1 /* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16 #ifndef _EXT4_H 17 #define _EXT4_H 18 19 #include <linux/types.h> 20 #include <linux/blkdev.h> 21 #include <linux/magic.h> 22 #include <linux/jbd2.h> 23 #include <linux/quota.h> 24 #include <linux/rwsem.h> 25 #include <linux/rbtree.h> 26 #include <linux/seqlock.h> 27 #include <linux/mutex.h> 28 #include <linux/timer.h> 29 #include <linux/wait.h> 30 #include <linux/blockgroup_lock.h> 31 #include <linux/percpu_counter.h> 32 #include <linux/ratelimit.h> 33 #include <crypto/hash.h> 34 #include <linux/falloc.h> 35 #ifdef __KERNEL__ 36 #include <linux/compat.h> 37 #endif 38 39 /* 40 * The fourth extended filesystem constants/structures 41 */ 42 43 /* 44 * Define EXT4FS_DEBUG to produce debug messages 45 */ 46 #undef EXT4FS_DEBUG 47 48 /* 49 * Debug code 50 */ 51 #ifdef EXT4FS_DEBUG 52 #define ext4_debug(f, a...) \ 53 do { \ 54 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 55 __FILE__, __LINE__, __func__); \ 56 printk(KERN_DEBUG f, ## a); \ 57 } while (0) 58 #else 59 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 60 #endif 61 62 /* 63 * Turn on EXT_DEBUG to get lots of info about extents operations. 64 */ 65 #define EXT_DEBUG__ 66 #ifdef EXT_DEBUG 67 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 68 #else 69 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 70 #endif 71 72 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 73 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 74 75 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 76 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 77 78 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 79 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 80 81 /* data type for block offset of block group */ 82 typedef int ext4_grpblk_t; 83 84 /* data type for filesystem-wide blocks number */ 85 typedef unsigned long long ext4_fsblk_t; 86 87 /* data type for file logical block number */ 88 typedef __u32 ext4_lblk_t; 89 90 /* data type for block group number */ 91 typedef unsigned int ext4_group_t; 92 93 /* 94 * Flags used in mballoc's allocation_context flags field. 95 * 96 * Also used to show what's going on for debugging purposes when the 97 * flag field is exported via the traceport interface 98 */ 99 100 /* prefer goal again. length */ 101 #define EXT4_MB_HINT_MERGE 0x0001 102 /* blocks already reserved */ 103 #define EXT4_MB_HINT_RESERVED 0x0002 104 /* metadata is being allocated */ 105 #define EXT4_MB_HINT_METADATA 0x0004 106 /* first blocks in the file */ 107 #define EXT4_MB_HINT_FIRST 0x0008 108 /* search for the best chunk */ 109 #define EXT4_MB_HINT_BEST 0x0010 110 /* data is being allocated */ 111 #define EXT4_MB_HINT_DATA 0x0020 112 /* don't preallocate (for tails) */ 113 #define EXT4_MB_HINT_NOPREALLOC 0x0040 114 /* allocate for locality group */ 115 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 116 /* allocate goal blocks or none */ 117 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 118 /* goal is meaningful */ 119 #define EXT4_MB_HINT_TRY_GOAL 0x0200 120 /* blocks already pre-reserved by delayed allocation */ 121 #define EXT4_MB_DELALLOC_RESERVED 0x0400 122 /* We are doing stream allocation */ 123 #define EXT4_MB_STREAM_ALLOC 0x0800 124 /* Use reserved root blocks if needed */ 125 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 126 /* Use blocks from reserved pool */ 127 #define EXT4_MB_USE_RESERVED 0x2000 128 129 struct ext4_allocation_request { 130 /* target inode for block we're allocating */ 131 struct inode *inode; 132 /* how many blocks we want to allocate */ 133 unsigned int len; 134 /* logical block in target inode */ 135 ext4_lblk_t logical; 136 /* the closest logical allocated block to the left */ 137 ext4_lblk_t lleft; 138 /* the closest logical allocated block to the right */ 139 ext4_lblk_t lright; 140 /* phys. target (a hint) */ 141 ext4_fsblk_t goal; 142 /* phys. block for the closest logical allocated block to the left */ 143 ext4_fsblk_t pleft; 144 /* phys. block for the closest logical allocated block to the right */ 145 ext4_fsblk_t pright; 146 /* flags. see above EXT4_MB_HINT_* */ 147 unsigned int flags; 148 }; 149 150 /* 151 * Logical to physical block mapping, used by ext4_map_blocks() 152 * 153 * This structure is used to pass requests into ext4_map_blocks() as 154 * well as to store the information returned by ext4_map_blocks(). It 155 * takes less room on the stack than a struct buffer_head. 156 */ 157 #define EXT4_MAP_NEW (1 << BH_New) 158 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 159 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 160 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 161 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 162 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 163 164 struct ext4_map_blocks { 165 ext4_fsblk_t m_pblk; 166 ext4_lblk_t m_lblk; 167 unsigned int m_len; 168 unsigned int m_flags; 169 }; 170 171 /* 172 * Flags for ext4_io_end->flags 173 */ 174 #define EXT4_IO_END_UNWRITTEN 0x0001 175 176 /* 177 * For converting unwritten extents on a work queue. 'handle' is used for 178 * buffered writeback. 179 */ 180 typedef struct ext4_io_end { 181 struct list_head list; /* per-file finished IO list */ 182 handle_t *handle; /* handle reserved for extent 183 * conversion */ 184 struct inode *inode; /* file being written to */ 185 struct bio *bio; /* Linked list of completed 186 * bios covering the extent */ 187 unsigned int flag; /* unwritten or not */ 188 loff_t offset; /* offset in the file */ 189 ssize_t size; /* size of the extent */ 190 atomic_t count; /* reference counter */ 191 } ext4_io_end_t; 192 193 struct ext4_io_submit { 194 int io_op; 195 struct bio *io_bio; 196 ext4_io_end_t *io_end; 197 sector_t io_next_block; 198 }; 199 200 /* 201 * Special inodes numbers 202 */ 203 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 204 #define EXT4_ROOT_INO 2 /* Root inode */ 205 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 206 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 207 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 208 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 209 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 210 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 211 212 /* First non-reserved inode for old ext4 filesystems */ 213 #define EXT4_GOOD_OLD_FIRST_INO 11 214 215 /* 216 * Maximal count of links to a file 217 */ 218 #define EXT4_LINK_MAX 65000 219 220 /* 221 * Macro-instructions used to manage several block sizes 222 */ 223 #define EXT4_MIN_BLOCK_SIZE 1024 224 #define EXT4_MAX_BLOCK_SIZE 65536 225 #define EXT4_MIN_BLOCK_LOG_SIZE 10 226 #define EXT4_MAX_BLOCK_LOG_SIZE 16 227 #ifdef __KERNEL__ 228 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 229 #else 230 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 231 #endif 232 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 233 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 234 EXT4_SB(s)->s_cluster_bits) 235 #ifdef __KERNEL__ 236 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 237 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 238 #else 239 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 240 #endif 241 #ifdef __KERNEL__ 242 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 243 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 244 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 245 #else 246 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 247 EXT4_GOOD_OLD_INODE_SIZE : \ 248 (s)->s_inode_size) 249 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 250 EXT4_GOOD_OLD_FIRST_INO : \ 251 (s)->s_first_ino) 252 #endif 253 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 254 255 /* Translate a block number to a cluster number */ 256 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 257 /* Translate a cluster number to a block number */ 258 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 259 /* Translate # of blks to # of clusters */ 260 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 261 (sbi)->s_cluster_bits) 262 /* Mask out the low bits to get the starting block of the cluster */ 263 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 264 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 265 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 266 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 267 /* Get the cluster offset */ 268 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 269 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 270 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 271 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 272 273 /* 274 * Structure of a blocks group descriptor 275 */ 276 struct ext4_group_desc 277 { 278 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 279 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 280 __le32 bg_inode_table_lo; /* Inodes table block */ 281 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 282 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 283 __le16 bg_used_dirs_count_lo; /* Directories count */ 284 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 285 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 286 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 287 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 288 __le16 bg_itable_unused_lo; /* Unused inodes count */ 289 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 290 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 291 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 292 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 293 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 294 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 295 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 296 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 297 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 298 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 299 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 300 __u32 bg_reserved; 301 }; 302 303 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 304 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 305 sizeof(__le16)) 306 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 307 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 308 sizeof(__le16)) 309 310 /* 311 * Structure of a flex block group info 312 */ 313 314 struct flex_groups { 315 atomic64_t free_clusters; 316 atomic_t free_inodes; 317 atomic_t used_dirs; 318 }; 319 320 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 321 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 322 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 323 324 /* 325 * Macro-instructions used to manage group descriptors 326 */ 327 #define EXT4_MIN_DESC_SIZE 32 328 #define EXT4_MIN_DESC_SIZE_64BIT 64 329 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 330 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 331 #ifdef __KERNEL__ 332 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 333 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 334 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 335 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 336 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 337 #else 338 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 339 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 340 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 341 #endif 342 343 /* 344 * Constants relative to the data blocks 345 */ 346 #define EXT4_NDIR_BLOCKS 12 347 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 348 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 349 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 350 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 351 352 /* 353 * Inode flags 354 */ 355 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 356 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 357 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 358 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 359 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 360 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 361 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 362 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 363 /* Reserved for compression usage... */ 364 #define EXT4_DIRTY_FL 0x00000100 365 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 366 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 367 /* nb: was previously EXT2_ECOMPR_FL */ 368 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 369 /* End compression flags --- maybe not all used */ 370 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 371 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 372 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 373 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 374 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 375 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 376 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 377 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 378 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 379 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 380 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 381 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 382 383 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 384 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 385 386 /* Flags that should be inherited by new inodes from their parent. */ 387 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 388 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 389 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 390 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 391 392 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 393 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 394 395 /* Flags that are appropriate for non-directories/regular files. */ 396 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 397 398 /* Mask out flags that are inappropriate for the given type of inode. */ 399 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 400 { 401 if (S_ISDIR(mode)) 402 return flags; 403 else if (S_ISREG(mode)) 404 return flags & EXT4_REG_FLMASK; 405 else 406 return flags & EXT4_OTHER_FLMASK; 407 } 408 409 /* 410 * Inode flags used for atomic set/get 411 */ 412 enum { 413 EXT4_INODE_SECRM = 0, /* Secure deletion */ 414 EXT4_INODE_UNRM = 1, /* Undelete */ 415 EXT4_INODE_COMPR = 2, /* Compress file */ 416 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 417 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 418 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 419 EXT4_INODE_NODUMP = 6, /* do not dump file */ 420 EXT4_INODE_NOATIME = 7, /* do not update atime */ 421 /* Reserved for compression usage... */ 422 EXT4_INODE_DIRTY = 8, 423 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 424 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 425 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 426 /* End compression flags --- maybe not all used */ 427 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 428 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 429 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 430 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 431 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 432 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 433 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 434 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 435 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 436 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 437 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 438 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 439 }; 440 441 /* 442 * Since it's pretty easy to mix up bit numbers and hex values, we use a 443 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 444 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 445 * any extra space in the compiled kernel image, otherwise, the build will fail. 446 * It's important that these values are the same, since we are using 447 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 448 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 449 * values found in ext2, ext3 and ext4 filesystems, and of course the values 450 * defined in e2fsprogs. 451 * 452 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 453 */ 454 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 455 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 456 457 static inline void ext4_check_flag_values(void) 458 { 459 CHECK_FLAG_VALUE(SECRM); 460 CHECK_FLAG_VALUE(UNRM); 461 CHECK_FLAG_VALUE(COMPR); 462 CHECK_FLAG_VALUE(SYNC); 463 CHECK_FLAG_VALUE(IMMUTABLE); 464 CHECK_FLAG_VALUE(APPEND); 465 CHECK_FLAG_VALUE(NODUMP); 466 CHECK_FLAG_VALUE(NOATIME); 467 CHECK_FLAG_VALUE(DIRTY); 468 CHECK_FLAG_VALUE(COMPRBLK); 469 CHECK_FLAG_VALUE(NOCOMPR); 470 CHECK_FLAG_VALUE(ENCRYPT); 471 CHECK_FLAG_VALUE(INDEX); 472 CHECK_FLAG_VALUE(IMAGIC); 473 CHECK_FLAG_VALUE(JOURNAL_DATA); 474 CHECK_FLAG_VALUE(NOTAIL); 475 CHECK_FLAG_VALUE(DIRSYNC); 476 CHECK_FLAG_VALUE(TOPDIR); 477 CHECK_FLAG_VALUE(HUGE_FILE); 478 CHECK_FLAG_VALUE(EXTENTS); 479 CHECK_FLAG_VALUE(EA_INODE); 480 CHECK_FLAG_VALUE(EOFBLOCKS); 481 CHECK_FLAG_VALUE(INLINE_DATA); 482 CHECK_FLAG_VALUE(RESERVED); 483 } 484 485 /* Used to pass group descriptor data when online resize is done */ 486 struct ext4_new_group_input { 487 __u32 group; /* Group number for this data */ 488 __u64 block_bitmap; /* Absolute block number of block bitmap */ 489 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 490 __u64 inode_table; /* Absolute block number of inode table start */ 491 __u32 blocks_count; /* Total number of blocks in this group */ 492 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 493 __u16 unused; 494 }; 495 496 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 497 struct compat_ext4_new_group_input { 498 u32 group; 499 compat_u64 block_bitmap; 500 compat_u64 inode_bitmap; 501 compat_u64 inode_table; 502 u32 blocks_count; 503 u16 reserved_blocks; 504 u16 unused; 505 }; 506 #endif 507 508 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 509 struct ext4_new_group_data { 510 __u32 group; 511 __u64 block_bitmap; 512 __u64 inode_bitmap; 513 __u64 inode_table; 514 __u32 blocks_count; 515 __u16 reserved_blocks; 516 __u16 unused; 517 __u32 free_blocks_count; 518 }; 519 520 /* Indexes used to index group tables in ext4_new_group_data */ 521 enum { 522 BLOCK_BITMAP = 0, /* block bitmap */ 523 INODE_BITMAP, /* inode bitmap */ 524 INODE_TABLE, /* inode tables */ 525 GROUP_TABLE_COUNT, 526 }; 527 528 /* 529 * Flags used by ext4_map_blocks() 530 */ 531 /* Allocate any needed blocks and/or convert an unwritten 532 extent to be an initialized ext4 */ 533 #define EXT4_GET_BLOCKS_CREATE 0x0001 534 /* Request the creation of an unwritten extent */ 535 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 536 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 537 EXT4_GET_BLOCKS_CREATE) 538 /* Caller is from the delayed allocation writeout path 539 * finally doing the actual allocation of delayed blocks */ 540 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 541 /* caller is from the direct IO path, request to creation of an 542 unwritten extents if not allocated, split the unwritten 543 extent if blocks has been preallocated already*/ 544 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 545 #define EXT4_GET_BLOCKS_CONVERT 0x0010 546 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 547 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 548 /* Convert extent to initialized after IO complete */ 549 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 550 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 551 /* Eventual metadata allocation (due to growing extent tree) 552 * should not fail, so try to use reserved blocks for that.*/ 553 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 554 /* Don't normalize allocation size (used for fallocate) */ 555 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 556 /* Request will not result in inode size update (user for fallocate) */ 557 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 558 /* Do not take i_data_sem locking in ext4_map_blocks */ 559 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 560 /* Convert written extents to unwritten */ 561 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200 562 563 /* 564 * The bit position of these flags must not overlap with any of the 565 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 566 * read_extent_tree_block(), ext4_split_extent_at(), 567 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 568 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 569 * caching the extents when reading from the extent tree while a 570 * truncate or punch hole operation is in progress. 571 */ 572 #define EXT4_EX_NOCACHE 0x40000000 573 #define EXT4_EX_FORCE_CACHE 0x20000000 574 575 /* 576 * Flags used by ext4_free_blocks 577 */ 578 #define EXT4_FREE_BLOCKS_METADATA 0x0001 579 #define EXT4_FREE_BLOCKS_FORGET 0x0002 580 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 581 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 582 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 583 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 584 585 /* Encryption algorithms */ 586 #define EXT4_ENCRYPTION_MODE_INVALID 0 587 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1 588 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2 589 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3 590 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4 591 592 /* 593 * ioctl commands 594 */ 595 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 596 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 597 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 598 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 599 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 600 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 601 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 602 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 603 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 604 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 605 #define EXT4_IOC_MIGRATE _IO('f', 9) 606 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 607 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 608 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 609 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 610 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 611 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 612 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 613 614 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 615 /* 616 * ioctl commands in 32 bit emulation 617 */ 618 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 619 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 620 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 621 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 622 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 623 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 624 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 625 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 626 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 627 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 628 #endif 629 630 /* Max physical block we can address w/o extents */ 631 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 632 633 /* 634 * Structure of an inode on the disk 635 */ 636 struct ext4_inode { 637 __le16 i_mode; /* File mode */ 638 __le16 i_uid; /* Low 16 bits of Owner Uid */ 639 __le32 i_size_lo; /* Size in bytes */ 640 __le32 i_atime; /* Access time */ 641 __le32 i_ctime; /* Inode Change time */ 642 __le32 i_mtime; /* Modification time */ 643 __le32 i_dtime; /* Deletion Time */ 644 __le16 i_gid; /* Low 16 bits of Group Id */ 645 __le16 i_links_count; /* Links count */ 646 __le32 i_blocks_lo; /* Blocks count */ 647 __le32 i_flags; /* File flags */ 648 union { 649 struct { 650 __le32 l_i_version; 651 } linux1; 652 struct { 653 __u32 h_i_translator; 654 } hurd1; 655 struct { 656 __u32 m_i_reserved1; 657 } masix1; 658 } osd1; /* OS dependent 1 */ 659 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 660 __le32 i_generation; /* File version (for NFS) */ 661 __le32 i_file_acl_lo; /* File ACL */ 662 __le32 i_size_high; 663 __le32 i_obso_faddr; /* Obsoleted fragment address */ 664 union { 665 struct { 666 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 667 __le16 l_i_file_acl_high; 668 __le16 l_i_uid_high; /* these 2 fields */ 669 __le16 l_i_gid_high; /* were reserved2[0] */ 670 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 671 __le16 l_i_reserved; 672 } linux2; 673 struct { 674 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 675 __u16 h_i_mode_high; 676 __u16 h_i_uid_high; 677 __u16 h_i_gid_high; 678 __u32 h_i_author; 679 } hurd2; 680 struct { 681 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 682 __le16 m_i_file_acl_high; 683 __u32 m_i_reserved2[2]; 684 } masix2; 685 } osd2; /* OS dependent 2 */ 686 __le16 i_extra_isize; 687 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 688 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 689 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 690 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 691 __le32 i_crtime; /* File Creation time */ 692 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 693 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 694 }; 695 696 struct move_extent { 697 __u32 reserved; /* should be zero */ 698 __u32 donor_fd; /* donor file descriptor */ 699 __u64 orig_start; /* logical start offset in block for orig */ 700 __u64 donor_start; /* logical start offset in block for donor */ 701 __u64 len; /* block length to be moved */ 702 __u64 moved_len; /* moved block length */ 703 }; 704 705 #define EXT4_EPOCH_BITS 2 706 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 707 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 708 709 /* 710 * Extended fields will fit into an inode if the filesystem was formatted 711 * with large inodes (-I 256 or larger) and there are not currently any EAs 712 * consuming all of the available space. For new inodes we always reserve 713 * enough space for the kernel's known extended fields, but for inodes 714 * created with an old kernel this might not have been the case. None of 715 * the extended inode fields is critical for correct filesystem operation. 716 * This macro checks if a certain field fits in the inode. Note that 717 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 718 */ 719 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 720 ((offsetof(typeof(*ext4_inode), field) + \ 721 sizeof((ext4_inode)->field)) \ 722 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 723 (einode)->i_extra_isize)) \ 724 725 static inline __le32 ext4_encode_extra_time(struct timespec *time) 726 { 727 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 728 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 729 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 730 } 731 732 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 733 { 734 if (sizeof(time->tv_sec) > 4) 735 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 736 << 32; 737 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 738 } 739 740 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 741 do { \ 742 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 743 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 744 (raw_inode)->xtime ## _extra = \ 745 ext4_encode_extra_time(&(inode)->xtime); \ 746 } while (0) 747 748 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 749 do { \ 750 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 751 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 752 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 753 (raw_inode)->xtime ## _extra = \ 754 ext4_encode_extra_time(&(einode)->xtime); \ 755 } while (0) 756 757 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 758 do { \ 759 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 760 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 761 ext4_decode_extra_time(&(inode)->xtime, \ 762 raw_inode->xtime ## _extra); \ 763 else \ 764 (inode)->xtime.tv_nsec = 0; \ 765 } while (0) 766 767 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 768 do { \ 769 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 770 (einode)->xtime.tv_sec = \ 771 (signed)le32_to_cpu((raw_inode)->xtime); \ 772 else \ 773 (einode)->xtime.tv_sec = 0; \ 774 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 775 ext4_decode_extra_time(&(einode)->xtime, \ 776 raw_inode->xtime ## _extra); \ 777 else \ 778 (einode)->xtime.tv_nsec = 0; \ 779 } while (0) 780 781 #define i_disk_version osd1.linux1.l_i_version 782 783 #if defined(__KERNEL__) || defined(__linux__) 784 #define i_reserved1 osd1.linux1.l_i_reserved1 785 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 786 #define i_blocks_high osd2.linux2.l_i_blocks_high 787 #define i_uid_low i_uid 788 #define i_gid_low i_gid 789 #define i_uid_high osd2.linux2.l_i_uid_high 790 #define i_gid_high osd2.linux2.l_i_gid_high 791 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 792 793 #elif defined(__GNU__) 794 795 #define i_translator osd1.hurd1.h_i_translator 796 #define i_uid_high osd2.hurd2.h_i_uid_high 797 #define i_gid_high osd2.hurd2.h_i_gid_high 798 #define i_author osd2.hurd2.h_i_author 799 800 #elif defined(__masix__) 801 802 #define i_reserved1 osd1.masix1.m_i_reserved1 803 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 804 #define i_reserved2 osd2.masix2.m_i_reserved2 805 806 #endif /* defined(__KERNEL__) || defined(__linux__) */ 807 808 #include "extents_status.h" 809 810 /* 811 * fourth extended file system inode data in memory 812 */ 813 struct ext4_inode_info { 814 __le32 i_data[15]; /* unconverted */ 815 __u32 i_dtime; 816 ext4_fsblk_t i_file_acl; 817 818 /* 819 * i_block_group is the number of the block group which contains 820 * this file's inode. Constant across the lifetime of the inode, 821 * it is ued for making block allocation decisions - we try to 822 * place a file's data blocks near its inode block, and new inodes 823 * near to their parent directory's inode. 824 */ 825 ext4_group_t i_block_group; 826 ext4_lblk_t i_dir_start_lookup; 827 #if (BITS_PER_LONG < 64) 828 unsigned long i_state_flags; /* Dynamic state flags */ 829 #endif 830 unsigned long i_flags; 831 832 /* 833 * Extended attributes can be read independently of the main file 834 * data. Taking i_mutex even when reading would cause contention 835 * between readers of EAs and writers of regular file data, so 836 * instead we synchronize on xattr_sem when reading or changing 837 * EAs. 838 */ 839 struct rw_semaphore xattr_sem; 840 841 struct list_head i_orphan; /* unlinked but open inodes */ 842 843 /* 844 * i_disksize keeps track of what the inode size is ON DISK, not 845 * in memory. During truncate, i_size is set to the new size by 846 * the VFS prior to calling ext4_truncate(), but the filesystem won't 847 * set i_disksize to 0 until the truncate is actually under way. 848 * 849 * The intent is that i_disksize always represents the blocks which 850 * are used by this file. This allows recovery to restart truncate 851 * on orphans if we crash during truncate. We actually write i_disksize 852 * into the on-disk inode when writing inodes out, instead of i_size. 853 * 854 * The only time when i_disksize and i_size may be different is when 855 * a truncate is in progress. The only things which change i_disksize 856 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 857 */ 858 loff_t i_disksize; 859 860 /* 861 * i_data_sem is for serialising ext4_truncate() against 862 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 863 * data tree are chopped off during truncate. We can't do that in 864 * ext4 because whenever we perform intermediate commits during 865 * truncate, the inode and all the metadata blocks *must* be in a 866 * consistent state which allows truncation of the orphans to restart 867 * during recovery. Hence we must fix the get_block-vs-truncate race 868 * by other means, so we have i_data_sem. 869 */ 870 struct rw_semaphore i_data_sem; 871 struct inode vfs_inode; 872 struct jbd2_inode *jinode; 873 874 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 875 876 /* 877 * File creation time. Its function is same as that of 878 * struct timespec i_{a,c,m}time in the generic inode. 879 */ 880 struct timespec i_crtime; 881 882 /* mballoc */ 883 struct list_head i_prealloc_list; 884 spinlock_t i_prealloc_lock; 885 886 /* extents status tree */ 887 struct ext4_es_tree i_es_tree; 888 rwlock_t i_es_lock; 889 struct list_head i_es_list; 890 unsigned int i_es_all_nr; /* protected by i_es_lock */ 891 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 892 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 893 extents to shrink. Protected by 894 i_es_lock */ 895 896 /* ialloc */ 897 ext4_group_t i_last_alloc_group; 898 899 /* allocation reservation info for delalloc */ 900 /* In case of bigalloc, these refer to clusters rather than blocks */ 901 unsigned int i_reserved_data_blocks; 902 unsigned int i_reserved_meta_blocks; 903 unsigned int i_allocated_meta_blocks; 904 ext4_lblk_t i_da_metadata_calc_last_lblock; 905 int i_da_metadata_calc_len; 906 907 /* on-disk additional length */ 908 __u16 i_extra_isize; 909 910 /* Indicate the inline data space. */ 911 u16 i_inline_off; 912 u16 i_inline_size; 913 914 #ifdef CONFIG_QUOTA 915 /* quota space reservation, managed internally by quota code */ 916 qsize_t i_reserved_quota; 917 #endif 918 919 /* Lock protecting lists below */ 920 spinlock_t i_completed_io_lock; 921 /* 922 * Completed IOs that need unwritten extents handling and have 923 * transaction reserved 924 */ 925 struct list_head i_rsv_conversion_list; 926 /* 927 * Completed IOs that need unwritten extents handling and don't have 928 * transaction reserved 929 */ 930 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 931 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 932 struct work_struct i_rsv_conversion_work; 933 934 spinlock_t i_block_reservation_lock; 935 936 /* 937 * Transactions that contain inode's metadata needed to complete 938 * fsync and fdatasync, respectively. 939 */ 940 tid_t i_sync_tid; 941 tid_t i_datasync_tid; 942 943 #ifdef CONFIG_QUOTA 944 struct dquot *i_dquot[MAXQUOTAS]; 945 #endif 946 947 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 948 __u32 i_csum_seed; 949 }; 950 951 /* 952 * File system states 953 */ 954 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 955 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 956 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 957 958 /* 959 * Misc. filesystem flags 960 */ 961 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 962 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 963 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 964 965 /* 966 * Mount flags set via mount options or defaults 967 */ 968 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 969 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 970 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 971 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 972 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 973 #define EXT4_MOUNT_ERRORS_MASK 0x00070 974 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 975 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 976 #ifdef CONFIG_FS_DAX 977 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 978 #else 979 #define EXT4_MOUNT_DAX 0 980 #endif 981 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 982 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 983 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 984 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 985 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 986 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 987 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 988 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 989 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 990 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 991 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 992 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 993 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 994 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 995 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 996 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 997 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 998 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 999 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1000 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1001 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1002 1003 /* 1004 * Mount flags set either automatically (could not be set by mount option) 1005 * based on per file system feature or property or in special cases such as 1006 * distinguishing between explicit mount option definition and default. 1007 */ 1008 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1009 specified delalloc */ 1010 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1011 size of blocksize * 8 1012 blocks */ 1013 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1014 file systems */ 1015 1016 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1017 ~EXT4_MOUNT_##opt 1018 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1019 EXT4_MOUNT_##opt 1020 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1021 EXT4_MOUNT_##opt) 1022 1023 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1024 ~EXT4_MOUNT2_##opt 1025 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1026 EXT4_MOUNT2_##opt 1027 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1028 EXT4_MOUNT2_##opt) 1029 1030 #define ext4_test_and_set_bit __test_and_set_bit_le 1031 #define ext4_set_bit __set_bit_le 1032 #define ext4_set_bit_atomic ext2_set_bit_atomic 1033 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1034 #define ext4_clear_bit __clear_bit_le 1035 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1036 #define ext4_test_bit test_bit_le 1037 #define ext4_find_next_zero_bit find_next_zero_bit_le 1038 #define ext4_find_next_bit find_next_bit_le 1039 1040 extern void ext4_set_bits(void *bm, int cur, int len); 1041 1042 /* 1043 * Maximal mount counts between two filesystem checks 1044 */ 1045 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1046 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1047 1048 /* 1049 * Behaviour when detecting errors 1050 */ 1051 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1052 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1053 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1054 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1055 1056 /* Metadata checksum algorithm codes */ 1057 #define EXT4_CRC32C_CHKSUM 1 1058 1059 /* Encryption algorithms */ 1060 #define EXT4_ENCRYPTION_MODE_INVALID 0 1061 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1 1062 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2 1063 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3 1064 1065 /* 1066 * Structure of the super block 1067 */ 1068 struct ext4_super_block { 1069 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1070 __le32 s_blocks_count_lo; /* Blocks count */ 1071 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1072 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1073 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1074 __le32 s_first_data_block; /* First Data Block */ 1075 __le32 s_log_block_size; /* Block size */ 1076 __le32 s_log_cluster_size; /* Allocation cluster size */ 1077 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1078 __le32 s_clusters_per_group; /* # Clusters per group */ 1079 __le32 s_inodes_per_group; /* # Inodes per group */ 1080 __le32 s_mtime; /* Mount time */ 1081 /*30*/ __le32 s_wtime; /* Write time */ 1082 __le16 s_mnt_count; /* Mount count */ 1083 __le16 s_max_mnt_count; /* Maximal mount count */ 1084 __le16 s_magic; /* Magic signature */ 1085 __le16 s_state; /* File system state */ 1086 __le16 s_errors; /* Behaviour when detecting errors */ 1087 __le16 s_minor_rev_level; /* minor revision level */ 1088 /*40*/ __le32 s_lastcheck; /* time of last check */ 1089 __le32 s_checkinterval; /* max. time between checks */ 1090 __le32 s_creator_os; /* OS */ 1091 __le32 s_rev_level; /* Revision level */ 1092 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1093 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1094 /* 1095 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1096 * 1097 * Note: the difference between the compatible feature set and 1098 * the incompatible feature set is that if there is a bit set 1099 * in the incompatible feature set that the kernel doesn't 1100 * know about, it should refuse to mount the filesystem. 1101 * 1102 * e2fsck's requirements are more strict; if it doesn't know 1103 * about a feature in either the compatible or incompatible 1104 * feature set, it must abort and not try to meddle with 1105 * things it doesn't understand... 1106 */ 1107 __le32 s_first_ino; /* First non-reserved inode */ 1108 __le16 s_inode_size; /* size of inode structure */ 1109 __le16 s_block_group_nr; /* block group # of this superblock */ 1110 __le32 s_feature_compat; /* compatible feature set */ 1111 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1112 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1113 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1114 /*78*/ char s_volume_name[16]; /* volume name */ 1115 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1116 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1117 /* 1118 * Performance hints. Directory preallocation should only 1119 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1120 */ 1121 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1122 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1123 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1124 /* 1125 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1126 */ 1127 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1128 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1129 __le32 s_journal_dev; /* device number of journal file */ 1130 __le32 s_last_orphan; /* start of list of inodes to delete */ 1131 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1132 __u8 s_def_hash_version; /* Default hash version to use */ 1133 __u8 s_jnl_backup_type; 1134 __le16 s_desc_size; /* size of group descriptor */ 1135 /*100*/ __le32 s_default_mount_opts; 1136 __le32 s_first_meta_bg; /* First metablock block group */ 1137 __le32 s_mkfs_time; /* When the filesystem was created */ 1138 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1139 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1140 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1141 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1142 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1143 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1144 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1145 __le32 s_flags; /* Miscellaneous flags */ 1146 __le16 s_raid_stride; /* RAID stride */ 1147 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1148 __le64 s_mmp_block; /* Block for multi-mount protection */ 1149 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1150 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1151 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1152 __u8 s_encryption_level; /* versioning level for encryption */ 1153 __u8 s_reserved_pad; /* Padding to next 32bits */ 1154 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1155 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1156 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1157 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1158 snapshot's future use */ 1159 __le32 s_snapshot_list; /* inode number of the head of the 1160 on-disk snapshot list */ 1161 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1162 __le32 s_error_count; /* number of fs errors */ 1163 __le32 s_first_error_time; /* first time an error happened */ 1164 __le32 s_first_error_ino; /* inode involved in first error */ 1165 __le64 s_first_error_block; /* block involved of first error */ 1166 __u8 s_first_error_func[32]; /* function where the error happened */ 1167 __le32 s_first_error_line; /* line number where error happened */ 1168 __le32 s_last_error_time; /* most recent time of an error */ 1169 __le32 s_last_error_ino; /* inode involved in last error */ 1170 __le32 s_last_error_line; /* line number where error happened */ 1171 __le64 s_last_error_block; /* block involved of last error */ 1172 __u8 s_last_error_func[32]; /* function where the error happened */ 1173 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1174 __u8 s_mount_opts[64]; 1175 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1176 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1177 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1178 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1179 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1180 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1181 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1182 __le32 s_reserved[100]; /* Padding to the end of the block */ 1183 __le32 s_checksum; /* crc32c(superblock) */ 1184 }; 1185 1186 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1187 1188 #ifdef __KERNEL__ 1189 1190 /* 1191 * run-time mount flags 1192 */ 1193 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1194 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1195 1196 /* Number of quota types we support */ 1197 #define EXT4_MAXQUOTAS 2 1198 1199 /* 1200 * fourth extended-fs super-block data in memory 1201 */ 1202 struct ext4_sb_info { 1203 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1204 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1205 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1206 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1207 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1208 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1209 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1210 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1211 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1212 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1213 unsigned long s_overhead; /* # of fs overhead clusters */ 1214 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1215 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1216 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1217 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1218 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1219 struct buffer_head **s_group_desc; 1220 unsigned int s_mount_opt; 1221 unsigned int s_mount_opt2; 1222 unsigned int s_mount_flags; 1223 unsigned int s_def_mount_opt; 1224 ext4_fsblk_t s_sb_block; 1225 atomic64_t s_resv_clusters; 1226 kuid_t s_resuid; 1227 kgid_t s_resgid; 1228 unsigned short s_mount_state; 1229 unsigned short s_pad; 1230 int s_addr_per_block_bits; 1231 int s_desc_per_block_bits; 1232 int s_inode_size; 1233 int s_first_ino; 1234 unsigned int s_inode_readahead_blks; 1235 unsigned int s_inode_goal; 1236 spinlock_t s_next_gen_lock; 1237 u32 s_next_generation; 1238 u32 s_hash_seed[4]; 1239 int s_def_hash_version; 1240 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1241 struct percpu_counter s_freeclusters_counter; 1242 struct percpu_counter s_freeinodes_counter; 1243 struct percpu_counter s_dirs_counter; 1244 struct percpu_counter s_dirtyclusters_counter; 1245 struct blockgroup_lock *s_blockgroup_lock; 1246 struct proc_dir_entry *s_proc; 1247 struct kobject s_kobj; 1248 struct completion s_kobj_unregister; 1249 struct super_block *s_sb; 1250 1251 /* Journaling */ 1252 struct journal_s *s_journal; 1253 struct list_head s_orphan; 1254 struct mutex s_orphan_lock; 1255 unsigned long s_resize_flags; /* Flags indicating if there 1256 is a resizer */ 1257 unsigned long s_commit_interval; 1258 u32 s_max_batch_time; 1259 u32 s_min_batch_time; 1260 struct block_device *journal_bdev; 1261 #ifdef CONFIG_QUOTA 1262 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1263 int s_jquota_fmt; /* Format of quota to use */ 1264 #endif 1265 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1266 struct rb_root system_blks; 1267 1268 #ifdef EXTENTS_STATS 1269 /* ext4 extents stats */ 1270 unsigned long s_ext_min; 1271 unsigned long s_ext_max; 1272 unsigned long s_depth_max; 1273 spinlock_t s_ext_stats_lock; 1274 unsigned long s_ext_blocks; 1275 unsigned long s_ext_extents; 1276 #endif 1277 1278 /* for buddy allocator */ 1279 struct ext4_group_info ***s_group_info; 1280 struct inode *s_buddy_cache; 1281 spinlock_t s_md_lock; 1282 unsigned short *s_mb_offsets; 1283 unsigned int *s_mb_maxs; 1284 unsigned int s_group_info_size; 1285 1286 /* tunables */ 1287 unsigned long s_stripe; 1288 unsigned int s_mb_stream_request; 1289 unsigned int s_mb_max_to_scan; 1290 unsigned int s_mb_min_to_scan; 1291 unsigned int s_mb_stats; 1292 unsigned int s_mb_order2_reqs; 1293 unsigned int s_mb_group_prealloc; 1294 unsigned int s_max_dir_size_kb; 1295 /* where last allocation was done - for stream allocation */ 1296 unsigned long s_mb_last_group; 1297 unsigned long s_mb_last_start; 1298 1299 /* stats for buddy allocator */ 1300 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1301 atomic_t s_bal_success; /* we found long enough chunks */ 1302 atomic_t s_bal_allocated; /* in blocks */ 1303 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1304 atomic_t s_bal_goals; /* goal hits */ 1305 atomic_t s_bal_breaks; /* too long searches */ 1306 atomic_t s_bal_2orders; /* 2^order hits */ 1307 spinlock_t s_bal_lock; 1308 unsigned long s_mb_buddies_generated; 1309 unsigned long long s_mb_generation_time; 1310 atomic_t s_mb_lost_chunks; 1311 atomic_t s_mb_preallocated; 1312 atomic_t s_mb_discarded; 1313 atomic_t s_lock_busy; 1314 1315 /* locality groups */ 1316 struct ext4_locality_group __percpu *s_locality_groups; 1317 1318 /* for write statistics */ 1319 unsigned long s_sectors_written_start; 1320 u64 s_kbytes_written; 1321 1322 /* the size of zero-out chunk */ 1323 unsigned int s_extent_max_zeroout_kb; 1324 1325 unsigned int s_log_groups_per_flex; 1326 struct flex_groups *s_flex_groups; 1327 ext4_group_t s_flex_groups_allocated; 1328 1329 /* workqueue for reserved extent conversions (buffered io) */ 1330 struct workqueue_struct *rsv_conversion_wq; 1331 1332 /* timer for periodic error stats printing */ 1333 struct timer_list s_err_report; 1334 1335 /* Lazy inode table initialization info */ 1336 struct ext4_li_request *s_li_request; 1337 /* Wait multiplier for lazy initialization thread */ 1338 unsigned int s_li_wait_mult; 1339 1340 /* Kernel thread for multiple mount protection */ 1341 struct task_struct *s_mmp_tsk; 1342 1343 /* record the last minlen when FITRIM is called. */ 1344 atomic_t s_last_trim_minblks; 1345 1346 /* Reference to checksum algorithm driver via cryptoapi */ 1347 struct crypto_shash *s_chksum_driver; 1348 1349 /* Precomputed FS UUID checksum for seeding other checksums */ 1350 __u32 s_csum_seed; 1351 1352 /* Reclaim extents from extent status tree */ 1353 struct shrinker s_es_shrinker; 1354 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1355 long s_es_nr_inode; 1356 struct ext4_es_stats s_es_stats; 1357 struct mb_cache *s_mb_cache; 1358 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1359 1360 /* Ratelimit ext4 messages. */ 1361 struct ratelimit_state s_err_ratelimit_state; 1362 struct ratelimit_state s_warning_ratelimit_state; 1363 struct ratelimit_state s_msg_ratelimit_state; 1364 }; 1365 1366 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1367 { 1368 return sb->s_fs_info; 1369 } 1370 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1371 { 1372 return container_of(inode, struct ext4_inode_info, vfs_inode); 1373 } 1374 1375 static inline struct timespec ext4_current_time(struct inode *inode) 1376 { 1377 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1378 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1379 } 1380 1381 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1382 { 1383 return ino == EXT4_ROOT_INO || 1384 ino == EXT4_USR_QUOTA_INO || 1385 ino == EXT4_GRP_QUOTA_INO || 1386 ino == EXT4_BOOT_LOADER_INO || 1387 ino == EXT4_JOURNAL_INO || 1388 ino == EXT4_RESIZE_INO || 1389 (ino >= EXT4_FIRST_INO(sb) && 1390 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1391 } 1392 1393 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1394 struct ext4_io_end *io_end) 1395 { 1396 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1397 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1398 atomic_inc(&EXT4_I(inode)->i_unwritten); 1399 } 1400 } 1401 1402 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1403 { 1404 return inode->i_private; 1405 } 1406 1407 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1408 { 1409 inode->i_private = io; 1410 } 1411 1412 /* 1413 * Inode dynamic state flags 1414 */ 1415 enum { 1416 EXT4_STATE_JDATA, /* journaled data exists */ 1417 EXT4_STATE_NEW, /* inode is newly created */ 1418 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1419 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1420 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1421 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1422 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1423 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1424 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1425 nolocking */ 1426 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1427 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1428 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1429 }; 1430 1431 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1432 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1433 { \ 1434 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1435 } \ 1436 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1437 { \ 1438 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1439 } \ 1440 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1441 { \ 1442 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1443 } 1444 1445 /* Add these declarations here only so that these functions can be 1446 * found by name. Otherwise, they are very hard to locate. */ 1447 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1448 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1449 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1450 EXT4_INODE_BIT_FNS(flag, flags, 0) 1451 1452 /* Add these declarations here only so that these functions can be 1453 * found by name. Otherwise, they are very hard to locate. */ 1454 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1455 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1456 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1457 #if (BITS_PER_LONG < 64) 1458 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1459 1460 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1461 { 1462 (ei)->i_state_flags = 0; 1463 } 1464 #else 1465 EXT4_INODE_BIT_FNS(state, flags, 32) 1466 1467 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1468 { 1469 /* We depend on the fact that callers will set i_flags */ 1470 } 1471 #endif 1472 #else 1473 /* Assume that user mode programs are passing in an ext4fs superblock, not 1474 * a kernel struct super_block. This will allow us to call the feature-test 1475 * macros from user land. */ 1476 #define EXT4_SB(sb) (sb) 1477 #endif 1478 1479 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1480 1481 /* 1482 * Codes for operating systems 1483 */ 1484 #define EXT4_OS_LINUX 0 1485 #define EXT4_OS_HURD 1 1486 #define EXT4_OS_MASIX 2 1487 #define EXT4_OS_FREEBSD 3 1488 #define EXT4_OS_LITES 4 1489 1490 /* 1491 * Revision levels 1492 */ 1493 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1494 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1495 1496 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1497 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1498 1499 #define EXT4_GOOD_OLD_INODE_SIZE 128 1500 1501 /* 1502 * Feature set definitions 1503 */ 1504 1505 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1506 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1507 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1508 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1509 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1510 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1511 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1512 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1513 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1514 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1515 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1516 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1517 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1518 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1519 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1520 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1521 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1522 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1523 1524 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1525 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1526 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1527 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1528 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1529 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1530 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1531 1532 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1533 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1534 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1535 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1536 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1537 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1538 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1539 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1540 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1541 /* 1542 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1543 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1544 * all other data structures' checksums. However, the METADATA_CSUM and 1545 * GDT_CSUM bits are mutually exclusive. 1546 */ 1547 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1548 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1549 1550 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1551 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1552 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1553 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1554 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1555 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1556 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1557 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1558 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1559 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1560 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1561 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1562 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1563 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1564 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1565 1566 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1567 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1568 EXT4_FEATURE_INCOMPAT_META_BG) 1569 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1570 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1571 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1572 1573 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1574 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1575 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1576 EXT4_FEATURE_INCOMPAT_META_BG) 1577 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1578 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1579 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1580 1581 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1582 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1583 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1584 EXT4_FEATURE_INCOMPAT_META_BG| \ 1585 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1586 EXT4_FEATURE_INCOMPAT_64BIT| \ 1587 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1588 EXT4_FEATURE_INCOMPAT_MMP | \ 1589 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1590 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1591 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1592 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1593 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1594 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1595 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1596 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1597 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1598 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1599 EXT4_FEATURE_RO_COMPAT_QUOTA) 1600 1601 /* 1602 * Default values for user and/or group using reserved blocks 1603 */ 1604 #define EXT4_DEF_RESUID 0 1605 #define EXT4_DEF_RESGID 0 1606 1607 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1608 1609 /* 1610 * Default mount options 1611 */ 1612 #define EXT4_DEFM_DEBUG 0x0001 1613 #define EXT4_DEFM_BSDGROUPS 0x0002 1614 #define EXT4_DEFM_XATTR_USER 0x0004 1615 #define EXT4_DEFM_ACL 0x0008 1616 #define EXT4_DEFM_UID16 0x0010 1617 #define EXT4_DEFM_JMODE 0x0060 1618 #define EXT4_DEFM_JMODE_DATA 0x0020 1619 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1620 #define EXT4_DEFM_JMODE_WBACK 0x0060 1621 #define EXT4_DEFM_NOBARRIER 0x0100 1622 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1623 #define EXT4_DEFM_DISCARD 0x0400 1624 #define EXT4_DEFM_NODELALLOC 0x0800 1625 1626 /* 1627 * Default journal batch times 1628 */ 1629 #define EXT4_DEF_MIN_BATCH_TIME 0 1630 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1631 1632 /* 1633 * Minimum number of groups in a flexgroup before we separate out 1634 * directories into the first block group of a flexgroup 1635 */ 1636 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1637 1638 /* 1639 * Structure of a directory entry 1640 */ 1641 #define EXT4_NAME_LEN 255 1642 1643 struct ext4_dir_entry { 1644 __le32 inode; /* Inode number */ 1645 __le16 rec_len; /* Directory entry length */ 1646 __le16 name_len; /* Name length */ 1647 char name[EXT4_NAME_LEN]; /* File name */ 1648 }; 1649 1650 /* 1651 * The new version of the directory entry. Since EXT4 structures are 1652 * stored in intel byte order, and the name_len field could never be 1653 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1654 * file_type field. 1655 */ 1656 struct ext4_dir_entry_2 { 1657 __le32 inode; /* Inode number */ 1658 __le16 rec_len; /* Directory entry length */ 1659 __u8 name_len; /* Name length */ 1660 __u8 file_type; 1661 char name[EXT4_NAME_LEN]; /* File name */ 1662 }; 1663 1664 /* 1665 * This is a bogus directory entry at the end of each leaf block that 1666 * records checksums. 1667 */ 1668 struct ext4_dir_entry_tail { 1669 __le32 det_reserved_zero1; /* Pretend to be unused */ 1670 __le16 det_rec_len; /* 12 */ 1671 __u8 det_reserved_zero2; /* Zero name length */ 1672 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1673 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1674 }; 1675 1676 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1677 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1678 ((blocksize) - \ 1679 sizeof(struct ext4_dir_entry_tail)))) 1680 1681 /* 1682 * Ext4 directory file types. Only the low 3 bits are used. The 1683 * other bits are reserved for now. 1684 */ 1685 #define EXT4_FT_UNKNOWN 0 1686 #define EXT4_FT_REG_FILE 1 1687 #define EXT4_FT_DIR 2 1688 #define EXT4_FT_CHRDEV 3 1689 #define EXT4_FT_BLKDEV 4 1690 #define EXT4_FT_FIFO 5 1691 #define EXT4_FT_SOCK 6 1692 #define EXT4_FT_SYMLINK 7 1693 1694 #define EXT4_FT_MAX 8 1695 1696 #define EXT4_FT_DIR_CSUM 0xDE 1697 1698 /* 1699 * EXT4_DIR_PAD defines the directory entries boundaries 1700 * 1701 * NOTE: It must be a multiple of 4 1702 */ 1703 #define EXT4_DIR_PAD 4 1704 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1705 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1706 ~EXT4_DIR_ROUND) 1707 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1708 1709 /* 1710 * If we ever get support for fs block sizes > page_size, we'll need 1711 * to remove the #if statements in the next two functions... 1712 */ 1713 static inline unsigned int 1714 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1715 { 1716 unsigned len = le16_to_cpu(dlen); 1717 1718 #if (PAGE_CACHE_SIZE >= 65536) 1719 if (len == EXT4_MAX_REC_LEN || len == 0) 1720 return blocksize; 1721 return (len & 65532) | ((len & 3) << 16); 1722 #else 1723 return len; 1724 #endif 1725 } 1726 1727 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1728 { 1729 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1730 BUG(); 1731 #if (PAGE_CACHE_SIZE >= 65536) 1732 if (len < 65536) 1733 return cpu_to_le16(len); 1734 if (len == blocksize) { 1735 if (blocksize == 65536) 1736 return cpu_to_le16(EXT4_MAX_REC_LEN); 1737 else 1738 return cpu_to_le16(0); 1739 } 1740 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1741 #else 1742 return cpu_to_le16(len); 1743 #endif 1744 } 1745 1746 /* 1747 * Hash Tree Directory indexing 1748 * (c) Daniel Phillips, 2001 1749 */ 1750 1751 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1752 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1753 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1754 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1755 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1756 1757 /* Legal values for the dx_root hash_version field: */ 1758 1759 #define DX_HASH_LEGACY 0 1760 #define DX_HASH_HALF_MD4 1 1761 #define DX_HASH_TEA 2 1762 #define DX_HASH_LEGACY_UNSIGNED 3 1763 #define DX_HASH_HALF_MD4_UNSIGNED 4 1764 #define DX_HASH_TEA_UNSIGNED 5 1765 1766 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1767 const void *address, unsigned int length) 1768 { 1769 struct { 1770 struct shash_desc shash; 1771 char ctx[4]; 1772 } desc; 1773 int err; 1774 1775 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1776 1777 desc.shash.tfm = sbi->s_chksum_driver; 1778 desc.shash.flags = 0; 1779 *(u32 *)desc.ctx = crc; 1780 1781 err = crypto_shash_update(&desc.shash, address, length); 1782 BUG_ON(err); 1783 1784 return *(u32 *)desc.ctx; 1785 } 1786 1787 #ifdef __KERNEL__ 1788 1789 /* hash info structure used by the directory hash */ 1790 struct dx_hash_info 1791 { 1792 u32 hash; 1793 u32 minor_hash; 1794 int hash_version; 1795 u32 *seed; 1796 }; 1797 1798 1799 /* 32 and 64 bit signed EOF for dx directories */ 1800 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1801 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1802 1803 1804 /* 1805 * Control parameters used by ext4_htree_next_block 1806 */ 1807 #define HASH_NB_ALWAYS 1 1808 1809 1810 /* 1811 * Describe an inode's exact location on disk and in memory 1812 */ 1813 struct ext4_iloc 1814 { 1815 struct buffer_head *bh; 1816 unsigned long offset; 1817 ext4_group_t block_group; 1818 }; 1819 1820 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1821 { 1822 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1823 } 1824 1825 /* 1826 * This structure is stuffed into the struct file's private_data field 1827 * for directories. It is where we put information so that we can do 1828 * readdir operations in hash tree order. 1829 */ 1830 struct dir_private_info { 1831 struct rb_root root; 1832 struct rb_node *curr_node; 1833 struct fname *extra_fname; 1834 loff_t last_pos; 1835 __u32 curr_hash; 1836 __u32 curr_minor_hash; 1837 __u32 next_hash; 1838 }; 1839 1840 /* calculate the first block number of the group */ 1841 static inline ext4_fsblk_t 1842 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1843 { 1844 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1845 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1846 } 1847 1848 /* 1849 * Special error return code only used by dx_probe() and its callers. 1850 */ 1851 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 1852 1853 /* 1854 * Timeout and state flag for lazy initialization inode thread. 1855 */ 1856 #define EXT4_DEF_LI_WAIT_MULT 10 1857 #define EXT4_DEF_LI_MAX_START_DELAY 5 1858 #define EXT4_LAZYINIT_QUIT 0x0001 1859 #define EXT4_LAZYINIT_RUNNING 0x0002 1860 1861 /* 1862 * Lazy inode table initialization info 1863 */ 1864 struct ext4_lazy_init { 1865 unsigned long li_state; 1866 struct list_head li_request_list; 1867 struct mutex li_list_mtx; 1868 }; 1869 1870 struct ext4_li_request { 1871 struct super_block *lr_super; 1872 struct ext4_sb_info *lr_sbi; 1873 ext4_group_t lr_next_group; 1874 struct list_head lr_request; 1875 unsigned long lr_next_sched; 1876 unsigned long lr_timeout; 1877 }; 1878 1879 struct ext4_features { 1880 struct kobject f_kobj; 1881 struct completion f_kobj_unregister; 1882 }; 1883 1884 /* 1885 * This structure will be used for multiple mount protection. It will be 1886 * written into the block number saved in the s_mmp_block field in the 1887 * superblock. Programs that check MMP should assume that if 1888 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1889 * to use the filesystem, regardless of how old the timestamp is. 1890 */ 1891 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1892 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1893 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1894 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1895 1896 struct mmp_struct { 1897 __le32 mmp_magic; /* Magic number for MMP */ 1898 __le32 mmp_seq; /* Sequence no. updated periodically */ 1899 1900 /* 1901 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1902 * purposes and do not affect the correctness of the algorithm 1903 */ 1904 __le64 mmp_time; /* Time last updated */ 1905 char mmp_nodename[64]; /* Node which last updated MMP block */ 1906 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1907 1908 /* 1909 * mmp_check_interval is used to verify if the MMP block has been 1910 * updated on the block device. The value is updated based on the 1911 * maximum time to write the MMP block during an update cycle. 1912 */ 1913 __le16 mmp_check_interval; 1914 1915 __le16 mmp_pad1; 1916 __le32 mmp_pad2[226]; 1917 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1918 }; 1919 1920 /* arguments passed to the mmp thread */ 1921 struct mmpd_data { 1922 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1923 struct super_block *sb; /* super block of the fs */ 1924 }; 1925 1926 /* 1927 * Check interval multiplier 1928 * The MMP block is written every update interval and initially checked every 1929 * update interval x the multiplier (the value is then adapted based on the 1930 * write latency). The reason is that writes can be delayed under load and we 1931 * don't want readers to incorrectly assume that the filesystem is no longer 1932 * in use. 1933 */ 1934 #define EXT4_MMP_CHECK_MULT 2UL 1935 1936 /* 1937 * Minimum interval for MMP checking in seconds. 1938 */ 1939 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1940 1941 /* 1942 * Maximum interval for MMP checking in seconds. 1943 */ 1944 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1945 1946 /* 1947 * Function prototypes 1948 */ 1949 1950 /* 1951 * Ok, these declarations are also in <linux/kernel.h> but none of the 1952 * ext4 source programs needs to include it so they are duplicated here. 1953 */ 1954 # define NORET_TYPE /**/ 1955 # define ATTRIB_NORET __attribute__((noreturn)) 1956 # define NORET_AND noreturn, 1957 1958 /* bitmap.c */ 1959 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1960 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1961 struct ext4_group_desc *gdp, 1962 struct buffer_head *bh, int sz); 1963 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1964 struct ext4_group_desc *gdp, 1965 struct buffer_head *bh, int sz); 1966 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1967 struct ext4_group_desc *gdp, 1968 struct buffer_head *bh); 1969 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1970 struct ext4_group_desc *gdp, 1971 struct buffer_head *bh); 1972 1973 /* balloc.c */ 1974 extern void ext4_get_group_no_and_offset(struct super_block *sb, 1975 ext4_fsblk_t blocknr, 1976 ext4_group_t *blockgrpp, 1977 ext4_grpblk_t *offsetp); 1978 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 1979 ext4_fsblk_t block); 1980 1981 extern unsigned int ext4_block_group(struct super_block *sb, 1982 ext4_fsblk_t blocknr); 1983 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1984 ext4_fsblk_t blocknr); 1985 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1986 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1987 ext4_group_t group); 1988 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1989 ext4_fsblk_t goal, 1990 unsigned int flags, 1991 unsigned long *count, 1992 int *errp); 1993 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1994 s64 nclusters, unsigned int flags); 1995 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1996 extern void ext4_check_blocks_bitmap(struct super_block *); 1997 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1998 ext4_group_t block_group, 1999 struct buffer_head ** bh); 2000 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2001 2002 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2003 ext4_group_t block_group); 2004 extern int ext4_wait_block_bitmap(struct super_block *sb, 2005 ext4_group_t block_group, 2006 struct buffer_head *bh); 2007 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2008 ext4_group_t block_group); 2009 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2010 ext4_group_t block_group, 2011 struct ext4_group_desc *gdp); 2012 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2013 2014 /* dir.c */ 2015 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2016 struct file *, 2017 struct ext4_dir_entry_2 *, 2018 struct buffer_head *, char *, int, 2019 unsigned int); 2020 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2021 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2022 (de), (bh), (buf), (size), (offset))) 2023 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2024 __u32 minor_hash, 2025 struct ext4_dir_entry_2 *dirent); 2026 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2027 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2028 struct buffer_head *bh, 2029 void *buf, int buf_size, 2030 const char *name, int namelen, 2031 struct ext4_dir_entry_2 **dest_de); 2032 void ext4_insert_dentry(struct inode *inode, 2033 struct ext4_dir_entry_2 *de, 2034 int buf_size, 2035 const char *name, int namelen); 2036 static inline void ext4_update_dx_flag(struct inode *inode) 2037 { 2038 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 2039 EXT4_FEATURE_COMPAT_DIR_INDEX)) 2040 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2041 } 2042 static unsigned char ext4_filetype_table[] = { 2043 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2044 }; 2045 2046 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2047 { 2048 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 2049 (filetype >= EXT4_FT_MAX)) 2050 return DT_UNKNOWN; 2051 2052 return ext4_filetype_table[filetype]; 2053 } 2054 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2055 void *buf, int buf_size); 2056 2057 /* fsync.c */ 2058 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2059 2060 /* hash.c */ 2061 extern int ext4fs_dirhash(const char *name, int len, struct 2062 dx_hash_info *hinfo); 2063 2064 /* ialloc.c */ 2065 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2066 const struct qstr *qstr, __u32 goal, 2067 uid_t *owner, int handle_type, 2068 unsigned int line_no, int nblocks); 2069 2070 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2071 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2072 0, 0, 0) 2073 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2074 type, nblocks) \ 2075 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2076 (type), __LINE__, (nblocks)) 2077 2078 2079 extern void ext4_free_inode(handle_t *, struct inode *); 2080 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2081 extern unsigned long ext4_count_free_inodes(struct super_block *); 2082 extern unsigned long ext4_count_dirs(struct super_block *); 2083 extern void ext4_check_inodes_bitmap(struct super_block *); 2084 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2085 extern int ext4_init_inode_table(struct super_block *sb, 2086 ext4_group_t group, int barrier); 2087 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2088 2089 /* mballoc.c */ 2090 extern long ext4_mb_stats; 2091 extern long ext4_mb_max_to_scan; 2092 extern int ext4_mb_init(struct super_block *); 2093 extern int ext4_mb_release(struct super_block *); 2094 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2095 struct ext4_allocation_request *, int *); 2096 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2097 extern void ext4_discard_preallocations(struct inode *); 2098 extern int __init ext4_init_mballoc(void); 2099 extern void ext4_exit_mballoc(void); 2100 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2101 struct buffer_head *bh, ext4_fsblk_t block, 2102 unsigned long count, int flags); 2103 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2104 ext4_group_t ngroups); 2105 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2106 ext4_group_t i, struct ext4_group_desc *desc); 2107 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2108 ext4_fsblk_t block, unsigned long count); 2109 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2110 2111 /* inode.c */ 2112 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2113 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2114 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2115 struct buffer_head *bh_result, int create); 2116 int ext4_get_block(struct inode *inode, sector_t iblock, 2117 struct buffer_head *bh_result, int create); 2118 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2119 struct buffer_head *bh, int create); 2120 int ext4_walk_page_buffers(handle_t *handle, 2121 struct buffer_head *head, 2122 unsigned from, 2123 unsigned to, 2124 int *partial, 2125 int (*fn)(handle_t *handle, 2126 struct buffer_head *bh)); 2127 int do_journal_get_write_access(handle_t *handle, 2128 struct buffer_head *bh); 2129 #define FALL_BACK_TO_NONDELALLOC 1 2130 #define CONVERT_INLINE_DATA 2 2131 2132 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2133 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2134 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2135 extern int ext4_setattr(struct dentry *, struct iattr *); 2136 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2137 struct kstat *stat); 2138 extern void ext4_evict_inode(struct inode *); 2139 extern void ext4_clear_inode(struct inode *); 2140 extern int ext4_sync_inode(handle_t *, struct inode *); 2141 extern void ext4_dirty_inode(struct inode *, int); 2142 extern int ext4_change_inode_journal_flag(struct inode *, int); 2143 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2144 extern int ext4_inode_attach_jinode(struct inode *inode); 2145 extern int ext4_can_truncate(struct inode *inode); 2146 extern void ext4_truncate(struct inode *); 2147 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2148 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2149 extern void ext4_set_inode_flags(struct inode *); 2150 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2151 extern int ext4_alloc_da_blocks(struct inode *inode); 2152 extern void ext4_set_aops(struct inode *inode); 2153 extern int ext4_writepage_trans_blocks(struct inode *); 2154 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2155 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2156 loff_t lstart, loff_t lend); 2157 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2158 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2159 extern void ext4_da_update_reserve_space(struct inode *inode, 2160 int used, int quota_claim); 2161 2162 /* indirect.c */ 2163 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2164 struct ext4_map_blocks *map, int flags); 2165 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2166 struct iov_iter *iter, loff_t offset); 2167 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2168 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2169 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2170 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2171 ext4_lblk_t start, ext4_lblk_t end); 2172 2173 /* ioctl.c */ 2174 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2175 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2176 2177 /* migrate.c */ 2178 extern int ext4_ext_migrate(struct inode *); 2179 extern int ext4_ind_migrate(struct inode *inode); 2180 2181 /* namei.c */ 2182 extern int ext4_dirent_csum_verify(struct inode *inode, 2183 struct ext4_dir_entry *dirent); 2184 extern int ext4_orphan_add(handle_t *, struct inode *); 2185 extern int ext4_orphan_del(handle_t *, struct inode *); 2186 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2187 __u32 start_minor_hash, __u32 *next_hash); 2188 extern int search_dir(struct buffer_head *bh, 2189 char *search_buf, 2190 int buf_size, 2191 struct inode *dir, 2192 const struct qstr *d_name, 2193 unsigned int offset, 2194 struct ext4_dir_entry_2 **res_dir); 2195 extern int ext4_generic_delete_entry(handle_t *handle, 2196 struct inode *dir, 2197 struct ext4_dir_entry_2 *de_del, 2198 struct buffer_head *bh, 2199 void *entry_buf, 2200 int buf_size, 2201 int csum_size); 2202 2203 /* resize.c */ 2204 extern int ext4_group_add(struct super_block *sb, 2205 struct ext4_new_group_data *input); 2206 extern int ext4_group_extend(struct super_block *sb, 2207 struct ext4_super_block *es, 2208 ext4_fsblk_t n_blocks_count); 2209 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2210 2211 /* super.c */ 2212 extern int ext4_calculate_overhead(struct super_block *sb); 2213 extern void ext4_superblock_csum_set(struct super_block *sb); 2214 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2215 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2216 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2217 ext4_group_t ngroup); 2218 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2219 char nbuf[16]); 2220 2221 extern __printf(4, 5) 2222 void __ext4_error(struct super_block *, const char *, unsigned int, 2223 const char *, ...); 2224 extern __printf(5, 6) 2225 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2226 const char *, ...); 2227 extern __printf(5, 6) 2228 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2229 const char *, ...); 2230 extern void __ext4_std_error(struct super_block *, const char *, 2231 unsigned int, int); 2232 extern __printf(4, 5) 2233 void __ext4_abort(struct super_block *, const char *, unsigned int, 2234 const char *, ...); 2235 extern __printf(4, 5) 2236 void __ext4_warning(struct super_block *, const char *, unsigned int, 2237 const char *, ...); 2238 extern __printf(3, 4) 2239 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2240 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2241 const char *, unsigned int, const char *); 2242 extern __printf(7, 8) 2243 void __ext4_grp_locked_error(const char *, unsigned int, 2244 struct super_block *, ext4_group_t, 2245 unsigned long, ext4_fsblk_t, 2246 const char *, ...); 2247 2248 #ifdef CONFIG_PRINTK 2249 2250 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2251 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2252 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2253 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2254 #define ext4_error(sb, fmt, ...) \ 2255 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2256 #define ext4_abort(sb, fmt, ...) \ 2257 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2258 #define ext4_warning(sb, fmt, ...) \ 2259 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2260 #define ext4_msg(sb, level, fmt, ...) \ 2261 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2262 #define dump_mmp_msg(sb, mmp, msg) \ 2263 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2264 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2265 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2266 fmt, ##__VA_ARGS__) 2267 2268 #else 2269 2270 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2271 do { \ 2272 no_printk(fmt, ##__VA_ARGS__); \ 2273 __ext4_error_inode(inode, "", 0, block, " "); \ 2274 } while (0) 2275 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2276 do { \ 2277 no_printk(fmt, ##__VA_ARGS__); \ 2278 __ext4_error_file(file, "", 0, block, " "); \ 2279 } while (0) 2280 #define ext4_error(sb, fmt, ...) \ 2281 do { \ 2282 no_printk(fmt, ##__VA_ARGS__); \ 2283 __ext4_error(sb, "", 0, " "); \ 2284 } while (0) 2285 #define ext4_abort(sb, fmt, ...) \ 2286 do { \ 2287 no_printk(fmt, ##__VA_ARGS__); \ 2288 __ext4_abort(sb, "", 0, " "); \ 2289 } while (0) 2290 #define ext4_warning(sb, fmt, ...) \ 2291 do { \ 2292 no_printk(fmt, ##__VA_ARGS__); \ 2293 __ext4_warning(sb, "", 0, " "); \ 2294 } while (0) 2295 #define ext4_msg(sb, level, fmt, ...) \ 2296 do { \ 2297 no_printk(fmt, ##__VA_ARGS__); \ 2298 __ext4_msg(sb, "", " "); \ 2299 } while (0) 2300 #define dump_mmp_msg(sb, mmp, msg) \ 2301 __dump_mmp_msg(sb, mmp, "", 0, "") 2302 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2303 do { \ 2304 no_printk(fmt, ##__VA_ARGS__); \ 2305 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2306 } while (0) 2307 2308 #endif 2309 2310 extern void ext4_update_dynamic_rev(struct super_block *sb); 2311 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2312 __u32 compat); 2313 extern int ext4_update_rocompat_feature(handle_t *handle, 2314 struct super_block *sb, __u32 rocompat); 2315 extern int ext4_update_incompat_feature(handle_t *handle, 2316 struct super_block *sb, __u32 incompat); 2317 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2318 struct ext4_group_desc *bg); 2319 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2320 struct ext4_group_desc *bg); 2321 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2322 struct ext4_group_desc *bg); 2323 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2324 struct ext4_group_desc *bg); 2325 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2326 struct ext4_group_desc *bg); 2327 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2328 struct ext4_group_desc *bg); 2329 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2330 struct ext4_group_desc *bg); 2331 extern void ext4_block_bitmap_set(struct super_block *sb, 2332 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2333 extern void ext4_inode_bitmap_set(struct super_block *sb, 2334 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2335 extern void ext4_inode_table_set(struct super_block *sb, 2336 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2337 extern void ext4_free_group_clusters_set(struct super_block *sb, 2338 struct ext4_group_desc *bg, 2339 __u32 count); 2340 extern void ext4_free_inodes_set(struct super_block *sb, 2341 struct ext4_group_desc *bg, __u32 count); 2342 extern void ext4_used_dirs_set(struct super_block *sb, 2343 struct ext4_group_desc *bg, __u32 count); 2344 extern void ext4_itable_unused_set(struct super_block *sb, 2345 struct ext4_group_desc *bg, __u32 count); 2346 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2347 struct ext4_group_desc *gdp); 2348 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2349 struct ext4_group_desc *gdp); 2350 extern int ext4_register_li_request(struct super_block *sb, 2351 ext4_group_t first_not_zeroed); 2352 2353 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2354 { 2355 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2356 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) || 2357 (EXT4_SB(sb)->s_chksum_driver != NULL); 2358 } 2359 2360 static inline int ext4_has_metadata_csum(struct super_block *sb) 2361 { 2362 WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb, 2363 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) && 2364 !EXT4_SB(sb)->s_chksum_driver); 2365 2366 return (EXT4_SB(sb)->s_chksum_driver != NULL); 2367 } 2368 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2369 { 2370 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2371 le32_to_cpu(es->s_blocks_count_lo); 2372 } 2373 2374 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2375 { 2376 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2377 le32_to_cpu(es->s_r_blocks_count_lo); 2378 } 2379 2380 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2381 { 2382 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2383 le32_to_cpu(es->s_free_blocks_count_lo); 2384 } 2385 2386 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2387 ext4_fsblk_t blk) 2388 { 2389 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2390 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2391 } 2392 2393 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2394 ext4_fsblk_t blk) 2395 { 2396 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2397 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2398 } 2399 2400 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2401 ext4_fsblk_t blk) 2402 { 2403 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2404 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2405 } 2406 2407 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2408 { 2409 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2410 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2411 le32_to_cpu(raw_inode->i_size_lo); 2412 else 2413 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2414 } 2415 2416 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2417 { 2418 raw_inode->i_size_lo = cpu_to_le32(i_size); 2419 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2420 } 2421 2422 static inline 2423 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2424 ext4_group_t group) 2425 { 2426 struct ext4_group_info ***grp_info; 2427 long indexv, indexh; 2428 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2429 grp_info = EXT4_SB(sb)->s_group_info; 2430 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2431 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2432 return grp_info[indexv][indexh]; 2433 } 2434 2435 /* 2436 * Reading s_groups_count requires using smp_rmb() afterwards. See 2437 * the locking protocol documented in the comments of ext4_group_add() 2438 * in resize.c 2439 */ 2440 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2441 { 2442 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2443 2444 smp_rmb(); 2445 return ngroups; 2446 } 2447 2448 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2449 ext4_group_t block_group) 2450 { 2451 return block_group >> sbi->s_log_groups_per_flex; 2452 } 2453 2454 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2455 { 2456 return 1 << sbi->s_log_groups_per_flex; 2457 } 2458 2459 #define ext4_std_error(sb, errno) \ 2460 do { \ 2461 if ((errno)) \ 2462 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2463 } while (0) 2464 2465 #ifdef CONFIG_SMP 2466 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2467 * counters. So we need to make sure we have free clusters more 2468 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2469 */ 2470 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2471 #else 2472 #define EXT4_FREECLUSTERS_WATERMARK 0 2473 #endif 2474 2475 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2476 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2477 { 2478 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2479 !mutex_is_locked(&inode->i_mutex)); 2480 down_write(&EXT4_I(inode)->i_data_sem); 2481 if (newsize > EXT4_I(inode)->i_disksize) 2482 EXT4_I(inode)->i_disksize = newsize; 2483 up_write(&EXT4_I(inode)->i_data_sem); 2484 } 2485 2486 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2487 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2488 { 2489 int changed = 0; 2490 2491 if (newsize > inode->i_size) { 2492 i_size_write(inode, newsize); 2493 changed = 1; 2494 } 2495 if (newsize > EXT4_I(inode)->i_disksize) { 2496 ext4_update_i_disksize(inode, newsize); 2497 changed |= 2; 2498 } 2499 return changed; 2500 } 2501 2502 struct ext4_group_info { 2503 unsigned long bb_state; 2504 struct rb_root bb_free_root; 2505 ext4_grpblk_t bb_first_free; /* first free block */ 2506 ext4_grpblk_t bb_free; /* total free blocks */ 2507 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2508 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2509 struct list_head bb_prealloc_list; 2510 #ifdef DOUBLE_CHECK 2511 void *bb_bitmap; 2512 #endif 2513 struct rw_semaphore alloc_sem; 2514 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2515 * regions, index is order. 2516 * bb_counters[3] = 5 means 2517 * 5 free 8-block regions. */ 2518 }; 2519 2520 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2521 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2522 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2523 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2524 2525 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2526 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2527 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2528 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2529 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2530 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2531 2532 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2533 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2534 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2535 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2536 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2537 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2538 2539 #define EXT4_MAX_CONTENTION 8 2540 #define EXT4_CONTENTION_THRESHOLD 2 2541 2542 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2543 ext4_group_t group) 2544 { 2545 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2546 } 2547 2548 /* 2549 * Returns true if the filesystem is busy enough that attempts to 2550 * access the block group locks has run into contention. 2551 */ 2552 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2553 { 2554 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2555 } 2556 2557 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2558 { 2559 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2560 if (spin_trylock(lock)) 2561 /* 2562 * We're able to grab the lock right away, so drop the 2563 * lock contention counter. 2564 */ 2565 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2566 else { 2567 /* 2568 * The lock is busy, so bump the contention counter, 2569 * and then wait on the spin lock. 2570 */ 2571 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2572 EXT4_MAX_CONTENTION); 2573 spin_lock(lock); 2574 } 2575 } 2576 2577 static inline void ext4_unlock_group(struct super_block *sb, 2578 ext4_group_t group) 2579 { 2580 spin_unlock(ext4_group_lock_ptr(sb, group)); 2581 } 2582 2583 /* 2584 * Block validity checking 2585 */ 2586 #define ext4_check_indirect_blockref(inode, bh) \ 2587 ext4_check_blockref(__func__, __LINE__, inode, \ 2588 (__le32 *)(bh)->b_data, \ 2589 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2590 2591 #define ext4_ind_check_inode(inode) \ 2592 ext4_check_blockref(__func__, __LINE__, inode, \ 2593 EXT4_I(inode)->i_data, \ 2594 EXT4_NDIR_BLOCKS) 2595 2596 /* 2597 * Inodes and files operations 2598 */ 2599 2600 /* dir.c */ 2601 extern const struct file_operations ext4_dir_operations; 2602 2603 /* file.c */ 2604 extern const struct inode_operations ext4_file_inode_operations; 2605 extern const struct file_operations ext4_file_operations; 2606 extern const struct file_operations ext4_dax_file_operations; 2607 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2608 2609 /* inline.c */ 2610 extern int ext4_get_max_inline_size(struct inode *inode); 2611 extern int ext4_find_inline_data_nolock(struct inode *inode); 2612 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2613 unsigned int len); 2614 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2615 2616 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2617 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2618 struct inode *inode, 2619 loff_t pos, unsigned len, 2620 unsigned flags, 2621 struct page **pagep); 2622 extern int ext4_write_inline_data_end(struct inode *inode, 2623 loff_t pos, unsigned len, 2624 unsigned copied, 2625 struct page *page); 2626 extern struct buffer_head * 2627 ext4_journalled_write_inline_data(struct inode *inode, 2628 unsigned len, 2629 struct page *page); 2630 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2631 struct inode *inode, 2632 loff_t pos, unsigned len, 2633 unsigned flags, 2634 struct page **pagep, 2635 void **fsdata); 2636 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2637 unsigned len, unsigned copied, 2638 struct page *page); 2639 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry, 2640 struct inode *inode); 2641 extern int ext4_try_create_inline_dir(handle_t *handle, 2642 struct inode *parent, 2643 struct inode *inode); 2644 extern int ext4_read_inline_dir(struct file *filp, 2645 struct dir_context *ctx, 2646 int *has_inline_data); 2647 extern int htree_inlinedir_to_tree(struct file *dir_file, 2648 struct inode *dir, ext4_lblk_t block, 2649 struct dx_hash_info *hinfo, 2650 __u32 start_hash, __u32 start_minor_hash, 2651 int *has_inline_data); 2652 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2653 const struct qstr *d_name, 2654 struct ext4_dir_entry_2 **res_dir, 2655 int *has_inline_data); 2656 extern int ext4_delete_inline_entry(handle_t *handle, 2657 struct inode *dir, 2658 struct ext4_dir_entry_2 *de_del, 2659 struct buffer_head *bh, 2660 int *has_inline_data); 2661 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 2662 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 2663 struct ext4_dir_entry_2 **parent_de, 2664 int *retval); 2665 extern int ext4_inline_data_fiemap(struct inode *inode, 2666 struct fiemap_extent_info *fieinfo, 2667 int *has_inline, __u64 start, __u64 len); 2668 extern int ext4_try_to_evict_inline_data(handle_t *handle, 2669 struct inode *inode, 2670 int needed); 2671 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 2672 2673 extern int ext4_convert_inline_data(struct inode *inode); 2674 2675 static inline int ext4_has_inline_data(struct inode *inode) 2676 { 2677 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 2678 EXT4_I(inode)->i_inline_off; 2679 } 2680 2681 /* namei.c */ 2682 extern const struct inode_operations ext4_dir_inode_operations; 2683 extern const struct inode_operations ext4_special_inode_operations; 2684 extern struct dentry *ext4_get_parent(struct dentry *child); 2685 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2686 struct ext4_dir_entry_2 *de, 2687 int blocksize, int csum_size, 2688 unsigned int parent_ino, int dotdot_real_len); 2689 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2690 unsigned int blocksize); 2691 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2692 struct inode *inode, 2693 struct buffer_head *bh); 2694 #define S_SHIFT 12 2695 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 2696 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 2697 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 2698 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 2699 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 2700 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 2701 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 2702 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 2703 }; 2704 2705 static inline void ext4_set_de_type(struct super_block *sb, 2706 struct ext4_dir_entry_2 *de, 2707 umode_t mode) { 2708 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE)) 2709 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 2710 } 2711 2712 /* readpages.c */ 2713 extern int ext4_mpage_readpages(struct address_space *mapping, 2714 struct list_head *pages, struct page *page, 2715 unsigned nr_pages); 2716 2717 /* symlink.c */ 2718 extern const struct inode_operations ext4_symlink_inode_operations; 2719 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2720 2721 /* block_validity */ 2722 extern void ext4_release_system_zone(struct super_block *sb); 2723 extern int ext4_setup_system_zone(struct super_block *sb); 2724 extern int __init ext4_init_system_zone(void); 2725 extern void ext4_exit_system_zone(void); 2726 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2727 ext4_fsblk_t start_blk, 2728 unsigned int count); 2729 extern int ext4_check_blockref(const char *, unsigned int, 2730 struct inode *, __le32 *, unsigned int); 2731 2732 /* extents.c */ 2733 struct ext4_ext_path; 2734 struct ext4_extent; 2735 2736 /* 2737 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 2738 * __le32. 2739 */ 2740 #define EXT_MAX_BLOCKS 0xffffffff 2741 2742 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2743 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2744 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 2745 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2746 struct ext4_map_blocks *map, int flags); 2747 extern void ext4_ext_truncate(handle_t *, struct inode *); 2748 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 2749 ext4_lblk_t end); 2750 extern void ext4_ext_init(struct super_block *); 2751 extern void ext4_ext_release(struct super_block *); 2752 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2753 loff_t len); 2754 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 2755 loff_t offset, ssize_t len); 2756 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2757 struct ext4_map_blocks *map, int flags); 2758 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2759 ext4_lblk_t lblocks); 2760 extern int ext4_extent_tree_init(handle_t *, struct inode *); 2761 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2762 int num, 2763 struct ext4_ext_path *path); 2764 extern int ext4_can_extents_be_merged(struct inode *inode, 2765 struct ext4_extent *ex1, 2766 struct ext4_extent *ex2); 2767 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2768 struct ext4_ext_path **, 2769 struct ext4_extent *, int); 2770 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 2771 struct ext4_ext_path **, 2772 int flags); 2773 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2774 extern int ext4_ext_check_inode(struct inode *inode); 2775 extern int ext4_find_delalloc_range(struct inode *inode, 2776 ext4_lblk_t lblk_start, 2777 ext4_lblk_t lblk_end); 2778 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2779 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 2780 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2781 __u64 start, __u64 len); 2782 extern int ext4_ext_precache(struct inode *inode); 2783 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 2784 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 2785 struct inode *inode2, ext4_lblk_t lblk1, 2786 ext4_lblk_t lblk2, ext4_lblk_t count, 2787 int mark_unwritten,int *err); 2788 2789 /* move_extent.c */ 2790 extern void ext4_double_down_write_data_sem(struct inode *first, 2791 struct inode *second); 2792 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 2793 struct inode *donor_inode); 2794 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2795 __u64 start_orig, __u64 start_donor, 2796 __u64 len, __u64 *moved_len); 2797 2798 /* page-io.c */ 2799 extern int __init ext4_init_pageio(void); 2800 extern void ext4_exit_pageio(void); 2801 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2802 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 2803 extern int ext4_put_io_end(ext4_io_end_t *io_end); 2804 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 2805 extern void ext4_io_submit_init(struct ext4_io_submit *io, 2806 struct writeback_control *wbc); 2807 extern void ext4_end_io_rsv_work(struct work_struct *work); 2808 extern void ext4_io_submit(struct ext4_io_submit *io); 2809 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2810 struct page *page, 2811 int len, 2812 struct writeback_control *wbc, 2813 bool keep_towrite); 2814 2815 /* mmp.c */ 2816 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2817 2818 /* 2819 * Add new method to test whether block and inode bitmaps are properly 2820 * initialized. With uninit_bg reading the block from disk is not enough 2821 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2822 */ 2823 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2824 2825 static inline int bitmap_uptodate(struct buffer_head *bh) 2826 { 2827 return (buffer_uptodate(bh) && 2828 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2829 } 2830 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2831 { 2832 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2833 } 2834 2835 /* 2836 * Disable DIO read nolock optimization, so new dioreaders will be forced 2837 * to grab i_mutex 2838 */ 2839 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2840 { 2841 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2842 smp_mb(); 2843 } 2844 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2845 { 2846 smp_mb(); 2847 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2848 } 2849 2850 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2851 2852 /* For ioend & aio unwritten conversion wait queues */ 2853 #define EXT4_WQ_HASH_SZ 37 2854 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2855 EXT4_WQ_HASH_SZ]) 2856 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2857 EXT4_WQ_HASH_SZ]) 2858 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2859 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2860 2861 #define EXT4_RESIZING 0 2862 extern int ext4_resize_begin(struct super_block *sb); 2863 extern void ext4_resize_end(struct super_block *sb); 2864 2865 #endif /* __KERNEL__ */ 2866 2867 #endif /* _EXT4_H */ 2868