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 #ifdef CONFIG_EXT4_FS_ENCRYPTION 38 #include <linux/fscrypt_supp.h> 39 #else 40 #include <linux/fscrypt_notsupp.h> 41 #endif 42 #include <linux/falloc.h> 43 #include <linux/percpu-rwsem.h> 44 #ifdef __KERNEL__ 45 #include <linux/compat.h> 46 #endif 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 unused; 550 __u32 free_blocks_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 622 623 /* 624 * ioctl commands 625 */ 626 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 627 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 628 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 629 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 630 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 631 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 632 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 633 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 634 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 635 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 636 #define EXT4_IOC_MIGRATE _IO('f', 9) 637 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 638 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 639 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 640 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 641 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 642 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 643 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 644 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 645 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 646 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 647 648 #ifndef FS_IOC_FSGETXATTR 649 /* Until the uapi changes get merged for project quota... */ 650 651 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr) 652 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr) 653 654 /* 655 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR. 656 */ 657 struct fsxattr { 658 __u32 fsx_xflags; /* xflags field value (get/set) */ 659 __u32 fsx_extsize; /* extsize field value (get/set)*/ 660 __u32 fsx_nextents; /* nextents field value (get) */ 661 __u32 fsx_projid; /* project identifier (get/set) */ 662 unsigned char fsx_pad[12]; 663 }; 664 665 /* 666 * Flags for the fsx_xflags field 667 */ 668 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */ 669 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */ 670 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */ 671 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */ 672 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */ 673 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */ 674 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */ 675 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */ 676 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */ 677 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */ 678 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */ 679 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */ 680 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */ 681 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */ 682 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */ 683 #endif /* !defined(FS_IOC_FSGETXATTR) */ 684 685 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 686 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 687 688 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 689 690 /* 691 * Flags for going down operation 692 */ 693 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 694 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 695 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 696 697 698 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 699 /* 700 * ioctl commands in 32 bit emulation 701 */ 702 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 703 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 704 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 705 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 706 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 707 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 708 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 709 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 710 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 711 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 712 #endif 713 714 /* Max physical block we can address w/o extents */ 715 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 716 717 /* 718 * Structure of an inode on the disk 719 */ 720 struct ext4_inode { 721 __le16 i_mode; /* File mode */ 722 __le16 i_uid; /* Low 16 bits of Owner Uid */ 723 __le32 i_size_lo; /* Size in bytes */ 724 __le32 i_atime; /* Access time */ 725 __le32 i_ctime; /* Inode Change time */ 726 __le32 i_mtime; /* Modification time */ 727 __le32 i_dtime; /* Deletion Time */ 728 __le16 i_gid; /* Low 16 bits of Group Id */ 729 __le16 i_links_count; /* Links count */ 730 __le32 i_blocks_lo; /* Blocks count */ 731 __le32 i_flags; /* File flags */ 732 union { 733 struct { 734 __le32 l_i_version; 735 } linux1; 736 struct { 737 __u32 h_i_translator; 738 } hurd1; 739 struct { 740 __u32 m_i_reserved1; 741 } masix1; 742 } osd1; /* OS dependent 1 */ 743 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 744 __le32 i_generation; /* File version (for NFS) */ 745 __le32 i_file_acl_lo; /* File ACL */ 746 __le32 i_size_high; 747 __le32 i_obso_faddr; /* Obsoleted fragment address */ 748 union { 749 struct { 750 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 751 __le16 l_i_file_acl_high; 752 __le16 l_i_uid_high; /* these 2 fields */ 753 __le16 l_i_gid_high; /* were reserved2[0] */ 754 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 755 __le16 l_i_reserved; 756 } linux2; 757 struct { 758 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 759 __u16 h_i_mode_high; 760 __u16 h_i_uid_high; 761 __u16 h_i_gid_high; 762 __u32 h_i_author; 763 } hurd2; 764 struct { 765 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 766 __le16 m_i_file_acl_high; 767 __u32 m_i_reserved2[2]; 768 } masix2; 769 } osd2; /* OS dependent 2 */ 770 __le16 i_extra_isize; 771 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 772 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 773 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 774 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 775 __le32 i_crtime; /* File Creation time */ 776 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 777 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 778 __le32 i_projid; /* Project ID */ 779 }; 780 781 struct move_extent { 782 __u32 reserved; /* should be zero */ 783 __u32 donor_fd; /* donor file descriptor */ 784 __u64 orig_start; /* logical start offset in block for orig */ 785 __u64 donor_start; /* logical start offset in block for donor */ 786 __u64 len; /* block length to be moved */ 787 __u64 moved_len; /* moved block length */ 788 }; 789 790 #define EXT4_EPOCH_BITS 2 791 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 792 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 793 794 /* 795 * Extended fields will fit into an inode if the filesystem was formatted 796 * with large inodes (-I 256 or larger) and there are not currently any EAs 797 * consuming all of the available space. For new inodes we always reserve 798 * enough space for the kernel's known extended fields, but for inodes 799 * created with an old kernel this might not have been the case. None of 800 * the extended inode fields is critical for correct filesystem operation. 801 * This macro checks if a certain field fits in the inode. Note that 802 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 803 */ 804 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 805 ((offsetof(typeof(*ext4_inode), field) + \ 806 sizeof((ext4_inode)->field)) \ 807 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 808 (einode)->i_extra_isize)) \ 809 810 /* 811 * We use an encoding that preserves the times for extra epoch "00": 812 * 813 * extra msb of adjust for signed 814 * epoch 32-bit 32-bit tv_sec to 815 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 816 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 817 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 818 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 819 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 820 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 821 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 822 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 823 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 824 * 825 * Note that previous versions of the kernel on 64-bit systems would 826 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 827 * 1970. e2fsck will correct this, assuming that it is run on the 828 * affected filesystem before 2242. 829 */ 830 831 static inline __le32 ext4_encode_extra_time(struct timespec *time) 832 { 833 u32 extra = sizeof(time->tv_sec) > 4 ? 834 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0; 835 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 836 } 837 838 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 839 { 840 if (unlikely(sizeof(time->tv_sec) > 4 && 841 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) { 842 843 #if 1 844 /* Handle legacy encoding of pre-1970 dates with epoch 845 * bits 1,1. (This backwards compatibility may be removed 846 * at the discretion of the ext4 developers.) 847 */ 848 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 849 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 850 extra_bits = 0; 851 time->tv_sec += extra_bits << 32; 852 #else 853 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 854 #endif 855 } 856 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 857 } 858 859 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 860 do { \ 861 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 862 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 863 (raw_inode)->xtime ## _extra = \ 864 ext4_encode_extra_time(&(inode)->xtime); \ 865 } while (0) 866 867 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 868 do { \ 869 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 870 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 871 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 872 (raw_inode)->xtime ## _extra = \ 873 ext4_encode_extra_time(&(einode)->xtime); \ 874 } while (0) 875 876 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 877 do { \ 878 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 879 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 880 ext4_decode_extra_time(&(inode)->xtime, \ 881 raw_inode->xtime ## _extra); \ 882 else \ 883 (inode)->xtime.tv_nsec = 0; \ 884 } while (0) 885 886 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 887 do { \ 888 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 889 (einode)->xtime.tv_sec = \ 890 (signed)le32_to_cpu((raw_inode)->xtime); \ 891 else \ 892 (einode)->xtime.tv_sec = 0; \ 893 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 894 ext4_decode_extra_time(&(einode)->xtime, \ 895 raw_inode->xtime ## _extra); \ 896 else \ 897 (einode)->xtime.tv_nsec = 0; \ 898 } while (0) 899 900 #define i_disk_version osd1.linux1.l_i_version 901 902 #if defined(__KERNEL__) || defined(__linux__) 903 #define i_reserved1 osd1.linux1.l_i_reserved1 904 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 905 #define i_blocks_high osd2.linux2.l_i_blocks_high 906 #define i_uid_low i_uid 907 #define i_gid_low i_gid 908 #define i_uid_high osd2.linux2.l_i_uid_high 909 #define i_gid_high osd2.linux2.l_i_gid_high 910 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 911 912 #elif defined(__GNU__) 913 914 #define i_translator osd1.hurd1.h_i_translator 915 #define i_uid_high osd2.hurd2.h_i_uid_high 916 #define i_gid_high osd2.hurd2.h_i_gid_high 917 #define i_author osd2.hurd2.h_i_author 918 919 #elif defined(__masix__) 920 921 #define i_reserved1 osd1.masix1.m_i_reserved1 922 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 923 #define i_reserved2 osd2.masix2.m_i_reserved2 924 925 #endif /* defined(__KERNEL__) || defined(__linux__) */ 926 927 #include "extents_status.h" 928 929 /* 930 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 931 * 932 * These are needed to avoid lockdep false positives when we need to 933 * allocate blocks to the quota inode during ext4_map_blocks(), while 934 * holding i_data_sem for a normal (non-quota) inode. Since we don't 935 * do quota tracking for the quota inode, this avoids deadlock (as 936 * well as infinite recursion, since it isn't turtles all the way 937 * down...) 938 * 939 * I_DATA_SEM_NORMAL - Used for most inodes 940 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 941 * where the second inode has larger inode number 942 * than the first 943 * I_DATA_SEM_QUOTA - Used for quota inodes only 944 */ 945 enum { 946 I_DATA_SEM_NORMAL = 0, 947 I_DATA_SEM_OTHER, 948 I_DATA_SEM_QUOTA, 949 }; 950 951 952 /* 953 * fourth extended file system inode data in memory 954 */ 955 struct ext4_inode_info { 956 __le32 i_data[15]; /* unconverted */ 957 __u32 i_dtime; 958 ext4_fsblk_t i_file_acl; 959 960 /* 961 * i_block_group is the number of the block group which contains 962 * this file's inode. Constant across the lifetime of the inode, 963 * it is used for making block allocation decisions - we try to 964 * place a file's data blocks near its inode block, and new inodes 965 * near to their parent directory's inode. 966 */ 967 ext4_group_t i_block_group; 968 ext4_lblk_t i_dir_start_lookup; 969 #if (BITS_PER_LONG < 64) 970 unsigned long i_state_flags; /* Dynamic state flags */ 971 #endif 972 unsigned long i_flags; 973 974 /* 975 * Extended attributes can be read independently of the main file 976 * data. Taking i_mutex even when reading would cause contention 977 * between readers of EAs and writers of regular file data, so 978 * instead we synchronize on xattr_sem when reading or changing 979 * EAs. 980 */ 981 struct rw_semaphore xattr_sem; 982 983 struct list_head i_orphan; /* unlinked but open inodes */ 984 985 /* 986 * i_disksize keeps track of what the inode size is ON DISK, not 987 * in memory. During truncate, i_size is set to the new size by 988 * the VFS prior to calling ext4_truncate(), but the filesystem won't 989 * set i_disksize to 0 until the truncate is actually under way. 990 * 991 * The intent is that i_disksize always represents the blocks which 992 * are used by this file. This allows recovery to restart truncate 993 * on orphans if we crash during truncate. We actually write i_disksize 994 * into the on-disk inode when writing inodes out, instead of i_size. 995 * 996 * The only time when i_disksize and i_size may be different is when 997 * a truncate is in progress. The only things which change i_disksize 998 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 999 */ 1000 loff_t i_disksize; 1001 1002 /* 1003 * i_data_sem is for serialising ext4_truncate() against 1004 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1005 * data tree are chopped off during truncate. We can't do that in 1006 * ext4 because whenever we perform intermediate commits during 1007 * truncate, the inode and all the metadata blocks *must* be in a 1008 * consistent state which allows truncation of the orphans to restart 1009 * during recovery. Hence we must fix the get_block-vs-truncate race 1010 * by other means, so we have i_data_sem. 1011 */ 1012 struct rw_semaphore i_data_sem; 1013 /* 1014 * i_mmap_sem is for serializing page faults with truncate / punch hole 1015 * operations. We have to make sure that new page cannot be faulted in 1016 * a section of the inode that is being punched. We cannot easily use 1017 * i_data_sem for this since we need protection for the whole punch 1018 * operation and i_data_sem ranks below transaction start so we have 1019 * to occasionally drop it. 1020 */ 1021 struct rw_semaphore i_mmap_sem; 1022 struct inode vfs_inode; 1023 struct jbd2_inode *jinode; 1024 1025 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1026 1027 /* 1028 * File creation time. Its function is same as that of 1029 * struct timespec i_{a,c,m}time in the generic inode. 1030 */ 1031 struct timespec i_crtime; 1032 1033 /* mballoc */ 1034 struct list_head i_prealloc_list; 1035 spinlock_t i_prealloc_lock; 1036 1037 /* extents status tree */ 1038 struct ext4_es_tree i_es_tree; 1039 rwlock_t i_es_lock; 1040 struct list_head i_es_list; 1041 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1042 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1043 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1044 extents to shrink. Protected by 1045 i_es_lock */ 1046 1047 /* ialloc */ 1048 ext4_group_t i_last_alloc_group; 1049 1050 /* allocation reservation info for delalloc */ 1051 /* In case of bigalloc, this refer to clusters rather than blocks */ 1052 unsigned int i_reserved_data_blocks; 1053 ext4_lblk_t i_da_metadata_calc_last_lblock; 1054 int i_da_metadata_calc_len; 1055 1056 /* on-disk additional length */ 1057 __u16 i_extra_isize; 1058 1059 /* Indicate the inline data space. */ 1060 u16 i_inline_off; 1061 u16 i_inline_size; 1062 1063 #ifdef CONFIG_QUOTA 1064 /* quota space reservation, managed internally by quota code */ 1065 qsize_t i_reserved_quota; 1066 #endif 1067 1068 /* Lock protecting lists below */ 1069 spinlock_t i_completed_io_lock; 1070 /* 1071 * Completed IOs that need unwritten extents handling and have 1072 * transaction reserved 1073 */ 1074 struct list_head i_rsv_conversion_list; 1075 struct work_struct i_rsv_conversion_work; 1076 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1077 1078 spinlock_t i_block_reservation_lock; 1079 1080 /* 1081 * Transactions that contain inode's metadata needed to complete 1082 * fsync and fdatasync, respectively. 1083 */ 1084 tid_t i_sync_tid; 1085 tid_t i_datasync_tid; 1086 1087 #ifdef CONFIG_QUOTA 1088 struct dquot *i_dquot[MAXQUOTAS]; 1089 #endif 1090 1091 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1092 __u32 i_csum_seed; 1093 1094 kprojid_t i_projid; 1095 }; 1096 1097 /* 1098 * File system states 1099 */ 1100 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1101 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1102 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1103 1104 /* 1105 * Misc. filesystem flags 1106 */ 1107 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1108 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1109 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1110 1111 /* 1112 * Mount flags set via mount options or defaults 1113 */ 1114 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1115 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1116 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1117 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1118 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1119 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1120 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1121 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1122 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1123 #ifdef CONFIG_FS_DAX 1124 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1125 #else 1126 #define EXT4_MOUNT_DAX 0 1127 #endif 1128 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1129 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1130 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1131 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1132 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1133 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1134 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1135 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1136 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1137 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1138 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1139 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1140 * enable enforcement for hidden 1141 * quota files */ 1142 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1143 * enforcement for hidden quota 1144 * files */ 1145 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1146 * enforcement */ 1147 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1148 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1149 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1150 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1151 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1152 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1153 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1154 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1155 1156 /* 1157 * Mount flags set either automatically (could not be set by mount option) 1158 * based on per file system feature or property or in special cases such as 1159 * distinguishing between explicit mount option definition and default. 1160 */ 1161 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1162 specified delalloc */ 1163 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1164 size of blocksize * 8 1165 blocks */ 1166 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1167 file systems */ 1168 1169 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1170 specified journal checksum */ 1171 1172 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1173 ~EXT4_MOUNT_##opt 1174 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1175 EXT4_MOUNT_##opt 1176 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1177 EXT4_MOUNT_##opt) 1178 1179 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1180 ~EXT4_MOUNT2_##opt 1181 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1182 EXT4_MOUNT2_##opt 1183 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1184 EXT4_MOUNT2_##opt) 1185 1186 #define ext4_test_and_set_bit __test_and_set_bit_le 1187 #define ext4_set_bit __set_bit_le 1188 #define ext4_set_bit_atomic ext2_set_bit_atomic 1189 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1190 #define ext4_clear_bit __clear_bit_le 1191 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1192 #define ext4_test_bit test_bit_le 1193 #define ext4_find_next_zero_bit find_next_zero_bit_le 1194 #define ext4_find_next_bit find_next_bit_le 1195 1196 extern void ext4_set_bits(void *bm, int cur, int len); 1197 1198 /* 1199 * Maximal mount counts between two filesystem checks 1200 */ 1201 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1202 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1203 1204 /* 1205 * Behaviour when detecting errors 1206 */ 1207 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1208 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1209 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1210 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1211 1212 /* Metadata checksum algorithm codes */ 1213 #define EXT4_CRC32C_CHKSUM 1 1214 1215 /* 1216 * Structure of the super block 1217 */ 1218 struct ext4_super_block { 1219 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1220 __le32 s_blocks_count_lo; /* Blocks count */ 1221 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1222 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1223 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1224 __le32 s_first_data_block; /* First Data Block */ 1225 __le32 s_log_block_size; /* Block size */ 1226 __le32 s_log_cluster_size; /* Allocation cluster size */ 1227 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1228 __le32 s_clusters_per_group; /* # Clusters per group */ 1229 __le32 s_inodes_per_group; /* # Inodes per group */ 1230 __le32 s_mtime; /* Mount time */ 1231 /*30*/ __le32 s_wtime; /* Write time */ 1232 __le16 s_mnt_count; /* Mount count */ 1233 __le16 s_max_mnt_count; /* Maximal mount count */ 1234 __le16 s_magic; /* Magic signature */ 1235 __le16 s_state; /* File system state */ 1236 __le16 s_errors; /* Behaviour when detecting errors */ 1237 __le16 s_minor_rev_level; /* minor revision level */ 1238 /*40*/ __le32 s_lastcheck; /* time of last check */ 1239 __le32 s_checkinterval; /* max. time between checks */ 1240 __le32 s_creator_os; /* OS */ 1241 __le32 s_rev_level; /* Revision level */ 1242 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1243 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1244 /* 1245 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1246 * 1247 * Note: the difference between the compatible feature set and 1248 * the incompatible feature set is that if there is a bit set 1249 * in the incompatible feature set that the kernel doesn't 1250 * know about, it should refuse to mount the filesystem. 1251 * 1252 * e2fsck's requirements are more strict; if it doesn't know 1253 * about a feature in either the compatible or incompatible 1254 * feature set, it must abort and not try to meddle with 1255 * things it doesn't understand... 1256 */ 1257 __le32 s_first_ino; /* First non-reserved inode */ 1258 __le16 s_inode_size; /* size of inode structure */ 1259 __le16 s_block_group_nr; /* block group # of this superblock */ 1260 __le32 s_feature_compat; /* compatible feature set */ 1261 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1262 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1263 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1264 /*78*/ char s_volume_name[16]; /* volume name */ 1265 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1266 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1267 /* 1268 * Performance hints. Directory preallocation should only 1269 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1270 */ 1271 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1272 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1273 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1274 /* 1275 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1276 */ 1277 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1278 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1279 __le32 s_journal_dev; /* device number of journal file */ 1280 __le32 s_last_orphan; /* start of list of inodes to delete */ 1281 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1282 __u8 s_def_hash_version; /* Default hash version to use */ 1283 __u8 s_jnl_backup_type; 1284 __le16 s_desc_size; /* size of group descriptor */ 1285 /*100*/ __le32 s_default_mount_opts; 1286 __le32 s_first_meta_bg; /* First metablock block group */ 1287 __le32 s_mkfs_time; /* When the filesystem was created */ 1288 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1289 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1290 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1291 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1292 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1293 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1294 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1295 __le32 s_flags; /* Miscellaneous flags */ 1296 __le16 s_raid_stride; /* RAID stride */ 1297 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1298 __le64 s_mmp_block; /* Block for multi-mount protection */ 1299 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1300 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1301 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1302 __u8 s_encryption_level; /* versioning level for encryption */ 1303 __u8 s_reserved_pad; /* Padding to next 32bits */ 1304 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1305 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1306 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1307 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1308 snapshot's future use */ 1309 __le32 s_snapshot_list; /* inode number of the head of the 1310 on-disk snapshot list */ 1311 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1312 __le32 s_error_count; /* number of fs errors */ 1313 __le32 s_first_error_time; /* first time an error happened */ 1314 __le32 s_first_error_ino; /* inode involved in first error */ 1315 __le64 s_first_error_block; /* block involved of first error */ 1316 __u8 s_first_error_func[32]; /* function where the error happened */ 1317 __le32 s_first_error_line; /* line number where error happened */ 1318 __le32 s_last_error_time; /* most recent time of an error */ 1319 __le32 s_last_error_ino; /* inode involved in last error */ 1320 __le32 s_last_error_line; /* line number where error happened */ 1321 __le64 s_last_error_block; /* block involved of last error */ 1322 __u8 s_last_error_func[32]; /* function where the error happened */ 1323 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1324 __u8 s_mount_opts[64]; 1325 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1326 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1327 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1328 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1329 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1330 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1331 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1332 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1333 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1334 __le32 s_reserved[98]; /* Padding to the end of the block */ 1335 __le32 s_checksum; /* crc32c(superblock) */ 1336 }; 1337 1338 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1339 1340 #ifdef __KERNEL__ 1341 1342 /* 1343 * run-time mount flags 1344 */ 1345 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1346 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1347 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1348 1349 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1350 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1351 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1352 #else 1353 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1354 #endif 1355 1356 /* Number of quota types we support */ 1357 #define EXT4_MAXQUOTAS 3 1358 1359 /* 1360 * fourth extended-fs super-block data in memory 1361 */ 1362 struct ext4_sb_info { 1363 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1364 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1365 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1366 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1367 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1368 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1369 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1370 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1371 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1372 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1373 unsigned long s_overhead; /* # of fs overhead clusters */ 1374 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1375 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1376 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1377 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1378 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1379 struct buffer_head **s_group_desc; 1380 unsigned int s_mount_opt; 1381 unsigned int s_mount_opt2; 1382 unsigned int s_mount_flags; 1383 unsigned int s_def_mount_opt; 1384 ext4_fsblk_t s_sb_block; 1385 atomic64_t s_resv_clusters; 1386 kuid_t s_resuid; 1387 kgid_t s_resgid; 1388 unsigned short s_mount_state; 1389 unsigned short s_pad; 1390 int s_addr_per_block_bits; 1391 int s_desc_per_block_bits; 1392 int s_inode_size; 1393 int s_first_ino; 1394 unsigned int s_inode_readahead_blks; 1395 unsigned int s_inode_goal; 1396 spinlock_t s_next_gen_lock; 1397 u32 s_next_generation; 1398 u32 s_hash_seed[4]; 1399 int s_def_hash_version; 1400 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1401 struct percpu_counter s_freeclusters_counter; 1402 struct percpu_counter s_freeinodes_counter; 1403 struct percpu_counter s_dirs_counter; 1404 struct percpu_counter s_dirtyclusters_counter; 1405 struct blockgroup_lock *s_blockgroup_lock; 1406 struct proc_dir_entry *s_proc; 1407 struct kobject s_kobj; 1408 struct completion s_kobj_unregister; 1409 struct super_block *s_sb; 1410 1411 /* Journaling */ 1412 struct journal_s *s_journal; 1413 struct list_head s_orphan; 1414 struct mutex s_orphan_lock; 1415 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1416 unsigned long s_commit_interval; 1417 u32 s_max_batch_time; 1418 u32 s_min_batch_time; 1419 struct block_device *journal_bdev; 1420 #ifdef CONFIG_QUOTA 1421 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1422 int s_jquota_fmt; /* Format of quota to use */ 1423 #endif 1424 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1425 struct rb_root system_blks; 1426 1427 #ifdef EXTENTS_STATS 1428 /* ext4 extents stats */ 1429 unsigned long s_ext_min; 1430 unsigned long s_ext_max; 1431 unsigned long s_depth_max; 1432 spinlock_t s_ext_stats_lock; 1433 unsigned long s_ext_blocks; 1434 unsigned long s_ext_extents; 1435 #endif 1436 1437 /* for buddy allocator */ 1438 struct ext4_group_info ***s_group_info; 1439 struct inode *s_buddy_cache; 1440 spinlock_t s_md_lock; 1441 unsigned short *s_mb_offsets; 1442 unsigned int *s_mb_maxs; 1443 unsigned int s_group_info_size; 1444 unsigned int s_mb_free_pending; 1445 struct list_head s_freed_data_list; /* List of blocks to be freed 1446 after commit completed */ 1447 1448 /* tunables */ 1449 unsigned long s_stripe; 1450 unsigned int s_mb_stream_request; 1451 unsigned int s_mb_max_to_scan; 1452 unsigned int s_mb_min_to_scan; 1453 unsigned int s_mb_stats; 1454 unsigned int s_mb_order2_reqs; 1455 unsigned int s_mb_group_prealloc; 1456 unsigned int s_max_dir_size_kb; 1457 /* where last allocation was done - for stream allocation */ 1458 unsigned long s_mb_last_group; 1459 unsigned long s_mb_last_start; 1460 1461 /* stats for buddy allocator */ 1462 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1463 atomic_t s_bal_success; /* we found long enough chunks */ 1464 atomic_t s_bal_allocated; /* in blocks */ 1465 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1466 atomic_t s_bal_goals; /* goal hits */ 1467 atomic_t s_bal_breaks; /* too long searches */ 1468 atomic_t s_bal_2orders; /* 2^order hits */ 1469 spinlock_t s_bal_lock; 1470 unsigned long s_mb_buddies_generated; 1471 unsigned long long s_mb_generation_time; 1472 atomic_t s_mb_lost_chunks; 1473 atomic_t s_mb_preallocated; 1474 atomic_t s_mb_discarded; 1475 atomic_t s_lock_busy; 1476 1477 /* locality groups */ 1478 struct ext4_locality_group __percpu *s_locality_groups; 1479 1480 /* for write statistics */ 1481 unsigned long s_sectors_written_start; 1482 u64 s_kbytes_written; 1483 1484 /* the size of zero-out chunk */ 1485 unsigned int s_extent_max_zeroout_kb; 1486 1487 unsigned int s_log_groups_per_flex; 1488 struct flex_groups *s_flex_groups; 1489 ext4_group_t s_flex_groups_allocated; 1490 1491 /* workqueue for reserved extent conversions (buffered io) */ 1492 struct workqueue_struct *rsv_conversion_wq; 1493 1494 /* timer for periodic error stats printing */ 1495 struct timer_list s_err_report; 1496 1497 /* Lazy inode table initialization info */ 1498 struct ext4_li_request *s_li_request; 1499 /* Wait multiplier for lazy initialization thread */ 1500 unsigned int s_li_wait_mult; 1501 1502 /* Kernel thread for multiple mount protection */ 1503 struct task_struct *s_mmp_tsk; 1504 1505 /* record the last minlen when FITRIM is called. */ 1506 atomic_t s_last_trim_minblks; 1507 1508 /* Reference to checksum algorithm driver via cryptoapi */ 1509 struct crypto_shash *s_chksum_driver; 1510 1511 /* Precomputed FS UUID checksum for seeding other checksums */ 1512 __u32 s_csum_seed; 1513 1514 /* Reclaim extents from extent status tree */ 1515 struct shrinker s_es_shrinker; 1516 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1517 long s_es_nr_inode; 1518 struct ext4_es_stats s_es_stats; 1519 struct mb_cache *s_ea_block_cache; 1520 struct mb_cache *s_ea_inode_cache; 1521 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1522 1523 /* Ratelimit ext4 messages. */ 1524 struct ratelimit_state s_err_ratelimit_state; 1525 struct ratelimit_state s_warning_ratelimit_state; 1526 struct ratelimit_state s_msg_ratelimit_state; 1527 1528 /* Barrier between changing inodes' journal flags and writepages ops. */ 1529 struct percpu_rw_semaphore s_journal_flag_rwsem; 1530 struct dax_device *s_daxdev; 1531 }; 1532 1533 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1534 { 1535 return sb->s_fs_info; 1536 } 1537 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1538 { 1539 return container_of(inode, struct ext4_inode_info, vfs_inode); 1540 } 1541 1542 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1543 { 1544 return ino == EXT4_ROOT_INO || 1545 ino == EXT4_USR_QUOTA_INO || 1546 ino == EXT4_GRP_QUOTA_INO || 1547 ino == EXT4_BOOT_LOADER_INO || 1548 ino == EXT4_JOURNAL_INO || 1549 ino == EXT4_RESIZE_INO || 1550 (ino >= EXT4_FIRST_INO(sb) && 1551 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1552 } 1553 1554 /* 1555 * Inode dynamic state flags 1556 */ 1557 enum { 1558 EXT4_STATE_JDATA, /* journaled data exists */ 1559 EXT4_STATE_NEW, /* inode is newly created */ 1560 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1561 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1562 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1563 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1564 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1565 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1566 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1567 nolocking */ 1568 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1569 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1570 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1571 }; 1572 1573 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1574 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1575 { \ 1576 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1577 } \ 1578 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1579 { \ 1580 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1581 } \ 1582 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1583 { \ 1584 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1585 } 1586 1587 /* Add these declarations here only so that these functions can be 1588 * found by name. Otherwise, they are very hard to locate. */ 1589 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1590 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1591 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1592 EXT4_INODE_BIT_FNS(flag, flags, 0) 1593 1594 /* Add these declarations here only so that these functions can be 1595 * found by name. Otherwise, they are very hard to locate. */ 1596 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1597 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1598 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1599 #if (BITS_PER_LONG < 64) 1600 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1601 1602 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1603 { 1604 (ei)->i_state_flags = 0; 1605 } 1606 #else 1607 EXT4_INODE_BIT_FNS(state, flags, 32) 1608 1609 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1610 { 1611 /* We depend on the fact that callers will set i_flags */ 1612 } 1613 #endif 1614 #else 1615 /* Assume that user mode programs are passing in an ext4fs superblock, not 1616 * a kernel struct super_block. This will allow us to call the feature-test 1617 * macros from user land. */ 1618 #define EXT4_SB(sb) (sb) 1619 #endif 1620 1621 /* 1622 * Returns true if the inode is inode is encrypted 1623 */ 1624 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1625 1626 /* 1627 * Codes for operating systems 1628 */ 1629 #define EXT4_OS_LINUX 0 1630 #define EXT4_OS_HURD 1 1631 #define EXT4_OS_MASIX 2 1632 #define EXT4_OS_FREEBSD 3 1633 #define EXT4_OS_LITES 4 1634 1635 /* 1636 * Revision levels 1637 */ 1638 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1639 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1640 1641 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1642 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1643 1644 #define EXT4_GOOD_OLD_INODE_SIZE 128 1645 1646 /* 1647 * Feature set definitions 1648 */ 1649 1650 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1651 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1652 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1653 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1654 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1655 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1656 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1657 1658 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1659 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1660 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1661 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1662 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1663 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1664 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1665 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1666 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1667 /* 1668 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1669 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1670 * all other data structures' checksums. However, the METADATA_CSUM and 1671 * GDT_CSUM bits are mutually exclusive. 1672 */ 1673 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1674 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1675 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1676 1677 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1678 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1679 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1680 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1681 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1682 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1683 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1684 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1685 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1686 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1687 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1688 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1689 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1690 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1691 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1692 1693 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1694 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1695 { \ 1696 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1697 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1698 } \ 1699 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1700 { \ 1701 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1702 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1703 } \ 1704 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1705 { \ 1706 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1707 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1708 } 1709 1710 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1711 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1712 { \ 1713 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1714 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1715 } \ 1716 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1717 { \ 1718 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1719 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1720 } \ 1721 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1722 { \ 1723 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1724 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1725 } 1726 1727 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1728 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1729 { \ 1730 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1731 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1732 } \ 1733 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1734 { \ 1735 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1736 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1737 } \ 1738 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1739 { \ 1740 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1741 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1742 } 1743 1744 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1745 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1746 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1747 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1748 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1749 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1750 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1751 1752 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1753 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1754 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1755 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1756 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1757 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1758 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1759 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1760 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1761 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1762 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1763 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1764 1765 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1766 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1767 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1768 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1769 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1770 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1771 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1772 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1773 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1774 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1775 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1776 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1777 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1778 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1779 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1780 1781 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1782 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1783 EXT4_FEATURE_INCOMPAT_META_BG) 1784 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1785 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1786 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1787 1788 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1789 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1790 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1791 EXT4_FEATURE_INCOMPAT_META_BG) 1792 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1793 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1794 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1795 1796 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1797 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1798 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1799 EXT4_FEATURE_INCOMPAT_META_BG| \ 1800 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1801 EXT4_FEATURE_INCOMPAT_64BIT| \ 1802 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1803 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1804 EXT4_FEATURE_INCOMPAT_MMP | \ 1805 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1806 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1807 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1808 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1809 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1810 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1811 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1812 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1813 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1814 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1815 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1816 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1817 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1818 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1819 EXT4_FEATURE_RO_COMPAT_PROJECT) 1820 1821 #define EXTN_FEATURE_FUNCS(ver) \ 1822 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1823 { \ 1824 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1825 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1826 } \ 1827 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1828 { \ 1829 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1830 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1831 } \ 1832 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1833 { \ 1834 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1835 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1836 } 1837 1838 EXTN_FEATURE_FUNCS(2) 1839 EXTN_FEATURE_FUNCS(3) 1840 EXTN_FEATURE_FUNCS(4) 1841 1842 static inline bool ext4_has_compat_features(struct super_block *sb) 1843 { 1844 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1845 } 1846 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1847 { 1848 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1849 } 1850 static inline bool ext4_has_incompat_features(struct super_block *sb) 1851 { 1852 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1853 } 1854 1855 /* 1856 * Superblock flags 1857 */ 1858 #define EXT4_FLAGS_RESIZING 0 1859 #define EXT4_FLAGS_SHUTDOWN 1 1860 1861 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1862 { 1863 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1864 } 1865 1866 1867 /* 1868 * Default values for user and/or group using reserved blocks 1869 */ 1870 #define EXT4_DEF_RESUID 0 1871 #define EXT4_DEF_RESGID 0 1872 1873 /* 1874 * Default project ID 1875 */ 1876 #define EXT4_DEF_PROJID 0 1877 1878 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1879 1880 /* 1881 * Default mount options 1882 */ 1883 #define EXT4_DEFM_DEBUG 0x0001 1884 #define EXT4_DEFM_BSDGROUPS 0x0002 1885 #define EXT4_DEFM_XATTR_USER 0x0004 1886 #define EXT4_DEFM_ACL 0x0008 1887 #define EXT4_DEFM_UID16 0x0010 1888 #define EXT4_DEFM_JMODE 0x0060 1889 #define EXT4_DEFM_JMODE_DATA 0x0020 1890 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1891 #define EXT4_DEFM_JMODE_WBACK 0x0060 1892 #define EXT4_DEFM_NOBARRIER 0x0100 1893 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1894 #define EXT4_DEFM_DISCARD 0x0400 1895 #define EXT4_DEFM_NODELALLOC 0x0800 1896 1897 /* 1898 * Default journal batch times 1899 */ 1900 #define EXT4_DEF_MIN_BATCH_TIME 0 1901 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1902 1903 /* 1904 * Minimum number of groups in a flexgroup before we separate out 1905 * directories into the first block group of a flexgroup 1906 */ 1907 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1908 1909 /* 1910 * Structure of a directory entry 1911 */ 1912 #define EXT4_NAME_LEN 255 1913 1914 struct ext4_dir_entry { 1915 __le32 inode; /* Inode number */ 1916 __le16 rec_len; /* Directory entry length */ 1917 __le16 name_len; /* Name length */ 1918 char name[EXT4_NAME_LEN]; /* File name */ 1919 }; 1920 1921 /* 1922 * The new version of the directory entry. Since EXT4 structures are 1923 * stored in intel byte order, and the name_len field could never be 1924 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1925 * file_type field. 1926 */ 1927 struct ext4_dir_entry_2 { 1928 __le32 inode; /* Inode number */ 1929 __le16 rec_len; /* Directory entry length */ 1930 __u8 name_len; /* Name length */ 1931 __u8 file_type; 1932 char name[EXT4_NAME_LEN]; /* File name */ 1933 }; 1934 1935 /* 1936 * This is a bogus directory entry at the end of each leaf block that 1937 * records checksums. 1938 */ 1939 struct ext4_dir_entry_tail { 1940 __le32 det_reserved_zero1; /* Pretend to be unused */ 1941 __le16 det_rec_len; /* 12 */ 1942 __u8 det_reserved_zero2; /* Zero name length */ 1943 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1944 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1945 }; 1946 1947 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1948 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1949 ((blocksize) - \ 1950 sizeof(struct ext4_dir_entry_tail)))) 1951 1952 /* 1953 * Ext4 directory file types. Only the low 3 bits are used. The 1954 * other bits are reserved for now. 1955 */ 1956 #define EXT4_FT_UNKNOWN 0 1957 #define EXT4_FT_REG_FILE 1 1958 #define EXT4_FT_DIR 2 1959 #define EXT4_FT_CHRDEV 3 1960 #define EXT4_FT_BLKDEV 4 1961 #define EXT4_FT_FIFO 5 1962 #define EXT4_FT_SOCK 6 1963 #define EXT4_FT_SYMLINK 7 1964 1965 #define EXT4_FT_MAX 8 1966 1967 #define EXT4_FT_DIR_CSUM 0xDE 1968 1969 /* 1970 * EXT4_DIR_PAD defines the directory entries boundaries 1971 * 1972 * NOTE: It must be a multiple of 4 1973 */ 1974 #define EXT4_DIR_PAD 4 1975 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1976 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1977 ~EXT4_DIR_ROUND) 1978 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1979 1980 /* 1981 * If we ever get support for fs block sizes > page_size, we'll need 1982 * to remove the #if statements in the next two functions... 1983 */ 1984 static inline unsigned int 1985 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1986 { 1987 unsigned len = le16_to_cpu(dlen); 1988 1989 #if (PAGE_SIZE >= 65536) 1990 if (len == EXT4_MAX_REC_LEN || len == 0) 1991 return blocksize; 1992 return (len & 65532) | ((len & 3) << 16); 1993 #else 1994 return len; 1995 #endif 1996 } 1997 1998 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1999 { 2000 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 2001 BUG(); 2002 #if (PAGE_SIZE >= 65536) 2003 if (len < 65536) 2004 return cpu_to_le16(len); 2005 if (len == blocksize) { 2006 if (blocksize == 65536) 2007 return cpu_to_le16(EXT4_MAX_REC_LEN); 2008 else 2009 return cpu_to_le16(0); 2010 } 2011 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2012 #else 2013 return cpu_to_le16(len); 2014 #endif 2015 } 2016 2017 /* 2018 * Hash Tree Directory indexing 2019 * (c) Daniel Phillips, 2001 2020 */ 2021 2022 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2023 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2024 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2025 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2026 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2027 2028 /* Legal values for the dx_root hash_version field: */ 2029 2030 #define DX_HASH_LEGACY 0 2031 #define DX_HASH_HALF_MD4 1 2032 #define DX_HASH_TEA 2 2033 #define DX_HASH_LEGACY_UNSIGNED 3 2034 #define DX_HASH_HALF_MD4_UNSIGNED 4 2035 #define DX_HASH_TEA_UNSIGNED 5 2036 2037 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2038 const void *address, unsigned int length) 2039 { 2040 struct { 2041 struct shash_desc shash; 2042 char ctx[4]; 2043 } desc; 2044 int err; 2045 2046 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2047 2048 desc.shash.tfm = sbi->s_chksum_driver; 2049 desc.shash.flags = 0; 2050 *(u32 *)desc.ctx = crc; 2051 2052 err = crypto_shash_update(&desc.shash, address, length); 2053 BUG_ON(err); 2054 2055 return *(u32 *)desc.ctx; 2056 } 2057 2058 #ifdef __KERNEL__ 2059 2060 /* hash info structure used by the directory hash */ 2061 struct dx_hash_info 2062 { 2063 u32 hash; 2064 u32 minor_hash; 2065 int hash_version; 2066 u32 *seed; 2067 }; 2068 2069 2070 /* 32 and 64 bit signed EOF for dx directories */ 2071 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2072 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2073 2074 2075 /* 2076 * Control parameters used by ext4_htree_next_block 2077 */ 2078 #define HASH_NB_ALWAYS 1 2079 2080 struct ext4_filename { 2081 const struct qstr *usr_fname; 2082 struct fscrypt_str disk_name; 2083 struct dx_hash_info hinfo; 2084 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2085 struct fscrypt_str crypto_buf; 2086 #endif 2087 }; 2088 2089 #define fname_name(p) ((p)->disk_name.name) 2090 #define fname_len(p) ((p)->disk_name.len) 2091 2092 /* 2093 * Describe an inode's exact location on disk and in memory 2094 */ 2095 struct ext4_iloc 2096 { 2097 struct buffer_head *bh; 2098 unsigned long offset; 2099 ext4_group_t block_group; 2100 }; 2101 2102 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2103 { 2104 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2105 } 2106 2107 static inline bool ext4_is_quota_file(struct inode *inode) 2108 { 2109 return IS_NOQUOTA(inode) && 2110 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2111 } 2112 2113 /* 2114 * This structure is stuffed into the struct file's private_data field 2115 * for directories. It is where we put information so that we can do 2116 * readdir operations in hash tree order. 2117 */ 2118 struct dir_private_info { 2119 struct rb_root root; 2120 struct rb_node *curr_node; 2121 struct fname *extra_fname; 2122 loff_t last_pos; 2123 __u32 curr_hash; 2124 __u32 curr_minor_hash; 2125 __u32 next_hash; 2126 }; 2127 2128 /* calculate the first block number of the group */ 2129 static inline ext4_fsblk_t 2130 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2131 { 2132 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2133 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2134 } 2135 2136 /* 2137 * Special error return code only used by dx_probe() and its callers. 2138 */ 2139 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2140 2141 /* htree levels for ext4 */ 2142 #define EXT4_HTREE_LEVEL_COMPAT 2 2143 #define EXT4_HTREE_LEVEL 3 2144 2145 static inline int ext4_dir_htree_level(struct super_block *sb) 2146 { 2147 return ext4_has_feature_largedir(sb) ? 2148 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2149 } 2150 2151 /* 2152 * Timeout and state flag for lazy initialization inode thread. 2153 */ 2154 #define EXT4_DEF_LI_WAIT_MULT 10 2155 #define EXT4_DEF_LI_MAX_START_DELAY 5 2156 #define EXT4_LAZYINIT_QUIT 0x0001 2157 #define EXT4_LAZYINIT_RUNNING 0x0002 2158 2159 /* 2160 * Lazy inode table initialization info 2161 */ 2162 struct ext4_lazy_init { 2163 unsigned long li_state; 2164 struct list_head li_request_list; 2165 struct mutex li_list_mtx; 2166 }; 2167 2168 struct ext4_li_request { 2169 struct super_block *lr_super; 2170 struct ext4_sb_info *lr_sbi; 2171 ext4_group_t lr_next_group; 2172 struct list_head lr_request; 2173 unsigned long lr_next_sched; 2174 unsigned long lr_timeout; 2175 }; 2176 2177 struct ext4_features { 2178 struct kobject f_kobj; 2179 struct completion f_kobj_unregister; 2180 }; 2181 2182 /* 2183 * This structure will be used for multiple mount protection. It will be 2184 * written into the block number saved in the s_mmp_block field in the 2185 * superblock. Programs that check MMP should assume that if 2186 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2187 * to use the filesystem, regardless of how old the timestamp is. 2188 */ 2189 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2190 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2191 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2192 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2193 2194 struct mmp_struct { 2195 __le32 mmp_magic; /* Magic number for MMP */ 2196 __le32 mmp_seq; /* Sequence no. updated periodically */ 2197 2198 /* 2199 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2200 * purposes and do not affect the correctness of the algorithm 2201 */ 2202 __le64 mmp_time; /* Time last updated */ 2203 char mmp_nodename[64]; /* Node which last updated MMP block */ 2204 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2205 2206 /* 2207 * mmp_check_interval is used to verify if the MMP block has been 2208 * updated on the block device. The value is updated based on the 2209 * maximum time to write the MMP block during an update cycle. 2210 */ 2211 __le16 mmp_check_interval; 2212 2213 __le16 mmp_pad1; 2214 __le32 mmp_pad2[226]; 2215 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2216 }; 2217 2218 /* arguments passed to the mmp thread */ 2219 struct mmpd_data { 2220 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2221 struct super_block *sb; /* super block of the fs */ 2222 }; 2223 2224 /* 2225 * Check interval multiplier 2226 * The MMP block is written every update interval and initially checked every 2227 * update interval x the multiplier (the value is then adapted based on the 2228 * write latency). The reason is that writes can be delayed under load and we 2229 * don't want readers to incorrectly assume that the filesystem is no longer 2230 * in use. 2231 */ 2232 #define EXT4_MMP_CHECK_MULT 2UL 2233 2234 /* 2235 * Minimum interval for MMP checking in seconds. 2236 */ 2237 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2238 2239 /* 2240 * Maximum interval for MMP checking in seconds. 2241 */ 2242 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2243 2244 /* 2245 * Function prototypes 2246 */ 2247 2248 /* 2249 * Ok, these declarations are also in <linux/kernel.h> but none of the 2250 * ext4 source programs needs to include it so they are duplicated here. 2251 */ 2252 # define NORET_TYPE /**/ 2253 # define ATTRIB_NORET __attribute__((noreturn)) 2254 # define NORET_AND noreturn, 2255 2256 /* bitmap.c */ 2257 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2258 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2259 struct ext4_group_desc *gdp, 2260 struct buffer_head *bh, int sz); 2261 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2262 struct ext4_group_desc *gdp, 2263 struct buffer_head *bh, int sz); 2264 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2265 struct ext4_group_desc *gdp, 2266 struct buffer_head *bh); 2267 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2268 struct ext4_group_desc *gdp, 2269 struct buffer_head *bh); 2270 2271 /* balloc.c */ 2272 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2273 ext4_fsblk_t blocknr, 2274 ext4_group_t *blockgrpp, 2275 ext4_grpblk_t *offsetp); 2276 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2277 ext4_fsblk_t block); 2278 2279 extern unsigned int ext4_block_group(struct super_block *sb, 2280 ext4_fsblk_t blocknr); 2281 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2282 ext4_fsblk_t blocknr); 2283 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2284 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2285 ext4_group_t group); 2286 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2287 ext4_fsblk_t goal, 2288 unsigned int flags, 2289 unsigned long *count, 2290 int *errp); 2291 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2292 s64 nclusters, unsigned int flags); 2293 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2294 extern void ext4_check_blocks_bitmap(struct super_block *); 2295 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2296 ext4_group_t block_group, 2297 struct buffer_head ** bh); 2298 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2299 2300 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2301 ext4_group_t block_group); 2302 extern int ext4_wait_block_bitmap(struct super_block *sb, 2303 ext4_group_t block_group, 2304 struct buffer_head *bh); 2305 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2306 ext4_group_t block_group); 2307 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2308 ext4_group_t block_group, 2309 struct ext4_group_desc *gdp); 2310 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2311 2312 static inline bool ext4_encrypted_inode(struct inode *inode) 2313 { 2314 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2315 } 2316 2317 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2318 static inline int ext4_fname_setup_filename(struct inode *dir, 2319 const struct qstr *iname, 2320 int lookup, struct ext4_filename *fname) 2321 { 2322 struct fscrypt_name name; 2323 int err; 2324 2325 memset(fname, 0, sizeof(struct ext4_filename)); 2326 2327 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2328 2329 fname->usr_fname = name.usr_fname; 2330 fname->disk_name = name.disk_name; 2331 fname->hinfo.hash = name.hash; 2332 fname->hinfo.minor_hash = name.minor_hash; 2333 fname->crypto_buf = name.crypto_buf; 2334 return err; 2335 } 2336 2337 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2338 { 2339 struct fscrypt_name name; 2340 2341 name.crypto_buf = fname->crypto_buf; 2342 fscrypt_free_filename(&name); 2343 2344 fname->crypto_buf.name = NULL; 2345 fname->usr_fname = NULL; 2346 fname->disk_name.name = NULL; 2347 } 2348 #else 2349 static inline int ext4_fname_setup_filename(struct inode *dir, 2350 const struct qstr *iname, 2351 int lookup, struct ext4_filename *fname) 2352 { 2353 fname->usr_fname = iname; 2354 fname->disk_name.name = (unsigned char *) iname->name; 2355 fname->disk_name.len = iname->len; 2356 return 0; 2357 } 2358 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2359 2360 #endif 2361 2362 /* dir.c */ 2363 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2364 struct file *, 2365 struct ext4_dir_entry_2 *, 2366 struct buffer_head *, char *, int, 2367 unsigned int); 2368 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2369 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2370 (de), (bh), (buf), (size), (offset))) 2371 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2372 __u32 minor_hash, 2373 struct ext4_dir_entry_2 *dirent, 2374 struct fscrypt_str *ent_name); 2375 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2376 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2377 struct buffer_head *bh, 2378 void *buf, int buf_size, 2379 struct ext4_filename *fname, 2380 struct ext4_dir_entry_2 **dest_de); 2381 void ext4_insert_dentry(struct inode *inode, 2382 struct ext4_dir_entry_2 *de, 2383 int buf_size, 2384 struct ext4_filename *fname); 2385 static inline void ext4_update_dx_flag(struct inode *inode) 2386 { 2387 if (!ext4_has_feature_dir_index(inode->i_sb)) 2388 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2389 } 2390 static const unsigned char ext4_filetype_table[] = { 2391 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2392 }; 2393 2394 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2395 { 2396 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2397 return DT_UNKNOWN; 2398 2399 return ext4_filetype_table[filetype]; 2400 } 2401 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2402 void *buf, int buf_size); 2403 2404 /* fsync.c */ 2405 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2406 2407 /* hash.c */ 2408 extern int ext4fs_dirhash(const char *name, int len, struct 2409 dx_hash_info *hinfo); 2410 2411 /* ialloc.c */ 2412 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2413 const struct qstr *qstr, __u32 goal, 2414 uid_t *owner, __u32 i_flags, 2415 int handle_type, unsigned int line_no, 2416 int nblocks); 2417 2418 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2419 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2420 i_flags, 0, 0, 0) 2421 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2422 type, nblocks) \ 2423 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2424 0, (type), __LINE__, (nblocks)) 2425 2426 2427 extern void ext4_free_inode(handle_t *, struct inode *); 2428 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2429 extern unsigned long ext4_count_free_inodes(struct super_block *); 2430 extern unsigned long ext4_count_dirs(struct super_block *); 2431 extern void ext4_check_inodes_bitmap(struct super_block *); 2432 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2433 extern int ext4_init_inode_table(struct super_block *sb, 2434 ext4_group_t group, int barrier); 2435 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2436 2437 /* mballoc.c */ 2438 extern const struct file_operations ext4_seq_mb_groups_fops; 2439 extern long ext4_mb_stats; 2440 extern long ext4_mb_max_to_scan; 2441 extern int ext4_mb_init(struct super_block *); 2442 extern int ext4_mb_release(struct super_block *); 2443 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2444 struct ext4_allocation_request *, int *); 2445 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2446 extern void ext4_discard_preallocations(struct inode *); 2447 extern int __init ext4_init_mballoc(void); 2448 extern void ext4_exit_mballoc(void); 2449 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2450 struct buffer_head *bh, ext4_fsblk_t block, 2451 unsigned long count, int flags); 2452 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2453 ext4_group_t ngroups); 2454 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2455 ext4_group_t i, struct ext4_group_desc *desc); 2456 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2457 ext4_fsblk_t block, unsigned long count); 2458 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2459 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2460 2461 /* inode.c */ 2462 int ext4_inode_is_fast_symlink(struct inode *inode); 2463 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2464 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2465 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2466 bool wait, struct buffer_head **bhs); 2467 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2468 struct buffer_head *bh_result, int create); 2469 int ext4_get_block(struct inode *inode, sector_t iblock, 2470 struct buffer_head *bh_result, int create); 2471 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2472 struct buffer_head *bh_result, int create); 2473 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2474 struct buffer_head *bh, int create); 2475 int ext4_walk_page_buffers(handle_t *handle, 2476 struct buffer_head *head, 2477 unsigned from, 2478 unsigned to, 2479 int *partial, 2480 int (*fn)(handle_t *handle, 2481 struct buffer_head *bh)); 2482 int do_journal_get_write_access(handle_t *handle, 2483 struct buffer_head *bh); 2484 #define FALL_BACK_TO_NONDELALLOC 1 2485 #define CONVERT_INLINE_DATA 2 2486 2487 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2488 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2489 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2490 extern int ext4_setattr(struct dentry *, struct iattr *); 2491 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2492 extern void ext4_evict_inode(struct inode *); 2493 extern void ext4_clear_inode(struct inode *); 2494 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2495 extern int ext4_sync_inode(handle_t *, struct inode *); 2496 extern void ext4_dirty_inode(struct inode *, int); 2497 extern int ext4_change_inode_journal_flag(struct inode *, int); 2498 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2499 extern int ext4_inode_attach_jinode(struct inode *inode); 2500 extern int ext4_can_truncate(struct inode *inode); 2501 extern int ext4_truncate(struct inode *); 2502 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2503 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2504 extern void ext4_set_inode_flags(struct inode *); 2505 extern int ext4_alloc_da_blocks(struct inode *inode); 2506 extern void ext4_set_aops(struct inode *inode); 2507 extern int ext4_writepage_trans_blocks(struct inode *); 2508 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2509 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2510 loff_t lstart, loff_t lend); 2511 extern int ext4_page_mkwrite(struct vm_fault *vmf); 2512 extern int ext4_filemap_fault(struct vm_fault *vmf); 2513 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2514 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2515 extern void ext4_da_update_reserve_space(struct inode *inode, 2516 int used, int quota_claim); 2517 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2518 ext4_fsblk_t pblk, ext4_lblk_t len); 2519 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk, 2520 unsigned int map_len, 2521 struct extent_status *result); 2522 2523 /* indirect.c */ 2524 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2525 struct ext4_map_blocks *map, int flags); 2526 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2527 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2528 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2529 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2530 ext4_lblk_t start, ext4_lblk_t end); 2531 2532 /* ioctl.c */ 2533 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2534 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2535 2536 /* migrate.c */ 2537 extern int ext4_ext_migrate(struct inode *); 2538 extern int ext4_ind_migrate(struct inode *inode); 2539 2540 /* namei.c */ 2541 extern int ext4_dirent_csum_verify(struct inode *inode, 2542 struct ext4_dir_entry *dirent); 2543 extern int ext4_orphan_add(handle_t *, struct inode *); 2544 extern int ext4_orphan_del(handle_t *, struct inode *); 2545 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2546 __u32 start_minor_hash, __u32 *next_hash); 2547 extern int ext4_search_dir(struct buffer_head *bh, 2548 char *search_buf, 2549 int buf_size, 2550 struct inode *dir, 2551 struct ext4_filename *fname, 2552 unsigned int offset, 2553 struct ext4_dir_entry_2 **res_dir); 2554 extern int ext4_generic_delete_entry(handle_t *handle, 2555 struct inode *dir, 2556 struct ext4_dir_entry_2 *de_del, 2557 struct buffer_head *bh, 2558 void *entry_buf, 2559 int buf_size, 2560 int csum_size); 2561 extern bool ext4_empty_dir(struct inode *inode); 2562 2563 /* resize.c */ 2564 extern int ext4_group_add(struct super_block *sb, 2565 struct ext4_new_group_data *input); 2566 extern int ext4_group_extend(struct super_block *sb, 2567 struct ext4_super_block *es, 2568 ext4_fsblk_t n_blocks_count); 2569 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2570 2571 /* super.c */ 2572 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2573 extern int ext4_calculate_overhead(struct super_block *sb); 2574 extern void ext4_superblock_csum_set(struct super_block *sb); 2575 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2576 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2577 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2578 ext4_group_t ngroup); 2579 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2580 char nbuf[16]); 2581 2582 extern __printf(4, 5) 2583 void __ext4_error(struct super_block *, const char *, unsigned int, 2584 const char *, ...); 2585 extern __printf(5, 6) 2586 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2587 const char *, ...); 2588 extern __printf(5, 6) 2589 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2590 const char *, ...); 2591 extern void __ext4_std_error(struct super_block *, const char *, 2592 unsigned int, int); 2593 extern __printf(4, 5) 2594 void __ext4_abort(struct super_block *, const char *, unsigned int, 2595 const char *, ...); 2596 extern __printf(4, 5) 2597 void __ext4_warning(struct super_block *, const char *, unsigned int, 2598 const char *, ...); 2599 extern __printf(4, 5) 2600 void __ext4_warning_inode(const struct inode *inode, const char *function, 2601 unsigned int line, const char *fmt, ...); 2602 extern __printf(3, 4) 2603 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2604 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2605 const char *, unsigned int, const char *); 2606 extern __printf(7, 8) 2607 void __ext4_grp_locked_error(const char *, unsigned int, 2608 struct super_block *, ext4_group_t, 2609 unsigned long, ext4_fsblk_t, 2610 const char *, ...); 2611 2612 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2613 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2614 2615 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2616 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2617 2618 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2619 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2620 2621 #ifdef CONFIG_PRINTK 2622 2623 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2624 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2625 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2626 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2627 #define ext4_error(sb, fmt, ...) \ 2628 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2629 #define ext4_abort(sb, fmt, ...) \ 2630 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2631 #define ext4_warning(sb, fmt, ...) \ 2632 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2633 #define ext4_warning_inode(inode, fmt, ...) \ 2634 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2635 #define ext4_msg(sb, level, fmt, ...) \ 2636 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2637 #define dump_mmp_msg(sb, mmp, msg) \ 2638 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2639 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2640 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2641 fmt, ##__VA_ARGS__) 2642 2643 #else 2644 2645 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2646 do { \ 2647 no_printk(fmt, ##__VA_ARGS__); \ 2648 __ext4_error_inode(inode, "", 0, block, " "); \ 2649 } while (0) 2650 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2651 do { \ 2652 no_printk(fmt, ##__VA_ARGS__); \ 2653 __ext4_error_file(file, "", 0, block, " "); \ 2654 } while (0) 2655 #define ext4_error(sb, fmt, ...) \ 2656 do { \ 2657 no_printk(fmt, ##__VA_ARGS__); \ 2658 __ext4_error(sb, "", 0, " "); \ 2659 } while (0) 2660 #define ext4_abort(sb, fmt, ...) \ 2661 do { \ 2662 no_printk(fmt, ##__VA_ARGS__); \ 2663 __ext4_abort(sb, "", 0, " "); \ 2664 } while (0) 2665 #define ext4_warning(sb, fmt, ...) \ 2666 do { \ 2667 no_printk(fmt, ##__VA_ARGS__); \ 2668 __ext4_warning(sb, "", 0, " "); \ 2669 } while (0) 2670 #define ext4_warning_inode(inode, fmt, ...) \ 2671 do { \ 2672 no_printk(fmt, ##__VA_ARGS__); \ 2673 __ext4_warning_inode(inode, "", 0, " "); \ 2674 } while (0) 2675 #define ext4_msg(sb, level, fmt, ...) \ 2676 do { \ 2677 no_printk(fmt, ##__VA_ARGS__); \ 2678 __ext4_msg(sb, "", " "); \ 2679 } while (0) 2680 #define dump_mmp_msg(sb, mmp, msg) \ 2681 __dump_mmp_msg(sb, mmp, "", 0, "") 2682 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2683 do { \ 2684 no_printk(fmt, ##__VA_ARGS__); \ 2685 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2686 } while (0) 2687 2688 #endif 2689 2690 extern void ext4_update_dynamic_rev(struct super_block *sb); 2691 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2692 __u32 compat); 2693 extern int ext4_update_rocompat_feature(handle_t *handle, 2694 struct super_block *sb, __u32 rocompat); 2695 extern int ext4_update_incompat_feature(handle_t *handle, 2696 struct super_block *sb, __u32 incompat); 2697 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2698 struct ext4_group_desc *bg); 2699 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2700 struct ext4_group_desc *bg); 2701 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2702 struct ext4_group_desc *bg); 2703 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2704 struct ext4_group_desc *bg); 2705 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2706 struct ext4_group_desc *bg); 2707 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2708 struct ext4_group_desc *bg); 2709 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2710 struct ext4_group_desc *bg); 2711 extern void ext4_block_bitmap_set(struct super_block *sb, 2712 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2713 extern void ext4_inode_bitmap_set(struct super_block *sb, 2714 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2715 extern void ext4_inode_table_set(struct super_block *sb, 2716 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2717 extern void ext4_free_group_clusters_set(struct super_block *sb, 2718 struct ext4_group_desc *bg, 2719 __u32 count); 2720 extern void ext4_free_inodes_set(struct super_block *sb, 2721 struct ext4_group_desc *bg, __u32 count); 2722 extern void ext4_used_dirs_set(struct super_block *sb, 2723 struct ext4_group_desc *bg, __u32 count); 2724 extern void ext4_itable_unused_set(struct super_block *sb, 2725 struct ext4_group_desc *bg, __u32 count); 2726 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2727 struct ext4_group_desc *gdp); 2728 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2729 struct ext4_group_desc *gdp); 2730 extern int ext4_register_li_request(struct super_block *sb, 2731 ext4_group_t first_not_zeroed); 2732 2733 static inline int ext4_has_metadata_csum(struct super_block *sb) 2734 { 2735 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2736 !EXT4_SB(sb)->s_chksum_driver); 2737 2738 return ext4_has_feature_metadata_csum(sb) && 2739 (EXT4_SB(sb)->s_chksum_driver != NULL); 2740 } 2741 2742 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2743 { 2744 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2745 } 2746 2747 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2748 { 2749 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2750 le32_to_cpu(es->s_blocks_count_lo); 2751 } 2752 2753 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2754 { 2755 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2756 le32_to_cpu(es->s_r_blocks_count_lo); 2757 } 2758 2759 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2760 { 2761 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2762 le32_to_cpu(es->s_free_blocks_count_lo); 2763 } 2764 2765 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2766 ext4_fsblk_t blk) 2767 { 2768 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2769 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2770 } 2771 2772 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2773 ext4_fsblk_t blk) 2774 { 2775 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2776 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2777 } 2778 2779 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2780 ext4_fsblk_t blk) 2781 { 2782 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2783 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2784 } 2785 2786 static inline loff_t ext4_isize(struct super_block *sb, 2787 struct ext4_inode *raw_inode) 2788 { 2789 if (ext4_has_feature_largedir(sb) || 2790 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2791 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2792 le32_to_cpu(raw_inode->i_size_lo); 2793 2794 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2795 } 2796 2797 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2798 { 2799 raw_inode->i_size_lo = cpu_to_le32(i_size); 2800 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2801 } 2802 2803 static inline 2804 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2805 ext4_group_t group) 2806 { 2807 struct ext4_group_info ***grp_info; 2808 long indexv, indexh; 2809 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2810 grp_info = EXT4_SB(sb)->s_group_info; 2811 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2812 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2813 return grp_info[indexv][indexh]; 2814 } 2815 2816 /* 2817 * Reading s_groups_count requires using smp_rmb() afterwards. See 2818 * the locking protocol documented in the comments of ext4_group_add() 2819 * in resize.c 2820 */ 2821 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2822 { 2823 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2824 2825 smp_rmb(); 2826 return ngroups; 2827 } 2828 2829 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2830 ext4_group_t block_group) 2831 { 2832 return block_group >> sbi->s_log_groups_per_flex; 2833 } 2834 2835 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2836 { 2837 return 1 << sbi->s_log_groups_per_flex; 2838 } 2839 2840 #define ext4_std_error(sb, errno) \ 2841 do { \ 2842 if ((errno)) \ 2843 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2844 } while (0) 2845 2846 #ifdef CONFIG_SMP 2847 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2848 * counters. So we need to make sure we have free clusters more 2849 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2850 */ 2851 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2852 #else 2853 #define EXT4_FREECLUSTERS_WATERMARK 0 2854 #endif 2855 2856 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2857 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2858 { 2859 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2860 !inode_is_locked(inode)); 2861 down_write(&EXT4_I(inode)->i_data_sem); 2862 if (newsize > EXT4_I(inode)->i_disksize) 2863 EXT4_I(inode)->i_disksize = newsize; 2864 up_write(&EXT4_I(inode)->i_data_sem); 2865 } 2866 2867 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2868 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2869 { 2870 int changed = 0; 2871 2872 if (newsize > inode->i_size) { 2873 i_size_write(inode, newsize); 2874 changed = 1; 2875 } 2876 if (newsize > EXT4_I(inode)->i_disksize) { 2877 ext4_update_i_disksize(inode, newsize); 2878 changed |= 2; 2879 } 2880 return changed; 2881 } 2882 2883 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2884 loff_t len); 2885 2886 struct ext4_group_info { 2887 unsigned long bb_state; 2888 struct rb_root bb_free_root; 2889 ext4_grpblk_t bb_first_free; /* first free block */ 2890 ext4_grpblk_t bb_free; /* total free blocks */ 2891 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2892 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2893 struct list_head bb_prealloc_list; 2894 #ifdef DOUBLE_CHECK 2895 void *bb_bitmap; 2896 #endif 2897 struct rw_semaphore alloc_sem; 2898 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2899 * regions, index is order. 2900 * bb_counters[3] = 5 means 2901 * 5 free 8-block regions. */ 2902 }; 2903 2904 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2905 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2906 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2907 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2908 2909 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2910 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2911 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2912 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2913 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2914 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2915 2916 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2917 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2918 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2919 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2920 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2921 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2922 2923 #define EXT4_MAX_CONTENTION 8 2924 #define EXT4_CONTENTION_THRESHOLD 2 2925 2926 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2927 ext4_group_t group) 2928 { 2929 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2930 } 2931 2932 /* 2933 * Returns true if the filesystem is busy enough that attempts to 2934 * access the block group locks has run into contention. 2935 */ 2936 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2937 { 2938 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2939 } 2940 2941 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2942 { 2943 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2944 if (spin_trylock(lock)) 2945 /* 2946 * We're able to grab the lock right away, so drop the 2947 * lock contention counter. 2948 */ 2949 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2950 else { 2951 /* 2952 * The lock is busy, so bump the contention counter, 2953 * and then wait on the spin lock. 2954 */ 2955 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2956 EXT4_MAX_CONTENTION); 2957 spin_lock(lock); 2958 } 2959 } 2960 2961 static inline void ext4_unlock_group(struct super_block *sb, 2962 ext4_group_t group) 2963 { 2964 spin_unlock(ext4_group_lock_ptr(sb, group)); 2965 } 2966 2967 /* 2968 * Block validity checking 2969 */ 2970 #define ext4_check_indirect_blockref(inode, bh) \ 2971 ext4_check_blockref(__func__, __LINE__, inode, \ 2972 (__le32 *)(bh)->b_data, \ 2973 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2974 2975 #define ext4_ind_check_inode(inode) \ 2976 ext4_check_blockref(__func__, __LINE__, inode, \ 2977 EXT4_I(inode)->i_data, \ 2978 EXT4_NDIR_BLOCKS) 2979 2980 /* 2981 * Inodes and files operations 2982 */ 2983 2984 /* dir.c */ 2985 extern const struct file_operations ext4_dir_operations; 2986 2987 /* file.c */ 2988 extern const struct inode_operations ext4_file_inode_operations; 2989 extern const struct file_operations ext4_file_operations; 2990 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2991 2992 /* inline.c */ 2993 extern int ext4_get_max_inline_size(struct inode *inode); 2994 extern int ext4_find_inline_data_nolock(struct inode *inode); 2995 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2996 unsigned int len); 2997 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2998 2999 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 3000 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3001 struct inode *inode, 3002 loff_t pos, unsigned len, 3003 unsigned flags, 3004 struct page **pagep); 3005 extern int ext4_write_inline_data_end(struct inode *inode, 3006 loff_t pos, unsigned len, 3007 unsigned copied, 3008 struct page *page); 3009 extern struct buffer_head * 3010 ext4_journalled_write_inline_data(struct inode *inode, 3011 unsigned len, 3012 struct page *page); 3013 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 3014 struct inode *inode, 3015 loff_t pos, unsigned len, 3016 unsigned flags, 3017 struct page **pagep, 3018 void **fsdata); 3019 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 3020 unsigned len, unsigned copied, 3021 struct page *page); 3022 extern int ext4_try_add_inline_entry(handle_t *handle, 3023 struct ext4_filename *fname, 3024 struct inode *dir, struct inode *inode); 3025 extern int ext4_try_create_inline_dir(handle_t *handle, 3026 struct inode *parent, 3027 struct inode *inode); 3028 extern int ext4_read_inline_dir(struct file *filp, 3029 struct dir_context *ctx, 3030 int *has_inline_data); 3031 extern int htree_inlinedir_to_tree(struct file *dir_file, 3032 struct inode *dir, ext4_lblk_t block, 3033 struct dx_hash_info *hinfo, 3034 __u32 start_hash, __u32 start_minor_hash, 3035 int *has_inline_data); 3036 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3037 struct ext4_filename *fname, 3038 struct ext4_dir_entry_2 **res_dir, 3039 int *has_inline_data); 3040 extern int ext4_delete_inline_entry(handle_t *handle, 3041 struct inode *dir, 3042 struct ext4_dir_entry_2 *de_del, 3043 struct buffer_head *bh, 3044 int *has_inline_data); 3045 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3046 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3047 struct ext4_dir_entry_2 **parent_de, 3048 int *retval); 3049 extern int ext4_inline_data_fiemap(struct inode *inode, 3050 struct fiemap_extent_info *fieinfo, 3051 int *has_inline, __u64 start, __u64 len); 3052 extern int ext4_try_to_evict_inline_data(handle_t *handle, 3053 struct inode *inode, 3054 int needed); 3055 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3056 3057 extern int ext4_convert_inline_data(struct inode *inode); 3058 3059 static inline int ext4_has_inline_data(struct inode *inode) 3060 { 3061 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3062 EXT4_I(inode)->i_inline_off; 3063 } 3064 3065 /* namei.c */ 3066 extern const struct inode_operations ext4_dir_inode_operations; 3067 extern const struct inode_operations ext4_special_inode_operations; 3068 extern struct dentry *ext4_get_parent(struct dentry *child); 3069 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3070 struct ext4_dir_entry_2 *de, 3071 int blocksize, int csum_size, 3072 unsigned int parent_ino, int dotdot_real_len); 3073 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3074 unsigned int blocksize); 3075 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3076 struct inode *inode, 3077 struct buffer_head *bh); 3078 #define S_SHIFT 12 3079 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3080 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3081 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3082 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3083 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3084 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3085 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3086 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3087 }; 3088 3089 static inline void ext4_set_de_type(struct super_block *sb, 3090 struct ext4_dir_entry_2 *de, 3091 umode_t mode) { 3092 if (ext4_has_feature_filetype(sb)) 3093 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3094 } 3095 3096 /* readpages.c */ 3097 extern int ext4_mpage_readpages(struct address_space *mapping, 3098 struct list_head *pages, struct page *page, 3099 unsigned nr_pages); 3100 3101 /* symlink.c */ 3102 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3103 extern const struct inode_operations ext4_symlink_inode_operations; 3104 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3105 3106 /* sysfs.c */ 3107 extern int ext4_register_sysfs(struct super_block *sb); 3108 extern void ext4_unregister_sysfs(struct super_block *sb); 3109 extern int __init ext4_init_sysfs(void); 3110 extern void ext4_exit_sysfs(void); 3111 3112 /* block_validity */ 3113 extern void ext4_release_system_zone(struct super_block *sb); 3114 extern int ext4_setup_system_zone(struct super_block *sb); 3115 extern int __init ext4_init_system_zone(void); 3116 extern void ext4_exit_system_zone(void); 3117 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3118 ext4_fsblk_t start_blk, 3119 unsigned int count); 3120 extern int ext4_check_blockref(const char *, unsigned int, 3121 struct inode *, __le32 *, unsigned int); 3122 3123 /* extents.c */ 3124 struct ext4_ext_path; 3125 struct ext4_extent; 3126 3127 /* 3128 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3129 * __le32. 3130 */ 3131 #define EXT_MAX_BLOCKS 0xffffffff 3132 3133 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3134 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3135 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3136 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3137 struct ext4_map_blocks *map, int flags); 3138 extern int ext4_ext_truncate(handle_t *, struct inode *); 3139 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3140 ext4_lblk_t end); 3141 extern void ext4_ext_init(struct super_block *); 3142 extern void ext4_ext_release(struct super_block *); 3143 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3144 loff_t len); 3145 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3146 loff_t offset, ssize_t len); 3147 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3148 struct ext4_map_blocks *map, int flags); 3149 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3150 ext4_lblk_t lblocks); 3151 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3152 int num, 3153 struct ext4_ext_path *path); 3154 extern int ext4_can_extents_be_merged(struct inode *inode, 3155 struct ext4_extent *ex1, 3156 struct ext4_extent *ex2); 3157 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3158 struct ext4_ext_path **, 3159 struct ext4_extent *, int); 3160 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3161 struct ext4_ext_path **, 3162 int flags); 3163 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3164 extern int ext4_ext_check_inode(struct inode *inode); 3165 extern int ext4_find_delalloc_range(struct inode *inode, 3166 ext4_lblk_t lblk_start, 3167 ext4_lblk_t lblk_end); 3168 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3169 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3170 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3171 __u64 start, __u64 len); 3172 extern int ext4_ext_precache(struct inode *inode); 3173 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3174 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3175 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3176 struct inode *inode2, ext4_lblk_t lblk1, 3177 ext4_lblk_t lblk2, ext4_lblk_t count, 3178 int mark_unwritten,int *err); 3179 3180 /* move_extent.c */ 3181 extern void ext4_double_down_write_data_sem(struct inode *first, 3182 struct inode *second); 3183 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3184 struct inode *donor_inode); 3185 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3186 __u64 start_orig, __u64 start_donor, 3187 __u64 len, __u64 *moved_len); 3188 3189 /* page-io.c */ 3190 extern int __init ext4_init_pageio(void); 3191 extern void ext4_exit_pageio(void); 3192 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3193 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3194 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3195 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3196 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3197 struct writeback_control *wbc); 3198 extern void ext4_end_io_rsv_work(struct work_struct *work); 3199 extern void ext4_io_submit(struct ext4_io_submit *io); 3200 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3201 struct page *page, 3202 int len, 3203 struct writeback_control *wbc, 3204 bool keep_towrite); 3205 3206 /* mmp.c */ 3207 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3208 3209 /* 3210 * Add new method to test whether block and inode bitmaps are properly 3211 * initialized. With uninit_bg reading the block from disk is not enough 3212 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3213 */ 3214 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3215 3216 static inline int bitmap_uptodate(struct buffer_head *bh) 3217 { 3218 return (buffer_uptodate(bh) && 3219 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3220 } 3221 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3222 { 3223 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3224 } 3225 3226 /* 3227 * Disable DIO read nolock optimization, so new dioreaders will be forced 3228 * to grab i_mutex 3229 */ 3230 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 3231 { 3232 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3233 smp_mb(); 3234 } 3235 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 3236 { 3237 smp_mb(); 3238 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3239 } 3240 3241 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3242 3243 /* For ioend & aio unwritten conversion wait queues */ 3244 #define EXT4_WQ_HASH_SZ 37 3245 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3246 EXT4_WQ_HASH_SZ]) 3247 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3248 3249 extern int ext4_resize_begin(struct super_block *sb); 3250 extern void ext4_resize_end(struct super_block *sb); 3251 3252 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3253 struct ext4_io_end *io_end) 3254 { 3255 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3256 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3257 atomic_inc(&EXT4_I(inode)->i_unwritten); 3258 } 3259 } 3260 3261 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3262 { 3263 struct inode *inode = io_end->inode; 3264 3265 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3266 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3267 /* Wake up anyone waiting on unwritten extent conversion */ 3268 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3269 wake_up_all(ext4_ioend_wq(inode)); 3270 } 3271 } 3272 3273 extern const struct iomap_ops ext4_iomap_ops; 3274 3275 #endif /* __KERNEL__ */ 3276 3277 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3278 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3279 3280 #endif /* _EXT4_H */ 3281