struct ext4_ext_path *path); 3144 extern int ext4_can_extents_be_merged(struct inode *inode, 3145 struct ext4_extent *ex1, 3146 struct ext4_extent *ex2); 3147 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3148 struct ext4_ext_path **, 3149 struct ext4_extent *, int); 3150 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3151 struct ext4_ext_path **, 3152 int flags); 3153 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3154 extern int ext4_ext_check_inode(struct inode *inode); 3155 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3156 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3157 __u64 start, __u64 len); 3158 extern int ext4_ext_precache(struct inode *inode); 3159 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3160 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3161 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3162 struct inode *inode2, ext4_lblk_t lblk1, 3163 ext4_lblk_t lblk2, ext4_lblk_t count, 3164 int mark_unwritten,int *err); 3165 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3166 3167 /* move_extent.c */ 3168 extern void ext4_double_down_write_data_sem(struct inode *first, 3169 struct inode *second); 3170 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3171 struct inode *donor_inode); 3172 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3173 __u64 start_orig, __u64 start_donor, 3174 __u64 len, __u64 *moved_len); 3175 3176 /* page-io.c */ 3177 extern int __init ext4_init_pageio(void); 3178 extern void ext4_exit_pageio(void); 3179 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3180 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3181 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3182 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3183 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3184 struct writeback_control *wbc); 3185 extern void ext4_end_io_rsv_work(struct work_struct *work); 3186 extern void ext4_io_submit(struct ext4_io_submit *io); 3187 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3188 struct page *page, 3189 int len, 3190 struct writeback_control *wbc, 3191 bool keep_towrite); 3192 3193 /* mmp.c */ 3194 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3195 3196 /* 3197 * Add new method to test whether block and inode bitmaps are properly 3198 * initialized. With uninit_bg reading the block from disk is not enough 3199 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3200 */ 3201 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3202 3203 static inline int bitmap_uptodate(struct buffer_head *bh) 3204 { 3205 return (buffer_uptodate(bh) && 3206 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3207 } 3208 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3209 { 3210 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3211 } 3212 3213 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3214 3215 /* For ioend & aio unwritten conversion wait queues */ 3216 #define EXT4_WQ_HASH_SZ 37 3217 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3218 EXT4_WQ_HASH_SZ]) 3219 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3220 3221 extern int ext4_resize_begin(struct super_block *sb); 3222 extern void ext4_resize_end(struct super_block *sb); 3223 3224 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3225 struct ext4_io_end *io_end) 3226 { 3227 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3228 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3229 atomic_inc(&EXT4_I(inode)->i_unwritten); 3230 } 3231 } 3232 3233 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3234 { 3235 struct inode *inode = io_end->inode; 3236 3237 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3238 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3239 /* Wake up anyone waiting on unwritten extent conversion */ 3240 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3241 wake_up_all(ext4_ioend_wq(inode)); 3242 } 3243 } 3244 3245 extern const struct iomap_ops ext4_iomap_ops; 3246 3247 #endif /* __KERNEL__ */ 3248 3249 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3250 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3251 3252 #endif /* _EXT4_H */ 3253