1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17 #ifndef _EXT4_H 18 #define _EXT4_H 19 20 #include <linux/types.h> 21 #include <linux/blkdev.h> 22 #include <linux/magic.h> 23 #include <linux/jbd2.h> 24 #include <linux/quota.h> 25 #include <linux/rwsem.h> 26 #include <linux/rbtree.h> 27 #include <linux/seqlock.h> 28 #include <linux/mutex.h> 29 #include <linux/timer.h> 30 #include <linux/version.h> 31 #include <linux/wait.h> 32 #include <linux/sched/signal.h> 33 #include <linux/blockgroup_lock.h> 34 #include <linux/percpu_counter.h> 35 #include <linux/ratelimit.h> 36 #include <crypto/hash.h> 37 #include <linux/falloc.h> 38 #include <linux/percpu-rwsem.h> 39 #ifdef __KERNEL__ 40 #include <linux/compat.h> 41 #endif 42 43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION) 44 #include <linux/fscrypt.h> 45 46 #include <linux/compiler.h> 47 48 /* 49 * The fourth extended filesystem constants/structures 50 */ 51 52 /* 53 * with AGGRESSIVE_CHECK allocator runs consistency checks over 54 * structures. these checks slow things down a lot 55 */ 56 #define AGGRESSIVE_CHECK__ 57 58 /* 59 * with DOUBLE_CHECK defined mballoc creates persistent in-core 60 * bitmaps, maintains and uses them to check for double allocations 61 */ 62 #define DOUBLE_CHECK__ 63 64 /* 65 * Define EXT4FS_DEBUG to produce debug messages 66 */ 67 #undef EXT4FS_DEBUG 68 69 /* 70 * Debug code 71 */ 72 #ifdef EXT4FS_DEBUG 73 #define ext4_debug(f, a...) \ 74 do { \ 75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 76 __FILE__, __LINE__, __func__); \ 77 printk(KERN_DEBUG f, ## a); \ 78 } while (0) 79 #else 80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 81 #endif 82 83 /* 84 * Turn on EXT_DEBUG to get lots of info about extents operations. 85 */ 86 #define EXT_DEBUG__ 87 #ifdef EXT_DEBUG 88 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 89 #else 90 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 91 #endif 92 93 /* data type for block offset of block group */ 94 typedef int ext4_grpblk_t; 95 96 /* data type for filesystem-wide blocks number */ 97 typedef unsigned long long ext4_fsblk_t; 98 99 /* data type for file logical block number */ 100 typedef __u32 ext4_lblk_t; 101 102 /* data type for block group number */ 103 typedef unsigned int ext4_group_t; 104 105 enum SHIFT_DIRECTION { 106 SHIFT_LEFT = 0, 107 SHIFT_RIGHT, 108 }; 109 110 /* 111 * Flags used in mballoc's allocation_context flags field. 112 * 113 * Also used to show what's going on for debugging purposes when the 114 * flag field is exported via the traceport interface 115 */ 116 117 /* prefer goal again. length */ 118 #define EXT4_MB_HINT_MERGE 0x0001 119 /* blocks already reserved */ 120 #define EXT4_MB_HINT_RESERVED 0x0002 121 /* metadata is being allocated */ 122 #define EXT4_MB_HINT_METADATA 0x0004 123 /* first blocks in the file */ 124 #define EXT4_MB_HINT_FIRST 0x0008 125 /* search for the best chunk */ 126 #define EXT4_MB_HINT_BEST 0x0010 127 /* data is being allocated */ 128 #define EXT4_MB_HINT_DATA 0x0020 129 /* don't preallocate (for tails) */ 130 #define EXT4_MB_HINT_NOPREALLOC 0x0040 131 /* allocate for locality group */ 132 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 133 /* allocate goal blocks or none */ 134 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 135 /* goal is meaningful */ 136 #define EXT4_MB_HINT_TRY_GOAL 0x0200 137 /* blocks already pre-reserved by delayed allocation */ 138 #define EXT4_MB_DELALLOC_RESERVED 0x0400 139 /* We are doing stream allocation */ 140 #define EXT4_MB_STREAM_ALLOC 0x0800 141 /* Use reserved root blocks if needed */ 142 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 143 /* Use blocks from reserved pool */ 144 #define EXT4_MB_USE_RESERVED 0x2000 145 146 struct ext4_allocation_request { 147 /* target inode for block we're allocating */ 148 struct inode *inode; 149 /* how many blocks we want to allocate */ 150 unsigned int len; 151 /* logical block in target inode */ 152 ext4_lblk_t logical; 153 /* the closest logical allocated block to the left */ 154 ext4_lblk_t lleft; 155 /* the closest logical allocated block to the right */ 156 ext4_lblk_t lright; 157 /* phys. target (a hint) */ 158 ext4_fsblk_t goal; 159 /* phys. block for the closest logical allocated block to the left */ 160 ext4_fsblk_t pleft; 161 /* phys. block for the closest logical allocated block to the right */ 162 ext4_fsblk_t pright; 163 /* flags. see above EXT4_MB_HINT_* */ 164 unsigned int flags; 165 }; 166 167 /* 168 * Logical to physical block mapping, used by ext4_map_blocks() 169 * 170 * This structure is used to pass requests into ext4_map_blocks() as 171 * well as to store the information returned by ext4_map_blocks(). It 172 * takes less room on the stack than a struct buffer_head. 173 */ 174 #define EXT4_MAP_NEW (1 << BH_New) 175 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 176 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 177 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 178 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 180 181 struct ext4_map_blocks { 182 ext4_fsblk_t m_pblk; 183 ext4_lblk_t m_lblk; 184 unsigned int m_len; 185 unsigned int m_flags; 186 }; 187 188 /* 189 * Flags for ext4_io_end->flags 190 */ 191 #define EXT4_IO_END_UNWRITTEN 0x0001 192 193 /* 194 * For converting unwritten extents on a work queue. 'handle' is used for 195 * buffered writeback. 196 */ 197 typedef struct ext4_io_end { 198 struct list_head list; /* per-file finished IO list */ 199 handle_t *handle; /* handle reserved for extent 200 * conversion */ 201 struct inode *inode; /* file being written to */ 202 struct bio *bio; /* Linked list of completed 203 * bios covering the extent */ 204 unsigned int flag; /* unwritten or not */ 205 atomic_t count; /* reference counter */ 206 loff_t offset; /* offset in the file */ 207 ssize_t size; /* size of the extent */ 208 } ext4_io_end_t; 209 210 struct ext4_io_submit { 211 struct writeback_control *io_wbc; 212 struct bio *io_bio; 213 ext4_io_end_t *io_end; 214 sector_t io_next_block; 215 }; 216 217 /* 218 * Special inodes numbers 219 */ 220 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 221 #define EXT4_ROOT_INO 2 /* Root inode */ 222 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 223 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 224 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 225 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 226 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 227 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 228 229 /* First non-reserved inode for old ext4 filesystems */ 230 #define EXT4_GOOD_OLD_FIRST_INO 11 231 232 /* 233 * Maximal count of links to a file 234 */ 235 #define EXT4_LINK_MAX 65000 236 237 /* 238 * Macro-instructions used to manage several block sizes 239 */ 240 #define EXT4_MIN_BLOCK_SIZE 1024 241 #define EXT4_MAX_BLOCK_SIZE 65536 242 #define EXT4_MIN_BLOCK_LOG_SIZE 10 243 #define EXT4_MAX_BLOCK_LOG_SIZE 16 244 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 245 #ifdef __KERNEL__ 246 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 247 #else 248 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 249 #endif 250 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 251 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 252 EXT4_SB(s)->s_cluster_bits) 253 #ifdef __KERNEL__ 254 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 255 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 256 #else 257 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 258 #endif 259 #ifdef __KERNEL__ 260 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 261 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 262 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 263 #else 264 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 265 EXT4_GOOD_OLD_INODE_SIZE : \ 266 (s)->s_inode_size) 267 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 268 EXT4_GOOD_OLD_FIRST_INO : \ 269 (s)->s_first_ino) 270 #endif 271 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 272 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 273 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 274 blkbits)) 275 276 /* Translate a block number to a cluster number */ 277 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 278 /* Translate a cluster number to a block number */ 279 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 280 /* Translate # of blks to # of clusters */ 281 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 282 (sbi)->s_cluster_bits) 283 /* Mask out the low bits to get the starting block of the cluster */ 284 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 285 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 286 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 287 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 288 /* Get the cluster offset */ 289 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 290 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 291 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 292 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 293 294 /* 295 * Structure of a blocks group descriptor 296 */ 297 struct ext4_group_desc 298 { 299 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 300 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 301 __le32 bg_inode_table_lo; /* Inodes table block */ 302 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 303 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 304 __le16 bg_used_dirs_count_lo; /* Directories count */ 305 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 306 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 307 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 308 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 309 __le16 bg_itable_unused_lo; /* Unused inodes count */ 310 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 311 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 312 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 313 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 314 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 315 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 316 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 317 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 318 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 319 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 320 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 321 __u32 bg_reserved; 322 }; 323 324 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 325 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 326 sizeof(__le16)) 327 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 328 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 329 sizeof(__le16)) 330 331 /* 332 * Structure of a flex block group info 333 */ 334 335 struct flex_groups { 336 atomic64_t free_clusters; 337 atomic_t free_inodes; 338 atomic_t used_dirs; 339 }; 340 341 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 342 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 343 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 344 345 /* 346 * Macro-instructions used to manage group descriptors 347 */ 348 #define EXT4_MIN_DESC_SIZE 32 349 #define EXT4_MIN_DESC_SIZE_64BIT 64 350 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 351 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 352 #ifdef __KERNEL__ 353 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 354 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 356 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 357 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 358 #else 359 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 361 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 362 #endif 363 364 /* 365 * Constants relative to the data blocks 366 */ 367 #define EXT4_NDIR_BLOCKS 12 368 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 369 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 370 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 371 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 372 373 /* 374 * Inode flags 375 */ 376 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 377 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 378 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 379 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 380 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 381 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 382 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 383 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 384 /* Reserved for compression usage... */ 385 #define EXT4_DIRTY_FL 0x00000100 386 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 387 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 388 /* nb: was previously EXT2_ECOMPR_FL */ 389 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 390 /* End compression flags --- maybe not all used */ 391 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 392 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 393 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 394 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 395 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 396 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 397 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 398 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 399 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 400 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 401 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 402 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 404 405 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 406 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 407 408 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 409 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 410 EXT4_IMMUTABLE_FL | \ 411 EXT4_APPEND_FL | \ 412 EXT4_NODUMP_FL | \ 413 EXT4_NOATIME_FL | \ 414 EXT4_PROJINHERIT_FL) 415 416 /* Flags that should be inherited by new inodes from their parent. */ 417 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 418 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 419 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 420 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 421 EXT4_PROJINHERIT_FL) 422 423 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 424 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 425 426 /* Flags that are appropriate for non-directories/regular files. */ 427 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 428 429 /* Mask out flags that are inappropriate for the given type of inode. */ 430 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 431 { 432 if (S_ISDIR(mode)) 433 return flags; 434 else if (S_ISREG(mode)) 435 return flags & EXT4_REG_FLMASK; 436 else 437 return flags & EXT4_OTHER_FLMASK; 438 } 439 440 /* 441 * Inode flags used for atomic set/get 442 */ 443 enum { 444 EXT4_INODE_SECRM = 0, /* Secure deletion */ 445 EXT4_INODE_UNRM = 1, /* Undelete */ 446 EXT4_INODE_COMPR = 2, /* Compress file */ 447 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 448 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 449 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 450 EXT4_INODE_NODUMP = 6, /* do not dump file */ 451 EXT4_INODE_NOATIME = 7, /* do not update atime */ 452 /* Reserved for compression usage... */ 453 EXT4_INODE_DIRTY = 8, 454 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 455 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 456 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 457 /* End compression flags --- maybe not all used */ 458 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 459 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 460 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 461 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 462 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 463 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 464 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 465 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 466 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 467 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 468 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 469 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 470 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 471 }; 472 473 /* 474 * Since it's pretty easy to mix up bit numbers and hex values, we use a 475 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 476 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 477 * any extra space in the compiled kernel image, otherwise, the build will fail. 478 * It's important that these values are the same, since we are using 479 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 480 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 481 * values found in ext2, ext3 and ext4 filesystems, and of course the values 482 * defined in e2fsprogs. 483 * 484 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 485 */ 486 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 487 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 488 489 static inline void ext4_check_flag_values(void) 490 { 491 CHECK_FLAG_VALUE(SECRM); 492 CHECK_FLAG_VALUE(UNRM); 493 CHECK_FLAG_VALUE(COMPR); 494 CHECK_FLAG_VALUE(SYNC); 495 CHECK_FLAG_VALUE(IMMUTABLE); 496 CHECK_FLAG_VALUE(APPEND); 497 CHECK_FLAG_VALUE(NODUMP); 498 CHECK_FLAG_VALUE(NOATIME); 499 CHECK_FLAG_VALUE(DIRTY); 500 CHECK_FLAG_VALUE(COMPRBLK); 501 CHECK_FLAG_VALUE(NOCOMPR); 502 CHECK_FLAG_VALUE(ENCRYPT); 503 CHECK_FLAG_VALUE(INDEX); 504 CHECK_FLAG_VALUE(IMAGIC); 505 CHECK_FLAG_VALUE(JOURNAL_DATA); 506 CHECK_FLAG_VALUE(NOTAIL); 507 CHECK_FLAG_VALUE(DIRSYNC); 508 CHECK_FLAG_VALUE(TOPDIR); 509 CHECK_FLAG_VALUE(HUGE_FILE); 510 CHECK_FLAG_VALUE(EXTENTS); 511 CHECK_FLAG_VALUE(EA_INODE); 512 CHECK_FLAG_VALUE(EOFBLOCKS); 513 CHECK_FLAG_VALUE(INLINE_DATA); 514 CHECK_FLAG_VALUE(PROJINHERIT); 515 CHECK_FLAG_VALUE(RESERVED); 516 } 517 518 /* Used to pass group descriptor data when online resize is done */ 519 struct ext4_new_group_input { 520 __u32 group; /* Group number for this data */ 521 __u64 block_bitmap; /* Absolute block number of block bitmap */ 522 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 523 __u64 inode_table; /* Absolute block number of inode table start */ 524 __u32 blocks_count; /* Total number of blocks in this group */ 525 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 526 __u16 unused; 527 }; 528 529 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 530 struct compat_ext4_new_group_input { 531 u32 group; 532 compat_u64 block_bitmap; 533 compat_u64 inode_bitmap; 534 compat_u64 inode_table; 535 u32 blocks_count; 536 u16 reserved_blocks; 537 u16 unused; 538 }; 539 #endif 540 541 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 542 struct ext4_new_group_data { 543 __u32 group; 544 __u64 block_bitmap; 545 __u64 inode_bitmap; 546 __u64 inode_table; 547 __u32 blocks_count; 548 __u16 reserved_blocks; 549 __u16 mdata_blocks; 550 __u32 free_clusters_count; 551 }; 552 553 /* Indexes used to index group tables in ext4_new_group_data */ 554 enum { 555 BLOCK_BITMAP = 0, /* block bitmap */ 556 INODE_BITMAP, /* inode bitmap */ 557 INODE_TABLE, /* inode tables */ 558 GROUP_TABLE_COUNT, 559 }; 560 561 /* 562 * Flags used by ext4_map_blocks() 563 */ 564 /* Allocate any needed blocks and/or convert an unwritten 565 extent to be an initialized ext4 */ 566 #define EXT4_GET_BLOCKS_CREATE 0x0001 567 /* Request the creation of an unwritten extent */ 568 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 569 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 570 EXT4_GET_BLOCKS_CREATE) 571 /* Caller is from the delayed allocation writeout path 572 * finally doing the actual allocation of delayed blocks */ 573 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 574 /* caller is from the direct IO path, request to creation of an 575 unwritten extents if not allocated, split the unwritten 576 extent if blocks has been preallocated already*/ 577 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 578 #define EXT4_GET_BLOCKS_CONVERT 0x0010 579 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 580 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 581 /* Convert extent to initialized after IO complete */ 582 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 583 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 584 /* Eventual metadata allocation (due to growing extent tree) 585 * should not fail, so try to use reserved blocks for that.*/ 586 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 587 /* Don't normalize allocation size (used for fallocate) */ 588 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 589 /* Request will not result in inode size update (user for fallocate) */ 590 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 591 /* Convert written extents to unwritten */ 592 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 593 /* Write zeros to newly created written extents */ 594 #define EXT4_GET_BLOCKS_ZERO 0x0200 595 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 596 EXT4_GET_BLOCKS_ZERO) 597 /* Caller will submit data before dropping transaction handle. This 598 * allows jbd2 to avoid submitting data before commit. */ 599 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 600 601 /* 602 * The bit position of these flags must not overlap with any of the 603 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 604 * read_extent_tree_block(), ext4_split_extent_at(), 605 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 606 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 607 * caching the extents when reading from the extent tree while a 608 * truncate or punch hole operation is in progress. 609 */ 610 #define EXT4_EX_NOCACHE 0x40000000 611 #define EXT4_EX_FORCE_CACHE 0x20000000 612 613 /* 614 * Flags used by ext4_free_blocks 615 */ 616 #define EXT4_FREE_BLOCKS_METADATA 0x0001 617 #define EXT4_FREE_BLOCKS_FORGET 0x0002 618 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 619 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 620 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 621 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 622 623 /* 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 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 649 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 650 651 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 652 653 /* 654 * Flags for going down operation 655 */ 656 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 657 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 658 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 659 660 661 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 662 /* 663 * ioctl commands in 32 bit emulation 664 */ 665 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 666 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 667 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 668 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 669 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 670 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 671 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 672 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 673 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 674 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 675 #endif 676 677 /* Max physical block we can address w/o extents */ 678 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 679 680 /* Max logical block we can support */ 681 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF 682 683 /* 684 * Structure of an inode on the disk 685 */ 686 struct ext4_inode { 687 __le16 i_mode; /* File mode */ 688 __le16 i_uid; /* Low 16 bits of Owner Uid */ 689 __le32 i_size_lo; /* Size in bytes */ 690 __le32 i_atime; /* Access time */ 691 __le32 i_ctime; /* Inode Change time */ 692 __le32 i_mtime; /* Modification time */ 693 __le32 i_dtime; /* Deletion Time */ 694 __le16 i_gid; /* Low 16 bits of Group Id */ 695 __le16 i_links_count; /* Links count */ 696 __le32 i_blocks_lo; /* Blocks count */ 697 __le32 i_flags; /* File flags */ 698 union { 699 struct { 700 __le32 l_i_version; 701 } linux1; 702 struct { 703 __u32 h_i_translator; 704 } hurd1; 705 struct { 706 __u32 m_i_reserved1; 707 } masix1; 708 } osd1; /* OS dependent 1 */ 709 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 710 __le32 i_generation; /* File version (for NFS) */ 711 __le32 i_file_acl_lo; /* File ACL */ 712 __le32 i_size_high; 713 __le32 i_obso_faddr; /* Obsoleted fragment address */ 714 union { 715 struct { 716 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 717 __le16 l_i_file_acl_high; 718 __le16 l_i_uid_high; /* these 2 fields */ 719 __le16 l_i_gid_high; /* were reserved2[0] */ 720 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 721 __le16 l_i_reserved; 722 } linux2; 723 struct { 724 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 725 __u16 h_i_mode_high; 726 __u16 h_i_uid_high; 727 __u16 h_i_gid_high; 728 __u32 h_i_author; 729 } hurd2; 730 struct { 731 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 732 __le16 m_i_file_acl_high; 733 __u32 m_i_reserved2[2]; 734 } masix2; 735 } osd2; /* OS dependent 2 */ 736 __le16 i_extra_isize; 737 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 738 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 739 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 740 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 741 __le32 i_crtime; /* File Creation time */ 742 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 743 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 744 __le32 i_projid; /* Project ID */ 745 }; 746 747 struct move_extent { 748 __u32 reserved; /* should be zero */ 749 __u32 donor_fd; /* donor file descriptor */ 750 __u64 orig_start; /* logical start offset in block for orig */ 751 __u64 donor_start; /* logical start offset in block for donor */ 752 __u64 len; /* block length to be moved */ 753 __u64 moved_len; /* moved block length */ 754 }; 755 756 #define EXT4_EPOCH_BITS 2 757 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 758 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 759 760 /* 761 * Extended fields will fit into an inode if the filesystem was formatted 762 * with large inodes (-I 256 or larger) and there are not currently any EAs 763 * consuming all of the available space. For new inodes we always reserve 764 * enough space for the kernel's known extended fields, but for inodes 765 * created with an old kernel this might not have been the case. None of 766 * the extended inode fields is critical for correct filesystem operation. 767 * This macro checks if a certain field fits in the inode. Note that 768 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 769 */ 770 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 771 ((offsetof(typeof(*ext4_inode), field) + \ 772 sizeof((ext4_inode)->field)) \ 773 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 774 (einode)->i_extra_isize)) \ 775 776 /* 777 * We use an encoding that preserves the times for extra epoch "00": 778 * 779 * extra msb of adjust for signed 780 * epoch 32-bit 32-bit tv_sec to 781 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 782 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 783 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 784 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 785 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 786 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 787 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 788 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 789 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 790 * 791 * Note that previous versions of the kernel on 64-bit systems would 792 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 793 * 1970. e2fsck will correct this, assuming that it is run on the 794 * affected filesystem before 2242. 795 */ 796 797 static inline __le32 ext4_encode_extra_time(struct timespec64 *time) 798 { 799 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK; 800 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 801 } 802 803 static inline void ext4_decode_extra_time(struct timespec64 *time, 804 __le32 extra) 805 { 806 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) { 807 808 #if 1 809 /* Handle legacy encoding of pre-1970 dates with epoch 810 * bits 1,1. (This backwards compatibility may be removed 811 * at the discretion of the ext4 developers.) 812 */ 813 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 814 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 815 extra_bits = 0; 816 time->tv_sec += extra_bits << 32; 817 #else 818 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 819 #endif 820 } 821 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 822 } 823 824 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 825 do { \ 826 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 827 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\ 828 (raw_inode)->xtime ## _extra = \ 829 ext4_encode_extra_time(&(inode)->xtime); \ 830 } \ 831 } while (0) 832 833 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 834 do { \ 835 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 836 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 837 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 838 (raw_inode)->xtime ## _extra = \ 839 ext4_encode_extra_time(&(einode)->xtime); \ 840 } while (0) 841 842 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 843 do { \ 844 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 845 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 846 ext4_decode_extra_time(&(inode)->xtime, \ 847 raw_inode->xtime ## _extra); \ 848 } \ 849 else \ 850 (inode)->xtime.tv_nsec = 0; \ 851 } while (0) 852 853 854 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 855 do { \ 856 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 857 (einode)->xtime.tv_sec = \ 858 (signed)le32_to_cpu((raw_inode)->xtime); \ 859 else \ 860 (einode)->xtime.tv_sec = 0; \ 861 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 862 ext4_decode_extra_time(&(einode)->xtime, \ 863 raw_inode->xtime ## _extra); \ 864 else \ 865 (einode)->xtime.tv_nsec = 0; \ 866 } while (0) 867 868 #define i_disk_version osd1.linux1.l_i_version 869 870 #if defined(__KERNEL__) || defined(__linux__) 871 #define i_reserved1 osd1.linux1.l_i_reserved1 872 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 873 #define i_blocks_high osd2.linux2.l_i_blocks_high 874 #define i_uid_low i_uid 875 #define i_gid_low i_gid 876 #define i_uid_high osd2.linux2.l_i_uid_high 877 #define i_gid_high osd2.linux2.l_i_gid_high 878 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 879 880 #elif defined(__GNU__) 881 882 #define i_translator osd1.hurd1.h_i_translator 883 #define i_uid_high osd2.hurd2.h_i_uid_high 884 #define i_gid_high osd2.hurd2.h_i_gid_high 885 #define i_author osd2.hurd2.h_i_author 886 887 #elif defined(__masix__) 888 889 #define i_reserved1 osd1.masix1.m_i_reserved1 890 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 891 #define i_reserved2 osd2.masix2.m_i_reserved2 892 893 #endif /* defined(__KERNEL__) || defined(__linux__) */ 894 895 #include "extents_status.h" 896 897 /* 898 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 899 * 900 * These are needed to avoid lockdep false positives when we need to 901 * allocate blocks to the quota inode during ext4_map_blocks(), while 902 * holding i_data_sem for a normal (non-quota) inode. Since we don't 903 * do quota tracking for the quota inode, this avoids deadlock (as 904 * well as infinite recursion, since it isn't turtles all the way 905 * down...) 906 * 907 * I_DATA_SEM_NORMAL - Used for most inodes 908 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 909 * where the second inode has larger inode number 910 * than the first 911 * I_DATA_SEM_QUOTA - Used for quota inodes only 912 */ 913 enum { 914 I_DATA_SEM_NORMAL = 0, 915 I_DATA_SEM_OTHER, 916 I_DATA_SEM_QUOTA, 917 }; 918 919 920 /* 921 * fourth extended file system inode data in memory 922 */ 923 struct ext4_inode_info { 924 __le32 i_data[15]; /* unconverted */ 925 __u32 i_dtime; 926 ext4_fsblk_t i_file_acl; 927 928 /* 929 * i_block_group is the number of the block group which contains 930 * this file's inode. Constant across the lifetime of the inode, 931 * it is used for making block allocation decisions - we try to 932 * place a file's data blocks near its inode block, and new inodes 933 * near to their parent directory's inode. 934 */ 935 ext4_group_t i_block_group; 936 ext4_lblk_t i_dir_start_lookup; 937 #if (BITS_PER_LONG < 64) 938 unsigned long i_state_flags; /* Dynamic state flags */ 939 #endif 940 unsigned long i_flags; 941 942 /* 943 * Extended attributes can be read independently of the main file 944 * data. Taking i_mutex even when reading would cause contention 945 * between readers of EAs and writers of regular file data, so 946 * instead we synchronize on xattr_sem when reading or changing 947 * EAs. 948 */ 949 struct rw_semaphore xattr_sem; 950 951 struct list_head i_orphan; /* unlinked but open inodes */ 952 953 /* 954 * i_disksize keeps track of what the inode size is ON DISK, not 955 * in memory. During truncate, i_size is set to the new size by 956 * the VFS prior to calling ext4_truncate(), but the filesystem won't 957 * set i_disksize to 0 until the truncate is actually under way. 958 * 959 * The intent is that i_disksize always represents the blocks which 960 * are used by this file. This allows recovery to restart truncate 961 * on orphans if we crash during truncate. We actually write i_disksize 962 * into the on-disk inode when writing inodes out, instead of i_size. 963 * 964 * The only time when i_disksize and i_size may be different is when 965 * a truncate is in progress. The only things which change i_disksize 966 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 967 */ 968 loff_t i_disksize; 969 970 /* 971 * i_data_sem is for serialising ext4_truncate() against 972 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 973 * data tree are chopped off during truncate. We can't do that in 974 * ext4 because whenever we perform intermediate commits during 975 * truncate, the inode and all the metadata blocks *must* be in a 976 * consistent state which allows truncation of the orphans to restart 977 * during recovery. Hence we must fix the get_block-vs-truncate race 978 * by other means, so we have i_data_sem. 979 */ 980 struct rw_semaphore i_data_sem; 981 /* 982 * i_mmap_sem is for serializing page faults with truncate / punch hole 983 * operations. We have to make sure that new page cannot be faulted in 984 * a section of the inode that is being punched. We cannot easily use 985 * i_data_sem for this since we need protection for the whole punch 986 * operation and i_data_sem ranks below transaction start so we have 987 * to occasionally drop it. 988 */ 989 struct rw_semaphore i_mmap_sem; 990 struct inode vfs_inode; 991 struct jbd2_inode *jinode; 992 993 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 994 995 /* 996 * File creation time. Its function is same as that of 997 * struct timespec64 i_{a,c,m}time in the generic inode. 998 */ 999 struct timespec64 i_crtime; 1000 1001 /* mballoc */ 1002 struct list_head i_prealloc_list; 1003 spinlock_t i_prealloc_lock; 1004 1005 /* extents status tree */ 1006 struct ext4_es_tree i_es_tree; 1007 rwlock_t i_es_lock; 1008 struct list_head i_es_list; 1009 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1010 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1011 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1012 extents to shrink. Protected by 1013 i_es_lock */ 1014 1015 /* ialloc */ 1016 ext4_group_t i_last_alloc_group; 1017 1018 /* allocation reservation info for delalloc */ 1019 /* In case of bigalloc, this refer to clusters rather than blocks */ 1020 unsigned int i_reserved_data_blocks; 1021 ext4_lblk_t i_da_metadata_calc_last_lblock; 1022 int i_da_metadata_calc_len; 1023 1024 /* on-disk additional length */ 1025 __u16 i_extra_isize; 1026 1027 /* Indicate the inline data space. */ 1028 u16 i_inline_off; 1029 u16 i_inline_size; 1030 1031 #ifdef CONFIG_QUOTA 1032 /* quota space reservation, managed internally by quota code */ 1033 qsize_t i_reserved_quota; 1034 #endif 1035 1036 /* Lock protecting lists below */ 1037 spinlock_t i_completed_io_lock; 1038 /* 1039 * Completed IOs that need unwritten extents handling and have 1040 * transaction reserved 1041 */ 1042 struct list_head i_rsv_conversion_list; 1043 struct work_struct i_rsv_conversion_work; 1044 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1045 1046 spinlock_t i_block_reservation_lock; 1047 1048 /* 1049 * Transactions that contain inode's metadata needed to complete 1050 * fsync and fdatasync, respectively. 1051 */ 1052 tid_t i_sync_tid; 1053 tid_t i_datasync_tid; 1054 1055 #ifdef CONFIG_QUOTA 1056 struct dquot *i_dquot[MAXQUOTAS]; 1057 #endif 1058 1059 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1060 __u32 i_csum_seed; 1061 1062 kprojid_t i_projid; 1063 }; 1064 1065 /* 1066 * File system states 1067 */ 1068 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1069 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1070 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1071 1072 /* 1073 * Misc. filesystem flags 1074 */ 1075 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1076 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1077 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1078 1079 /* 1080 * Mount flags set via mount options or defaults 1081 */ 1082 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1083 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1084 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1085 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1086 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1087 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1088 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1089 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1090 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1091 #ifdef CONFIG_FS_DAX 1092 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1093 #else 1094 #define EXT4_MOUNT_DAX 0 1095 #endif 1096 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1097 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1098 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1099 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1100 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1101 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1102 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1103 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1104 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1105 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1106 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1107 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1108 * enable enforcement for hidden 1109 * quota files */ 1110 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1111 * enforcement for hidden quota 1112 * files */ 1113 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1114 * enforcement */ 1115 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1116 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1117 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1118 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1119 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1120 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1121 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1122 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1123 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1124 1125 /* 1126 * Mount flags set either automatically (could not be set by mount option) 1127 * based on per file system feature or property or in special cases such as 1128 * distinguishing between explicit mount option definition and default. 1129 */ 1130 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1131 specified delalloc */ 1132 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1133 size of blocksize * 8 1134 blocks */ 1135 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1136 file systems */ 1137 1138 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1139 specified journal checksum */ 1140 1141 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1142 ~EXT4_MOUNT_##opt 1143 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1144 EXT4_MOUNT_##opt 1145 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1146 EXT4_MOUNT_##opt) 1147 1148 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1149 ~EXT4_MOUNT2_##opt 1150 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1151 EXT4_MOUNT2_##opt 1152 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1153 EXT4_MOUNT2_##opt) 1154 1155 #define ext4_test_and_set_bit __test_and_set_bit_le 1156 #define ext4_set_bit __set_bit_le 1157 #define ext4_set_bit_atomic ext2_set_bit_atomic 1158 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1159 #define ext4_clear_bit __clear_bit_le 1160 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1161 #define ext4_test_bit test_bit_le 1162 #define ext4_find_next_zero_bit find_next_zero_bit_le 1163 #define ext4_find_next_bit find_next_bit_le 1164 1165 extern void ext4_set_bits(void *bm, int cur, int len); 1166 1167 /* 1168 * Maximal mount counts between two filesystem checks 1169 */ 1170 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1171 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1172 1173 /* 1174 * Behaviour when detecting errors 1175 */ 1176 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1177 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1178 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1179 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1180 1181 /* Metadata checksum algorithm codes */ 1182 #define EXT4_CRC32C_CHKSUM 1 1183 1184 /* 1185 * Structure of the super block 1186 */ 1187 struct ext4_super_block { 1188 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1189 __le32 s_blocks_count_lo; /* Blocks count */ 1190 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1191 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1192 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1193 __le32 s_first_data_block; /* First Data Block */ 1194 __le32 s_log_block_size; /* Block size */ 1195 __le32 s_log_cluster_size; /* Allocation cluster size */ 1196 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1197 __le32 s_clusters_per_group; /* # Clusters per group */ 1198 __le32 s_inodes_per_group; /* # Inodes per group */ 1199 __le32 s_mtime; /* Mount time */ 1200 /*30*/ __le32 s_wtime; /* Write time */ 1201 __le16 s_mnt_count; /* Mount count */ 1202 __le16 s_max_mnt_count; /* Maximal mount count */ 1203 __le16 s_magic; /* Magic signature */ 1204 __le16 s_state; /* File system state */ 1205 __le16 s_errors; /* Behaviour when detecting errors */ 1206 __le16 s_minor_rev_level; /* minor revision level */ 1207 /*40*/ __le32 s_lastcheck; /* time of last check */ 1208 __le32 s_checkinterval; /* max. time between checks */ 1209 __le32 s_creator_os; /* OS */ 1210 __le32 s_rev_level; /* Revision level */ 1211 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1212 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1213 /* 1214 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1215 * 1216 * Note: the difference between the compatible feature set and 1217 * the incompatible feature set is that if there is a bit set 1218 * in the incompatible feature set that the kernel doesn't 1219 * know about, it should refuse to mount the filesystem. 1220 * 1221 * e2fsck's requirements are more strict; if it doesn't know 1222 * about a feature in either the compatible or incompatible 1223 * feature set, it must abort and not try to meddle with 1224 * things it doesn't understand... 1225 */ 1226 __le32 s_first_ino; /* First non-reserved inode */ 1227 __le16 s_inode_size; /* size of inode structure */ 1228 __le16 s_block_group_nr; /* block group # of this superblock */ 1229 __le32 s_feature_compat; /* compatible feature set */ 1230 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1231 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1232 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1233 /*78*/ char s_volume_name[16]; /* volume name */ 1234 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1235 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1236 /* 1237 * Performance hints. Directory preallocation should only 1238 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1239 */ 1240 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1241 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1242 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1243 /* 1244 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1245 */ 1246 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1247 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1248 __le32 s_journal_dev; /* device number of journal file */ 1249 __le32 s_last_orphan; /* start of list of inodes to delete */ 1250 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1251 __u8 s_def_hash_version; /* Default hash version to use */ 1252 __u8 s_jnl_backup_type; 1253 __le16 s_desc_size; /* size of group descriptor */ 1254 /*100*/ __le32 s_default_mount_opts; 1255 __le32 s_first_meta_bg; /* First metablock block group */ 1256 __le32 s_mkfs_time; /* When the filesystem was created */ 1257 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1258 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1259 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1260 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1261 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1262 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1263 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1264 __le32 s_flags; /* Miscellaneous flags */ 1265 __le16 s_raid_stride; /* RAID stride */ 1266 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1267 __le64 s_mmp_block; /* Block for multi-mount protection */ 1268 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1269 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1270 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1271 __u8 s_encryption_level; /* versioning level for encryption */ 1272 __u8 s_reserved_pad; /* Padding to next 32bits */ 1273 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1274 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1275 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1276 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1277 snapshot's future use */ 1278 __le32 s_snapshot_list; /* inode number of the head of the 1279 on-disk snapshot list */ 1280 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1281 __le32 s_error_count; /* number of fs errors */ 1282 __le32 s_first_error_time; /* first time an error happened */ 1283 __le32 s_first_error_ino; /* inode involved in first error */ 1284 __le64 s_first_error_block; /* block involved of first error */ 1285 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1286 __le32 s_first_error_line; /* line number where error happened */ 1287 __le32 s_last_error_time; /* most recent time of an error */ 1288 __le32 s_last_error_ino; /* inode involved in last error */ 1289 __le32 s_last_error_line; /* line number where error happened */ 1290 __le64 s_last_error_block; /* block involved of last error */ 1291 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1292 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1293 __u8 s_mount_opts[64]; 1294 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1295 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1296 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1297 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1298 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1299 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1300 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1301 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1302 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1303 __u8 s_wtime_hi; 1304 __u8 s_mtime_hi; 1305 __u8 s_mkfs_time_hi; 1306 __u8 s_lastcheck_hi; 1307 __u8 s_first_error_time_hi; 1308 __u8 s_last_error_time_hi; 1309 __u8 s_pad[2]; 1310 __le32 s_reserved[96]; /* Padding to the end of the block */ 1311 __le32 s_checksum; /* crc32c(superblock) */ 1312 }; 1313 1314 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1315 1316 #ifdef __KERNEL__ 1317 1318 /* 1319 * run-time mount flags 1320 */ 1321 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1322 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1323 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1324 1325 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1326 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1327 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1328 #else 1329 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1330 #endif 1331 1332 /* Number of quota types we support */ 1333 #define EXT4_MAXQUOTAS 3 1334 1335 /* 1336 * fourth extended-fs super-block data in memory 1337 */ 1338 struct ext4_sb_info { 1339 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1340 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1341 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1342 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1343 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1344 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1345 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1346 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1347 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1348 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1349 unsigned long s_overhead; /* # of fs overhead clusters */ 1350 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1351 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1352 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1353 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1354 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1355 struct buffer_head **s_group_desc; 1356 unsigned int s_mount_opt; 1357 unsigned int s_mount_opt2; 1358 unsigned int s_mount_flags; 1359 unsigned int s_def_mount_opt; 1360 ext4_fsblk_t s_sb_block; 1361 atomic64_t s_resv_clusters; 1362 kuid_t s_resuid; 1363 kgid_t s_resgid; 1364 unsigned short s_mount_state; 1365 unsigned short s_pad; 1366 int s_addr_per_block_bits; 1367 int s_desc_per_block_bits; 1368 int s_inode_size; 1369 int s_first_ino; 1370 unsigned int s_inode_readahead_blks; 1371 unsigned int s_inode_goal; 1372 u32 s_hash_seed[4]; 1373 int s_def_hash_version; 1374 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1375 struct percpu_counter s_freeclusters_counter; 1376 struct percpu_counter s_freeinodes_counter; 1377 struct percpu_counter s_dirs_counter; 1378 struct percpu_counter s_dirtyclusters_counter; 1379 struct blockgroup_lock *s_blockgroup_lock; 1380 struct proc_dir_entry *s_proc; 1381 struct kobject s_kobj; 1382 struct completion s_kobj_unregister; 1383 struct super_block *s_sb; 1384 1385 /* Journaling */ 1386 struct journal_s *s_journal; 1387 struct list_head s_orphan; 1388 struct mutex s_orphan_lock; 1389 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1390 unsigned long s_commit_interval; 1391 u32 s_max_batch_time; 1392 u32 s_min_batch_time; 1393 struct block_device *journal_bdev; 1394 #ifdef CONFIG_QUOTA 1395 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1396 int s_jquota_fmt; /* Format of quota to use */ 1397 #endif 1398 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1399 struct rb_root system_blks; 1400 1401 #ifdef EXTENTS_STATS 1402 /* ext4 extents stats */ 1403 unsigned long s_ext_min; 1404 unsigned long s_ext_max; 1405 unsigned long s_depth_max; 1406 spinlock_t s_ext_stats_lock; 1407 unsigned long s_ext_blocks; 1408 unsigned long s_ext_extents; 1409 #endif 1410 1411 /* for buddy allocator */ 1412 struct ext4_group_info ***s_group_info; 1413 struct inode *s_buddy_cache; 1414 spinlock_t s_md_lock; 1415 unsigned short *s_mb_offsets; 1416 unsigned int *s_mb_maxs; 1417 unsigned int s_group_info_size; 1418 unsigned int s_mb_free_pending; 1419 struct list_head s_freed_data_list; /* List of blocks to be freed 1420 after commit completed */ 1421 1422 /* tunables */ 1423 unsigned long s_stripe; 1424 unsigned int s_mb_stream_request; 1425 unsigned int s_mb_max_to_scan; 1426 unsigned int s_mb_min_to_scan; 1427 unsigned int s_mb_stats; 1428 unsigned int s_mb_order2_reqs; 1429 unsigned int s_mb_group_prealloc; 1430 unsigned int s_max_dir_size_kb; 1431 /* where last allocation was done - for stream allocation */ 1432 unsigned long s_mb_last_group; 1433 unsigned long s_mb_last_start; 1434 1435 /* stats for buddy allocator */ 1436 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1437 atomic_t s_bal_success; /* we found long enough chunks */ 1438 atomic_t s_bal_allocated; /* in blocks */ 1439 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1440 atomic_t s_bal_goals; /* goal hits */ 1441 atomic_t s_bal_breaks; /* too long searches */ 1442 atomic_t s_bal_2orders; /* 2^order hits */ 1443 spinlock_t s_bal_lock; 1444 unsigned long s_mb_buddies_generated; 1445 unsigned long long s_mb_generation_time; 1446 atomic_t s_mb_lost_chunks; 1447 atomic_t s_mb_preallocated; 1448 atomic_t s_mb_discarded; 1449 atomic_t s_lock_busy; 1450 1451 /* locality groups */ 1452 struct ext4_locality_group __percpu *s_locality_groups; 1453 1454 /* for write statistics */ 1455 unsigned long s_sectors_written_start; 1456 u64 s_kbytes_written; 1457 1458 /* the size of zero-out chunk */ 1459 unsigned int s_extent_max_zeroout_kb; 1460 1461 unsigned int s_log_groups_per_flex; 1462 struct flex_groups *s_flex_groups; 1463 ext4_group_t s_flex_groups_allocated; 1464 1465 /* workqueue for reserved extent conversions (buffered io) */ 1466 struct workqueue_struct *rsv_conversion_wq; 1467 1468 /* timer for periodic error stats printing */ 1469 struct timer_list s_err_report; 1470 1471 /* Lazy inode table initialization info */ 1472 struct ext4_li_request *s_li_request; 1473 /* Wait multiplier for lazy initialization thread */ 1474 unsigned int s_li_wait_mult; 1475 1476 /* Kernel thread for multiple mount protection */ 1477 struct task_struct *s_mmp_tsk; 1478 1479 /* record the last minlen when FITRIM is called. */ 1480 atomic_t s_last_trim_minblks; 1481 1482 /* Reference to checksum algorithm driver via cryptoapi */ 1483 struct crypto_shash *s_chksum_driver; 1484 1485 /* Precomputed FS UUID checksum for seeding other checksums */ 1486 __u32 s_csum_seed; 1487 1488 /* Reclaim extents from extent status tree */ 1489 struct shrinker s_es_shrinker; 1490 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1491 long s_es_nr_inode; 1492 struct ext4_es_stats s_es_stats; 1493 struct mb_cache *s_ea_block_cache; 1494 struct mb_cache *s_ea_inode_cache; 1495 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1496 1497 /* Ratelimit ext4 messages. */ 1498 struct ratelimit_state s_err_ratelimit_state; 1499 struct ratelimit_state s_warning_ratelimit_state; 1500 struct ratelimit_state s_msg_ratelimit_state; 1501 1502 /* Barrier between changing inodes' journal flags and writepages ops. */ 1503 struct percpu_rw_semaphore s_journal_flag_rwsem; 1504 struct dax_device *s_daxdev; 1505 }; 1506 1507 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1508 { 1509 return sb->s_fs_info; 1510 } 1511 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1512 { 1513 return container_of(inode, struct ext4_inode_info, vfs_inode); 1514 } 1515 1516 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1517 { 1518 return ino == EXT4_ROOT_INO || 1519 (ino >= EXT4_FIRST_INO(sb) && 1520 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1521 } 1522 1523 /* 1524 * Inode dynamic state flags 1525 */ 1526 enum { 1527 EXT4_STATE_JDATA, /* journaled data exists */ 1528 EXT4_STATE_NEW, /* inode is newly created */ 1529 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1530 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1531 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1532 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1533 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1534 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1535 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1536 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1537 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1538 }; 1539 1540 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1541 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1542 { \ 1543 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1544 } \ 1545 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1546 { \ 1547 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1548 } \ 1549 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1550 { \ 1551 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1552 } 1553 1554 /* Add these declarations here only so that these functions can be 1555 * found by name. Otherwise, they are very hard to locate. */ 1556 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1557 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1558 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1559 EXT4_INODE_BIT_FNS(flag, flags, 0) 1560 1561 /* Add these declarations here only so that these functions can be 1562 * found by name. Otherwise, they are very hard to locate. */ 1563 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1564 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1565 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1566 #if (BITS_PER_LONG < 64) 1567 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1568 1569 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1570 { 1571 (ei)->i_state_flags = 0; 1572 } 1573 #else 1574 EXT4_INODE_BIT_FNS(state, flags, 32) 1575 1576 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1577 { 1578 /* We depend on the fact that callers will set i_flags */ 1579 } 1580 #endif 1581 #else 1582 /* Assume that user mode programs are passing in an ext4fs superblock, not 1583 * a kernel struct super_block. This will allow us to call the feature-test 1584 * macros from user land. */ 1585 #define EXT4_SB(sb) (sb) 1586 #endif 1587 1588 /* 1589 * Returns true if the inode is inode is encrypted 1590 */ 1591 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1592 1593 /* 1594 * Codes for operating systems 1595 */ 1596 #define EXT4_OS_LINUX 0 1597 #define EXT4_OS_HURD 1 1598 #define EXT4_OS_MASIX 2 1599 #define EXT4_OS_FREEBSD 3 1600 #define EXT4_OS_LITES 4 1601 1602 /* 1603 * Revision levels 1604 */ 1605 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1606 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1607 1608 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1609 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1610 1611 #define EXT4_GOOD_OLD_INODE_SIZE 128 1612 1613 /* 1614 * Feature set definitions 1615 */ 1616 1617 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1618 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1619 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1620 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1621 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1622 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1623 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1624 1625 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1626 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1627 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1628 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1629 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1630 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1631 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1632 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1633 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1634 /* 1635 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1636 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1637 * all other data structures' checksums. However, the METADATA_CSUM and 1638 * GDT_CSUM bits are mutually exclusive. 1639 */ 1640 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1641 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1642 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1643 1644 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1645 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1646 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1647 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1648 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1649 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1650 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1651 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1652 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1653 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1654 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1655 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1656 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1657 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1658 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1659 1660 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1661 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1662 { \ 1663 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1664 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1665 } \ 1666 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1667 { \ 1668 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1669 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1670 } \ 1671 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1672 { \ 1673 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1674 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1675 } 1676 1677 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1678 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1679 { \ 1680 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1681 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1682 } \ 1683 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1684 { \ 1685 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1686 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1687 } \ 1688 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1689 { \ 1690 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1691 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1692 } 1693 1694 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1695 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1696 { \ 1697 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1698 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1699 } \ 1700 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1701 { \ 1702 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1703 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1704 } \ 1705 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1706 { \ 1707 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1708 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1709 } 1710 1711 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1712 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1713 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1714 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1715 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1716 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1717 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1718 1719 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1720 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1721 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1722 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1723 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1724 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1725 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1726 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1727 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1728 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1729 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1730 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1731 1732 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1733 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1734 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1735 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1736 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1737 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1738 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1739 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1740 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1741 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1742 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1743 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1744 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1745 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1746 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1747 1748 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1749 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1750 EXT4_FEATURE_INCOMPAT_META_BG) 1751 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1752 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1753 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1754 1755 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1756 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1757 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1758 EXT4_FEATURE_INCOMPAT_META_BG) 1759 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1760 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1761 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1762 1763 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1764 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1765 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1766 EXT4_FEATURE_INCOMPAT_META_BG| \ 1767 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1768 EXT4_FEATURE_INCOMPAT_64BIT| \ 1769 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1770 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1771 EXT4_FEATURE_INCOMPAT_MMP | \ 1772 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1773 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1774 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1775 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1776 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1777 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1778 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1779 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1780 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1781 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1782 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1783 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1784 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1785 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1786 EXT4_FEATURE_RO_COMPAT_PROJECT) 1787 1788 #define EXTN_FEATURE_FUNCS(ver) \ 1789 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1790 { \ 1791 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1792 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1793 } \ 1794 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1795 { \ 1796 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1797 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1798 } \ 1799 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1800 { \ 1801 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1802 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1803 } 1804 1805 EXTN_FEATURE_FUNCS(2) 1806 EXTN_FEATURE_FUNCS(3) 1807 EXTN_FEATURE_FUNCS(4) 1808 1809 static inline bool ext4_has_compat_features(struct super_block *sb) 1810 { 1811 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1812 } 1813 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1814 { 1815 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1816 } 1817 static inline bool ext4_has_incompat_features(struct super_block *sb) 1818 { 1819 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1820 } 1821 1822 /* 1823 * Superblock flags 1824 */ 1825 #define EXT4_FLAGS_RESIZING 0 1826 #define EXT4_FLAGS_SHUTDOWN 1 1827 1828 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1829 { 1830 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1831 } 1832 1833 1834 /* 1835 * Default values for user and/or group using reserved blocks 1836 */ 1837 #define EXT4_DEF_RESUID 0 1838 #define EXT4_DEF_RESGID 0 1839 1840 /* 1841 * Default project ID 1842 */ 1843 #define EXT4_DEF_PROJID 0 1844 1845 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1846 1847 /* 1848 * Default mount options 1849 */ 1850 #define EXT4_DEFM_DEBUG 0x0001 1851 #define EXT4_DEFM_BSDGROUPS 0x0002 1852 #define EXT4_DEFM_XATTR_USER 0x0004 1853 #define EXT4_DEFM_ACL 0x0008 1854 #define EXT4_DEFM_UID16 0x0010 1855 #define EXT4_DEFM_JMODE 0x0060 1856 #define EXT4_DEFM_JMODE_DATA 0x0020 1857 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1858 #define EXT4_DEFM_JMODE_WBACK 0x0060 1859 #define EXT4_DEFM_NOBARRIER 0x0100 1860 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1861 #define EXT4_DEFM_DISCARD 0x0400 1862 #define EXT4_DEFM_NODELALLOC 0x0800 1863 1864 /* 1865 * Default journal batch times 1866 */ 1867 #define EXT4_DEF_MIN_BATCH_TIME 0 1868 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1869 1870 /* 1871 * Minimum number of groups in a flexgroup before we separate out 1872 * directories into the first block group of a flexgroup 1873 */ 1874 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1875 1876 /* 1877 * Structure of a directory entry 1878 */ 1879 #define EXT4_NAME_LEN 255 1880 1881 struct ext4_dir_entry { 1882 __le32 inode; /* Inode number */ 1883 __le16 rec_len; /* Directory entry length */ 1884 __le16 name_len; /* Name length */ 1885 char name[EXT4_NAME_LEN]; /* File name */ 1886 }; 1887 1888 /* 1889 * The new version of the directory entry. Since EXT4 structures are 1890 * stored in intel byte order, and the name_len field could never be 1891 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1892 * file_type field. 1893 */ 1894 struct ext4_dir_entry_2 { 1895 __le32 inode; /* Inode number */ 1896 __le16 rec_len; /* Directory entry length */ 1897 __u8 name_len; /* Name length */ 1898 __u8 file_type; 1899 char name[EXT4_NAME_LEN]; /* File name */ 1900 }; 1901 1902 /* 1903 * This is a bogus directory entry at the end of each leaf block that 1904 * records checksums. 1905 */ 1906 struct ext4_dir_entry_tail { 1907 __le32 det_reserved_zero1; /* Pretend to be unused */ 1908 __le16 det_rec_len; /* 12 */ 1909 __u8 det_reserved_zero2; /* Zero name length */ 1910 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1911 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1912 }; 1913 1914 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1915 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1916 ((blocksize) - \ 1917 sizeof(struct ext4_dir_entry_tail)))) 1918 1919 /* 1920 * Ext4 directory file types. Only the low 3 bits are used. The 1921 * other bits are reserved for now. 1922 */ 1923 #define EXT4_FT_UNKNOWN 0 1924 #define EXT4_FT_REG_FILE 1 1925 #define EXT4_FT_DIR 2 1926 #define EXT4_FT_CHRDEV 3 1927 #define EXT4_FT_BLKDEV 4 1928 #define EXT4_FT_FIFO 5 1929 #define EXT4_FT_SOCK 6 1930 #define EXT4_FT_SYMLINK 7 1931 1932 #define EXT4_FT_MAX 8 1933 1934 #define EXT4_FT_DIR_CSUM 0xDE 1935 1936 /* 1937 * EXT4_DIR_PAD defines the directory entries boundaries 1938 * 1939 * NOTE: It must be a multiple of 4 1940 */ 1941 #define EXT4_DIR_PAD 4 1942 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1943 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1944 ~EXT4_DIR_ROUND) 1945 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1946 1947 /* 1948 * If we ever get support for fs block sizes > page_size, we'll need 1949 * to remove the #if statements in the next two functions... 1950 */ 1951 static inline unsigned int 1952 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1953 { 1954 unsigned len = le16_to_cpu(dlen); 1955 1956 #if (PAGE_SIZE >= 65536) 1957 if (len == EXT4_MAX_REC_LEN || len == 0) 1958 return blocksize; 1959 return (len & 65532) | ((len & 3) << 16); 1960 #else 1961 return len; 1962 #endif 1963 } 1964 1965 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1966 { 1967 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1968 BUG(); 1969 #if (PAGE_SIZE >= 65536) 1970 if (len < 65536) 1971 return cpu_to_le16(len); 1972 if (len == blocksize) { 1973 if (blocksize == 65536) 1974 return cpu_to_le16(EXT4_MAX_REC_LEN); 1975 else 1976 return cpu_to_le16(0); 1977 } 1978 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1979 #else 1980 return cpu_to_le16(len); 1981 #endif 1982 } 1983 1984 /* 1985 * Hash Tree Directory indexing 1986 * (c) Daniel Phillips, 2001 1987 */ 1988 1989 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1990 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1991 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 1992 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 1993 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1994 1995 /* Legal values for the dx_root hash_version field: */ 1996 1997 #define DX_HASH_LEGACY 0 1998 #define DX_HASH_HALF_MD4 1 1999 #define DX_HASH_TEA 2 2000 #define DX_HASH_LEGACY_UNSIGNED 3 2001 #define DX_HASH_HALF_MD4_UNSIGNED 4 2002 #define DX_HASH_TEA_UNSIGNED 5 2003 2004 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2005 const void *address, unsigned int length) 2006 { 2007 struct { 2008 struct shash_desc shash; 2009 char ctx[4]; 2010 } desc; 2011 2012 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2013 2014 desc.shash.tfm = sbi->s_chksum_driver; 2015 desc.shash.flags = 0; 2016 *(u32 *)desc.ctx = crc; 2017 2018 BUG_ON(crypto_shash_update(&desc.shash, address, length)); 2019 2020 return *(u32 *)desc.ctx; 2021 } 2022 2023 #ifdef __KERNEL__ 2024 2025 /* hash info structure used by the directory hash */ 2026 struct dx_hash_info 2027 { 2028 u32 hash; 2029 u32 minor_hash; 2030 int hash_version; 2031 u32 *seed; 2032 }; 2033 2034 2035 /* 32 and 64 bit signed EOF for dx directories */ 2036 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2037 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2038 2039 2040 /* 2041 * Control parameters used by ext4_htree_next_block 2042 */ 2043 #define HASH_NB_ALWAYS 1 2044 2045 struct ext4_filename { 2046 const struct qstr *usr_fname; 2047 struct fscrypt_str disk_name; 2048 struct dx_hash_info hinfo; 2049 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2050 struct fscrypt_str crypto_buf; 2051 #endif 2052 }; 2053 2054 #define fname_name(p) ((p)->disk_name.name) 2055 #define fname_len(p) ((p)->disk_name.len) 2056 2057 /* 2058 * Describe an inode's exact location on disk and in memory 2059 */ 2060 struct ext4_iloc 2061 { 2062 struct buffer_head *bh; 2063 unsigned long offset; 2064 ext4_group_t block_group; 2065 }; 2066 2067 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2068 { 2069 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2070 } 2071 2072 static inline bool ext4_is_quota_file(struct inode *inode) 2073 { 2074 return IS_NOQUOTA(inode) && 2075 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2076 } 2077 2078 /* 2079 * This structure is stuffed into the struct file's private_data field 2080 * for directories. It is where we put information so that we can do 2081 * readdir operations in hash tree order. 2082 */ 2083 struct dir_private_info { 2084 struct rb_root root; 2085 struct rb_node *curr_node; 2086 struct fname *extra_fname; 2087 loff_t last_pos; 2088 __u32 curr_hash; 2089 __u32 curr_minor_hash; 2090 __u32 next_hash; 2091 }; 2092 2093 /* calculate the first block number of the group */ 2094 static inline ext4_fsblk_t 2095 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2096 { 2097 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2098 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2099 } 2100 2101 /* 2102 * Special error return code only used by dx_probe() and its callers. 2103 */ 2104 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2105 2106 /* htree levels for ext4 */ 2107 #define EXT4_HTREE_LEVEL_COMPAT 2 2108 #define EXT4_HTREE_LEVEL 3 2109 2110 static inline int ext4_dir_htree_level(struct super_block *sb) 2111 { 2112 return ext4_has_feature_largedir(sb) ? 2113 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2114 } 2115 2116 /* 2117 * Timeout and state flag for lazy initialization inode thread. 2118 */ 2119 #define EXT4_DEF_LI_WAIT_MULT 10 2120 #define EXT4_DEF_LI_MAX_START_DELAY 5 2121 #define EXT4_LAZYINIT_QUIT 0x0001 2122 #define EXT4_LAZYINIT_RUNNING 0x0002 2123 2124 /* 2125 * Lazy inode table initialization info 2126 */ 2127 struct ext4_lazy_init { 2128 unsigned long li_state; 2129 struct list_head li_request_list; 2130 struct mutex li_list_mtx; 2131 }; 2132 2133 struct ext4_li_request { 2134 struct super_block *lr_super; 2135 struct ext4_sb_info *lr_sbi; 2136 ext4_group_t lr_next_group; 2137 struct list_head lr_request; 2138 unsigned long lr_next_sched; 2139 unsigned long lr_timeout; 2140 }; 2141 2142 struct ext4_features { 2143 struct kobject f_kobj; 2144 struct completion f_kobj_unregister; 2145 }; 2146 2147 /* 2148 * This structure will be used for multiple mount protection. It will be 2149 * written into the block number saved in the s_mmp_block field in the 2150 * superblock. Programs that check MMP should assume that if 2151 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2152 * to use the filesystem, regardless of how old the timestamp is. 2153 */ 2154 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2155 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2156 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2157 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2158 2159 struct mmp_struct { 2160 __le32 mmp_magic; /* Magic number for MMP */ 2161 __le32 mmp_seq; /* Sequence no. updated periodically */ 2162 2163 /* 2164 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2165 * purposes and do not affect the correctness of the algorithm 2166 */ 2167 __le64 mmp_time; /* Time last updated */ 2168 char mmp_nodename[64]; /* Node which last updated MMP block */ 2169 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2170 2171 /* 2172 * mmp_check_interval is used to verify if the MMP block has been 2173 * updated on the block device. The value is updated based on the 2174 * maximum time to write the MMP block during an update cycle. 2175 */ 2176 __le16 mmp_check_interval; 2177 2178 __le16 mmp_pad1; 2179 __le32 mmp_pad2[226]; 2180 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2181 }; 2182 2183 /* arguments passed to the mmp thread */ 2184 struct mmpd_data { 2185 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2186 struct super_block *sb; /* super block of the fs */ 2187 }; 2188 2189 /* 2190 * Check interval multiplier 2191 * The MMP block is written every update interval and initially checked every 2192 * update interval x the multiplier (the value is then adapted based on the 2193 * write latency). The reason is that writes can be delayed under load and we 2194 * don't want readers to incorrectly assume that the filesystem is no longer 2195 * in use. 2196 */ 2197 #define EXT4_MMP_CHECK_MULT 2UL 2198 2199 /* 2200 * Minimum interval for MMP checking in seconds. 2201 */ 2202 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2203 2204 /* 2205 * Maximum interval for MMP checking in seconds. 2206 */ 2207 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2208 2209 /* 2210 * Function prototypes 2211 */ 2212 2213 /* 2214 * Ok, these declarations are also in <linux/kernel.h> but none of the 2215 * ext4 source programs needs to include it so they are duplicated here. 2216 */ 2217 # define NORET_TYPE /**/ 2218 # define ATTRIB_NORET __attribute__((noreturn)) 2219 # define NORET_AND noreturn, 2220 2221 /* bitmap.c */ 2222 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2223 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2224 struct ext4_group_desc *gdp, 2225 struct buffer_head *bh, int sz); 2226 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2227 struct ext4_group_desc *gdp, 2228 struct buffer_head *bh, int sz); 2229 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2230 struct ext4_group_desc *gdp, 2231 struct buffer_head *bh); 2232 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2233 struct ext4_group_desc *gdp, 2234 struct buffer_head *bh); 2235 2236 /* balloc.c */ 2237 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2238 ext4_fsblk_t blocknr, 2239 ext4_group_t *blockgrpp, 2240 ext4_grpblk_t *offsetp); 2241 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2242 ext4_fsblk_t block); 2243 2244 extern unsigned int ext4_block_group(struct super_block *sb, 2245 ext4_fsblk_t blocknr); 2246 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2247 ext4_fsblk_t blocknr); 2248 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2249 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2250 ext4_group_t group); 2251 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2252 ext4_fsblk_t goal, 2253 unsigned int flags, 2254 unsigned long *count, 2255 int *errp); 2256 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2257 s64 nclusters, unsigned int flags); 2258 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2259 extern void ext4_check_blocks_bitmap(struct super_block *); 2260 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2261 ext4_group_t block_group, 2262 struct buffer_head ** bh); 2263 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2264 2265 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2266 ext4_group_t block_group); 2267 extern int ext4_wait_block_bitmap(struct super_block *sb, 2268 ext4_group_t block_group, 2269 struct buffer_head *bh); 2270 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2271 ext4_group_t block_group); 2272 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2273 ext4_group_t block_group, 2274 struct ext4_group_desc *gdp); 2275 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2276 2277 static inline bool ext4_encrypted_inode(struct inode *inode) 2278 { 2279 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2280 } 2281 2282 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2283 static inline int ext4_fname_setup_filename(struct inode *dir, 2284 const struct qstr *iname, 2285 int lookup, struct ext4_filename *fname) 2286 { 2287 struct fscrypt_name name; 2288 int err; 2289 2290 memset(fname, 0, sizeof(struct ext4_filename)); 2291 2292 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2293 2294 fname->usr_fname = name.usr_fname; 2295 fname->disk_name = name.disk_name; 2296 fname->hinfo.hash = name.hash; 2297 fname->hinfo.minor_hash = name.minor_hash; 2298 fname->crypto_buf = name.crypto_buf; 2299 return err; 2300 } 2301 2302 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2303 { 2304 struct fscrypt_name name; 2305 2306 name.crypto_buf = fname->crypto_buf; 2307 fscrypt_free_filename(&name); 2308 2309 fname->crypto_buf.name = NULL; 2310 fname->usr_fname = NULL; 2311 fname->disk_name.name = NULL; 2312 } 2313 #else 2314 static inline int ext4_fname_setup_filename(struct inode *dir, 2315 const struct qstr *iname, 2316 int lookup, struct ext4_filename *fname) 2317 { 2318 fname->usr_fname = iname; 2319 fname->disk_name.name = (unsigned char *) iname->name; 2320 fname->disk_name.len = iname->len; 2321 return 0; 2322 } 2323 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2324 2325 #endif 2326 2327 /* dir.c */ 2328 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2329 struct file *, 2330 struct ext4_dir_entry_2 *, 2331 struct buffer_head *, char *, int, 2332 unsigned int); 2333 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2334 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2335 (de), (bh), (buf), (size), (offset))) 2336 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2337 __u32 minor_hash, 2338 struct ext4_dir_entry_2 *dirent, 2339 struct fscrypt_str *ent_name); 2340 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2341 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2342 struct buffer_head *bh, 2343 void *buf, int buf_size, 2344 struct ext4_filename *fname, 2345 struct ext4_dir_entry_2 **dest_de); 2346 void ext4_insert_dentry(struct inode *inode, 2347 struct ext4_dir_entry_2 *de, 2348 int buf_size, 2349 struct ext4_filename *fname); 2350 static inline void ext4_update_dx_flag(struct inode *inode) 2351 { 2352 if (!ext4_has_feature_dir_index(inode->i_sb)) 2353 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2354 } 2355 static const unsigned char ext4_filetype_table[] = { 2356 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2357 }; 2358 2359 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2360 { 2361 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2362 return DT_UNKNOWN; 2363 2364 return ext4_filetype_table[filetype]; 2365 } 2366 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2367 void *buf, int buf_size); 2368 2369 /* fsync.c */ 2370 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2371 2372 /* hash.c */ 2373 extern int ext4fs_dirhash(const char *name, int len, struct 2374 dx_hash_info *hinfo); 2375 2376 /* ialloc.c */ 2377 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2378 const struct qstr *qstr, __u32 goal, 2379 uid_t *owner, __u32 i_flags, 2380 int handle_type, unsigned int line_no, 2381 int nblocks); 2382 2383 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2384 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2385 i_flags, 0, 0, 0) 2386 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2387 type, nblocks) \ 2388 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2389 0, (type), __LINE__, (nblocks)) 2390 2391 2392 extern void ext4_free_inode(handle_t *, struct inode *); 2393 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2394 extern unsigned long ext4_count_free_inodes(struct super_block *); 2395 extern unsigned long ext4_count_dirs(struct super_block *); 2396 extern void ext4_check_inodes_bitmap(struct super_block *); 2397 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2398 extern int ext4_init_inode_table(struct super_block *sb, 2399 ext4_group_t group, int barrier); 2400 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2401 2402 /* mballoc.c */ 2403 extern const struct seq_operations ext4_mb_seq_groups_ops; 2404 extern long ext4_mb_stats; 2405 extern long ext4_mb_max_to_scan; 2406 extern int ext4_mb_init(struct super_block *); 2407 extern int ext4_mb_release(struct super_block *); 2408 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2409 struct ext4_allocation_request *, int *); 2410 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2411 extern void ext4_discard_preallocations(struct inode *); 2412 extern int __init ext4_init_mballoc(void); 2413 extern void ext4_exit_mballoc(void); 2414 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2415 struct buffer_head *bh, ext4_fsblk_t block, 2416 unsigned long count, int flags); 2417 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2418 ext4_group_t ngroups); 2419 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2420 ext4_group_t i, struct ext4_group_desc *desc); 2421 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2422 ext4_fsblk_t block, unsigned long count); 2423 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2424 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2425 2426 /* inode.c */ 2427 int ext4_inode_is_fast_symlink(struct inode *inode); 2428 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2429 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2430 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2431 bool wait, struct buffer_head **bhs); 2432 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2433 struct buffer_head *bh_result, int create); 2434 int ext4_get_block(struct inode *inode, sector_t iblock, 2435 struct buffer_head *bh_result, int create); 2436 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2437 struct buffer_head *bh_result, int create); 2438 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2439 struct buffer_head *bh, int create); 2440 int ext4_walk_page_buffers(handle_t *handle, 2441 struct buffer_head *head, 2442 unsigned from, 2443 unsigned to, 2444 int *partial, 2445 int (*fn)(handle_t *handle, 2446 struct buffer_head *bh)); 2447 int do_journal_get_write_access(handle_t *handle, 2448 struct buffer_head *bh); 2449 #define FALL_BACK_TO_NONDELALLOC 1 2450 #define CONVERT_INLINE_DATA 2 2451 2452 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2453 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2454 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2455 extern int ext4_setattr(struct dentry *, struct iattr *); 2456 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2457 extern void ext4_evict_inode(struct inode *); 2458 extern void ext4_clear_inode(struct inode *); 2459 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2460 extern int ext4_sync_inode(handle_t *, struct inode *); 2461 extern void ext4_dirty_inode(struct inode *, int); 2462 extern int ext4_change_inode_journal_flag(struct inode *, int); 2463 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2464 extern int ext4_inode_attach_jinode(struct inode *inode); 2465 extern int ext4_can_truncate(struct inode *inode); 2466 extern int ext4_truncate(struct inode *); 2467 extern int ext4_break_layouts(struct inode *); 2468 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2469 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2470 extern void ext4_set_inode_flags(struct inode *); 2471 extern int ext4_alloc_da_blocks(struct inode *inode); 2472 extern void ext4_set_aops(struct inode *inode); 2473 extern int ext4_writepage_trans_blocks(struct inode *); 2474 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2475 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2476 loff_t lstart, loff_t lend); 2477 extern int ext4_page_mkwrite(struct vm_fault *vmf); 2478 extern int ext4_filemap_fault(struct vm_fault *vmf); 2479 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2480 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2481 extern void ext4_da_update_reserve_space(struct inode *inode, 2482 int used, int quota_claim); 2483 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2484 ext4_fsblk_t pblk, ext4_lblk_t len); 2485 2486 /* indirect.c */ 2487 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2488 struct ext4_map_blocks *map, int flags); 2489 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2490 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2491 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2492 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2493 ext4_lblk_t start, ext4_lblk_t end); 2494 2495 /* ioctl.c */ 2496 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2497 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2498 2499 /* migrate.c */ 2500 extern int ext4_ext_migrate(struct inode *); 2501 extern int ext4_ind_migrate(struct inode *inode); 2502 2503 /* namei.c */ 2504 extern int ext4_dirent_csum_verify(struct inode *inode, 2505 struct ext4_dir_entry *dirent); 2506 extern int ext4_orphan_add(handle_t *, struct inode *); 2507 extern int ext4_orphan_del(handle_t *, struct inode *); 2508 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2509 __u32 start_minor_hash, __u32 *next_hash); 2510 extern int ext4_search_dir(struct buffer_head *bh, 2511 char *search_buf, 2512 int buf_size, 2513 struct inode *dir, 2514 struct ext4_filename *fname, 2515 unsigned int offset, 2516 struct ext4_dir_entry_2 **res_dir); 2517 extern int ext4_generic_delete_entry(handle_t *handle, 2518 struct inode *dir, 2519 struct ext4_dir_entry_2 *de_del, 2520 struct buffer_head *bh, 2521 void *entry_buf, 2522 int buf_size, 2523 int csum_size); 2524 extern bool ext4_empty_dir(struct inode *inode); 2525 2526 /* resize.c */ 2527 extern int ext4_group_add(struct super_block *sb, 2528 struct ext4_new_group_data *input); 2529 extern int ext4_group_extend(struct super_block *sb, 2530 struct ext4_super_block *es, 2531 ext4_fsblk_t n_blocks_count); 2532 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2533 2534 /* super.c */ 2535 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2536 extern int ext4_calculate_overhead(struct super_block *sb); 2537 extern void ext4_superblock_csum_set(struct super_block *sb); 2538 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2539 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2540 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2541 ext4_group_t ngroup); 2542 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2543 char nbuf[16]); 2544 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 2545 ext4_group_t block_group, 2546 unsigned int flags); 2547 2548 extern __printf(4, 5) 2549 void __ext4_error(struct super_block *, const char *, unsigned int, 2550 const char *, ...); 2551 extern __printf(5, 6) 2552 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2553 const char *, ...); 2554 extern __printf(5, 6) 2555 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2556 const char *, ...); 2557 extern void __ext4_std_error(struct super_block *, const char *, 2558 unsigned int, int); 2559 extern __printf(4, 5) 2560 void __ext4_abort(struct super_block *, const char *, unsigned int, 2561 const char *, ...); 2562 extern __printf(4, 5) 2563 void __ext4_warning(struct super_block *, const char *, unsigned int, 2564 const char *, ...); 2565 extern __printf(4, 5) 2566 void __ext4_warning_inode(const struct inode *inode, const char *function, 2567 unsigned int line, const char *fmt, ...); 2568 extern __printf(3, 4) 2569 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2570 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2571 const char *, unsigned int, const char *); 2572 extern __printf(7, 8) 2573 void __ext4_grp_locked_error(const char *, unsigned int, 2574 struct super_block *, ext4_group_t, 2575 unsigned long, ext4_fsblk_t, 2576 const char *, ...); 2577 2578 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2579 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2580 2581 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2582 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2583 2584 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2585 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2586 2587 #ifdef CONFIG_PRINTK 2588 2589 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2590 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2591 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2592 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2593 #define ext4_error(sb, fmt, ...) \ 2594 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2595 #define ext4_abort(sb, fmt, ...) \ 2596 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2597 #define ext4_warning(sb, fmt, ...) \ 2598 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2599 #define ext4_warning_inode(inode, fmt, ...) \ 2600 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2601 #define ext4_msg(sb, level, fmt, ...) \ 2602 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2603 #define dump_mmp_msg(sb, mmp, msg) \ 2604 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2605 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2606 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2607 fmt, ##__VA_ARGS__) 2608 2609 #else 2610 2611 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2612 do { \ 2613 no_printk(fmt, ##__VA_ARGS__); \ 2614 __ext4_error_inode(inode, "", 0, block, " "); \ 2615 } while (0) 2616 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2617 do { \ 2618 no_printk(fmt, ##__VA_ARGS__); \ 2619 __ext4_error_file(file, "", 0, block, " "); \ 2620 } while (0) 2621 #define ext4_error(sb, fmt, ...) \ 2622 do { \ 2623 no_printk(fmt, ##__VA_ARGS__); \ 2624 __ext4_error(sb, "", 0, " "); \ 2625 } while (0) 2626 #define ext4_abort(sb, fmt, ...) \ 2627 do { \ 2628 no_printk(fmt, ##__VA_ARGS__); \ 2629 __ext4_abort(sb, "", 0, " "); \ 2630 } while (0) 2631 #define ext4_warning(sb, fmt, ...) \ 2632 do { \ 2633 no_printk(fmt, ##__VA_ARGS__); \ 2634 __ext4_warning(sb, "", 0, " "); \ 2635 } while (0) 2636 #define ext4_warning_inode(inode, fmt, ...) \ 2637 do { \ 2638 no_printk(fmt, ##__VA_ARGS__); \ 2639 __ext4_warning_inode(inode, "", 0, " "); \ 2640 } while (0) 2641 #define ext4_msg(sb, level, fmt, ...) \ 2642 do { \ 2643 no_printk(fmt, ##__VA_ARGS__); \ 2644 __ext4_msg(sb, "", " "); \ 2645 } while (0) 2646 #define dump_mmp_msg(sb, mmp, msg) \ 2647 __dump_mmp_msg(sb, mmp, "", 0, "") 2648 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2649 do { \ 2650 no_printk(fmt, ##__VA_ARGS__); \ 2651 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2652 } while (0) 2653 2654 #endif 2655 2656 extern void ext4_update_dynamic_rev(struct super_block *sb); 2657 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2658 __u32 compat); 2659 extern int ext4_update_rocompat_feature(handle_t *handle, 2660 struct super_block *sb, __u32 rocompat); 2661 extern int ext4_update_incompat_feature(handle_t *handle, 2662 struct super_block *sb, __u32 incompat); 2663 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2664 struct ext4_group_desc *bg); 2665 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2666 struct ext4_group_desc *bg); 2667 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2668 struct ext4_group_desc *bg); 2669 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2670 struct ext4_group_desc *bg); 2671 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2672 struct ext4_group_desc *bg); 2673 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2674 struct ext4_group_desc *bg); 2675 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2676 struct ext4_group_desc *bg); 2677 extern void ext4_block_bitmap_set(struct super_block *sb, 2678 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2679 extern void ext4_inode_bitmap_set(struct super_block *sb, 2680 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2681 extern void ext4_inode_table_set(struct super_block *sb, 2682 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2683 extern void ext4_free_group_clusters_set(struct super_block *sb, 2684 struct ext4_group_desc *bg, 2685 __u32 count); 2686 extern void ext4_free_inodes_set(struct super_block *sb, 2687 struct ext4_group_desc *bg, __u32 count); 2688 extern void ext4_used_dirs_set(struct super_block *sb, 2689 struct ext4_group_desc *bg, __u32 count); 2690 extern void ext4_itable_unused_set(struct super_block *sb, 2691 struct ext4_group_desc *bg, __u32 count); 2692 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2693 struct ext4_group_desc *gdp); 2694 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2695 struct ext4_group_desc *gdp); 2696 extern int ext4_register_li_request(struct super_block *sb, 2697 ext4_group_t first_not_zeroed); 2698 2699 static inline int ext4_has_metadata_csum(struct super_block *sb) 2700 { 2701 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2702 !EXT4_SB(sb)->s_chksum_driver); 2703 2704 return ext4_has_feature_metadata_csum(sb) && 2705 (EXT4_SB(sb)->s_chksum_driver != NULL); 2706 } 2707 2708 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2709 { 2710 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2711 } 2712 2713 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2714 { 2715 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2716 le32_to_cpu(es->s_blocks_count_lo); 2717 } 2718 2719 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2720 { 2721 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2722 le32_to_cpu(es->s_r_blocks_count_lo); 2723 } 2724 2725 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2726 { 2727 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2728 le32_to_cpu(es->s_free_blocks_count_lo); 2729 } 2730 2731 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2732 ext4_fsblk_t blk) 2733 { 2734 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2735 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2736 } 2737 2738 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2739 ext4_fsblk_t blk) 2740 { 2741 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2742 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2743 } 2744 2745 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2746 ext4_fsblk_t blk) 2747 { 2748 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2749 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2750 } 2751 2752 static inline loff_t ext4_isize(struct super_block *sb, 2753 struct ext4_inode *raw_inode) 2754 { 2755 if (ext4_has_feature_largedir(sb) || 2756 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2757 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2758 le32_to_cpu(raw_inode->i_size_lo); 2759 2760 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2761 } 2762 2763 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2764 { 2765 raw_inode->i_size_lo = cpu_to_le32(i_size); 2766 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2767 } 2768 2769 static inline 2770 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2771 ext4_group_t group) 2772 { 2773 struct ext4_group_info ***grp_info; 2774 long indexv, indexh; 2775 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2776 grp_info = EXT4_SB(sb)->s_group_info; 2777 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2778 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2779 return grp_info[indexv][indexh]; 2780 } 2781 2782 /* 2783 * Reading s_groups_count requires using smp_rmb() afterwards. See 2784 * the locking protocol documented in the comments of ext4_group_add() 2785 * in resize.c 2786 */ 2787 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2788 { 2789 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2790 2791 smp_rmb(); 2792 return ngroups; 2793 } 2794 2795 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2796 ext4_group_t block_group) 2797 { 2798 return block_group >> sbi->s_log_groups_per_flex; 2799 } 2800 2801 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2802 { 2803 return 1 << sbi->s_log_groups_per_flex; 2804 } 2805 2806 #define ext4_std_error(sb, errno) \ 2807 do { \ 2808 if ((errno)) \ 2809 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2810 } while (0) 2811 2812 #ifdef CONFIG_SMP 2813 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2814 * counters. So we need to make sure we have free clusters more 2815 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2816 */ 2817 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2818 #else 2819 #define EXT4_FREECLUSTERS_WATERMARK 0 2820 #endif 2821 2822 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2823 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2824 { 2825 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2826 !inode_is_locked(inode)); 2827 down_write(&EXT4_I(inode)->i_data_sem); 2828 if (newsize > EXT4_I(inode)->i_disksize) 2829 EXT4_I(inode)->i_disksize = newsize; 2830 up_write(&EXT4_I(inode)->i_data_sem); 2831 } 2832 2833 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2834 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2835 { 2836 int changed = 0; 2837 2838 if (newsize > inode->i_size) { 2839 i_size_write(inode, newsize); 2840 changed = 1; 2841 } 2842 if (newsize > EXT4_I(inode)->i_disksize) { 2843 ext4_update_i_disksize(inode, newsize); 2844 changed |= 2; 2845 } 2846 return changed; 2847 } 2848 2849 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2850 loff_t len); 2851 2852 struct ext4_group_info { 2853 unsigned long bb_state; 2854 struct rb_root bb_free_root; 2855 ext4_grpblk_t bb_first_free; /* first free block */ 2856 ext4_grpblk_t bb_free; /* total free blocks */ 2857 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2858 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2859 struct list_head bb_prealloc_list; 2860 #ifdef DOUBLE_CHECK 2861 void *bb_bitmap; 2862 #endif 2863 struct rw_semaphore alloc_sem; 2864 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2865 * regions, index is order. 2866 * bb_counters[3] = 5 means 2867 * 5 free 8-block regions. */ 2868 }; 2869 2870 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2871 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2872 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2873 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2874 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 2875 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 2876 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 2877 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 2878 2879 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2880 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2881 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2882 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2883 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2884 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2885 2886 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2887 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2888 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2889 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2890 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2891 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2892 2893 #define EXT4_MAX_CONTENTION 8 2894 #define EXT4_CONTENTION_THRESHOLD 2 2895 2896 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2897 ext4_group_t group) 2898 { 2899 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2900 } 2901 2902 /* 2903 * Returns true if the filesystem is busy enough that attempts to 2904 * access the block group locks has run into contention. 2905 */ 2906 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2907 { 2908 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2909 } 2910 2911 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2912 { 2913 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2914 if (spin_trylock(lock)) 2915 /* 2916 * We're able to grab the lock right away, so drop the 2917 * lock contention counter. 2918 */ 2919 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2920 else { 2921 /* 2922 * The lock is busy, so bump the contention counter, 2923 * and then wait on the spin lock. 2924 */ 2925 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2926 EXT4_MAX_CONTENTION); 2927 spin_lock(lock); 2928 } 2929 } 2930 2931 static inline void ext4_unlock_group(struct super_block *sb, 2932 ext4_group_t group) 2933 { 2934 spin_unlock(ext4_group_lock_ptr(sb, group)); 2935 } 2936 2937 /* 2938 * Block validity checking 2939 */ 2940 #define ext4_check_indirect_blockref(inode, bh) \ 2941 ext4_check_blockref(__func__, __LINE__, inode, \ 2942 (__le32 *)(bh)->b_data, \ 2943 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2944 2945 #define ext4_ind_check_inode(inode) \ 2946 ext4_check_blockref(__func__, __LINE__, inode, \ 2947 EXT4_I(inode)->i_data, \ 2948 EXT4_NDIR_BLOCKS) 2949 2950 /* 2951 * Inodes and files operations 2952 */ 2953 2954 /* dir.c */ 2955 extern const struct file_operations ext4_dir_operations; 2956 2957 /* file.c */ 2958 extern const struct inode_operations ext4_file_inode_operations; 2959 extern const struct file_operations ext4_file_operations; 2960 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2961 2962 /* inline.c */ 2963 extern int ext4_get_max_inline_size(struct inode *inode); 2964 extern int ext4_find_inline_data_nolock(struct inode *inode); 2965 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2966 unsigned int len); 2967 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2968 2969 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2970 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2971 struct inode *inode, 2972 loff_t pos, unsigned len, 2973 unsigned flags, 2974 struct page **pagep); 2975 extern int ext4_write_inline_data_end(struct inode *inode, 2976 loff_t pos, unsigned len, 2977 unsigned copied, 2978 struct page *page); 2979 extern struct buffer_head * 2980 ext4_journalled_write_inline_data(struct inode *inode, 2981 unsigned len, 2982 struct page *page); 2983 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2984 struct inode *inode, 2985 loff_t pos, unsigned len, 2986 unsigned flags, 2987 struct page **pagep, 2988 void **fsdata); 2989 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2990 unsigned len, unsigned copied, 2991 struct page *page); 2992 extern int ext4_try_add_inline_entry(handle_t *handle, 2993 struct ext4_filename *fname, 2994 struct inode *dir, struct inode *inode); 2995 extern int ext4_try_create_inline_dir(handle_t *handle, 2996 struct inode *parent, 2997 struct inode *inode); 2998 extern int ext4_read_inline_dir(struct file *filp, 2999 struct dir_context *ctx, 3000 int *has_inline_data); 3001 extern int htree_inlinedir_to_tree(struct file *dir_file, 3002 struct inode *dir, ext4_lblk_t block, 3003 struct dx_hash_info *hinfo, 3004 __u32 start_hash, __u32 start_minor_hash, 3005 int *has_inline_data); 3006 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3007 struct ext4_filename *fname, 3008 struct ext4_dir_entry_2 **res_dir, 3009 int *has_inline_data); 3010 extern int ext4_delete_inline_entry(handle_t *handle, 3011 struct inode *dir, 3012 struct ext4_dir_entry_2 *de_del, 3013 struct buffer_head *bh, 3014 int *has_inline_data); 3015 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3016 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3017 struct ext4_dir_entry_2 **parent_de, 3018 int *retval); 3019 extern int ext4_inline_data_fiemap(struct inode *inode, 3020 struct fiemap_extent_info *fieinfo, 3021 int *has_inline, __u64 start, __u64 len); 3022 3023 struct iomap; 3024 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3025 3026 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3027 3028 extern int ext4_convert_inline_data(struct inode *inode); 3029 3030 static inline int ext4_has_inline_data(struct inode *inode) 3031 { 3032 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3033 EXT4_I(inode)->i_inline_off; 3034 } 3035 3036 /* namei.c */ 3037 extern const struct inode_operations ext4_dir_inode_operations; 3038 extern const struct inode_operations ext4_special_inode_operations; 3039 extern struct dentry *ext4_get_parent(struct dentry *child); 3040 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3041 struct ext4_dir_entry_2 *de, 3042 int blocksize, int csum_size, 3043 unsigned int parent_ino, int dotdot_real_len); 3044 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3045 unsigned int blocksize); 3046 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3047 struct inode *inode, 3048 struct buffer_head *bh); 3049 #define S_SHIFT 12 3050 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3051 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3052 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3053 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3054 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3055 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3056 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3057 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3058 }; 3059 3060 static inline void ext4_set_de_type(struct super_block *sb, 3061 struct ext4_dir_entry_2 *de, 3062 umode_t mode) { 3063 if (ext4_has_feature_filetype(sb)) 3064 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3065 } 3066 3067 /* readpages.c */ 3068 extern int ext4_mpage_readpages(struct address_space *mapping, 3069 struct list_head *pages, struct page *page, 3070 unsigned nr_pages, bool is_readahead); 3071 3072 /* symlink.c */ 3073 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3074 extern const struct inode_operations ext4_symlink_inode_operations; 3075 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3076 3077 /* sysfs.c */ 3078 extern int ext4_register_sysfs(struct super_block *sb); 3079 extern void ext4_unregister_sysfs(struct super_block *sb); 3080 extern int __init ext4_init_sysfs(void); 3081 extern void ext4_exit_sysfs(void); 3082 3083 /* block_validity */ 3084 extern void ext4_release_system_zone(struct super_block *sb); 3085 extern int ext4_setup_system_zone(struct super_block *sb); 3086 extern int __init ext4_init_system_zone(void); 3087 extern void ext4_exit_system_zone(void); 3088 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3089 ext4_fsblk_t start_blk, 3090 unsigned int count); 3091 extern int ext4_check_blockref(const char *, unsigned int, 3092 struct inode *, __le32 *, unsigned int); 3093 3094 /* extents.c */ 3095 struct ext4_ext_path; 3096 struct ext4_extent; 3097 3098 /* 3099 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3100 * __le32. 3101 */ 3102 #define EXT_MAX_BLOCKS 0xffffffff 3103 3104 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3105 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3106 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3107 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3108 struct ext4_map_blocks *map, int flags); 3109 extern int ext4_ext_truncate(handle_t *, struct inode *); 3110 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3111 ext4_lblk_t end); 3112 extern void ext4_ext_init(struct super_block *); 3113 extern void ext4_ext_release(struct super_block *); 3114 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3115 loff_t len); 3116 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3117 loff_t offset, ssize_t len); 3118 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3119 struct ext4_map_blocks *map, int flags); 3120 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3121 ext4_lblk_t lblocks); 3122 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3123 int num, 3124 struct ext4_ext_path *path); 3125 extern int ext4_can_extents_be_merged(struct inode *inode, 3126 struct ext4_extent *ex1, 3127 struct ext4_extent *ex2); 3128 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3129 struct ext4_ext_path **, 3130 struct ext4_extent *, int); 3131 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3132 struct ext4_ext_path **, 3133 int flags); 3134 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3135 extern int ext4_ext_check_inode(struct inode *inode); 3136 extern int ext4_find_delalloc_range(struct inode *inode, 3137 ext4_lblk_t lblk_start, 3138 ext4_lblk_t lblk_end); 3139 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3140 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3141 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3142 __u64 start, __u64 len); 3143 extern int ext4_ext_precache(struct inode *inode); 3144 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3145 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3146 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3147 struct inode *inode2, ext4_lblk_t lblk1, 3148 ext4_lblk_t lblk2, ext4_lblk_t count, 3149 int mark_unwritten,int *err); 3150 3151 /* move_extent.c */ 3152 extern void ext4_double_down_write_data_sem(struct inode *first, 3153 struct inode *second); 3154 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3155 struct inode *donor_inode); 3156 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3157 __u64 start_orig, __u64 start_donor, 3158 __u64 len, __u64 *moved_len); 3159 3160 /* page-io.c */ 3161 extern int __init ext4_init_pageio(void); 3162 extern void ext4_exit_pageio(void); 3163 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3164 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3165 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3166 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3167 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3168 struct writeback_control *wbc); 3169 extern void ext4_end_io_rsv_work(struct work_struct *work); 3170 extern void ext4_io_submit(struct ext4_io_submit *io); 3171 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3172 struct page *page, 3173 int len, 3174 struct writeback_control *wbc, 3175 bool keep_towrite); 3176 3177 /* mmp.c */ 3178 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3179 3180 /* 3181 * Add new method to test whether block and inode bitmaps are properly 3182 * initialized. With uninit_bg reading the block from disk is not enough 3183 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3184 */ 3185 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3186 3187 static inline int bitmap_uptodate(struct buffer_head *bh) 3188 { 3189 return (buffer_uptodate(bh) && 3190 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3191 } 3192 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3193 { 3194 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3195 } 3196 3197 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3198 3199 /* For ioend & aio unwritten conversion wait queues */ 3200 #define EXT4_WQ_HASH_SZ 37 3201 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3202 EXT4_WQ_HASH_SZ]) 3203 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3204 3205 extern int ext4_resize_begin(struct super_block *sb); 3206 extern void ext4_resize_end(struct super_block *sb); 3207 3208 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3209 struct ext4_io_end *io_end) 3210 { 3211 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3212 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3213 atomic_inc(&EXT4_I(inode)->i_unwritten); 3214 } 3215 } 3216 3217 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3218 { 3219 struct inode *inode = io_end->inode; 3220 3221 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3222 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3223 /* Wake up anyone waiting on unwritten extent conversion */ 3224 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3225 wake_up_all(ext4_ioend_wq(inode)); 3226 } 3227 } 3228 3229 extern const struct iomap_ops ext4_iomap_ops; 3230 3231 #endif /* __KERNEL__ */ 3232 3233 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3234 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3235 3236 #endif /* _EXT4_H */ 3237