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 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 368 /* End compression flags --- maybe not all used */ 369 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 370 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 371 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 372 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 373 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 374 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 375 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 376 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 377 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 378 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 379 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 380 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 381 382 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 383 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 384 385 /* Flags that should be inherited by new inodes from their parent. */ 386 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 387 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 388 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 389 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 390 391 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 392 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 393 394 /* Flags that are appropriate for non-directories/regular files. */ 395 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 396 397 /* Mask out flags that are inappropriate for the given type of inode. */ 398 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 399 { 400 if (S_ISDIR(mode)) 401 return flags; 402 else if (S_ISREG(mode)) 403 return flags & EXT4_REG_FLMASK; 404 else 405 return flags & EXT4_OTHER_FLMASK; 406 } 407 408 /* 409 * Inode flags used for atomic set/get 410 */ 411 enum { 412 EXT4_INODE_SECRM = 0, /* Secure deletion */ 413 EXT4_INODE_UNRM = 1, /* Undelete */ 414 EXT4_INODE_COMPR = 2, /* Compress file */ 415 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 416 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 417 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 418 EXT4_INODE_NODUMP = 6, /* do not dump file */ 419 EXT4_INODE_NOATIME = 7, /* do not update atime */ 420 /* Reserved for compression usage... */ 421 EXT4_INODE_DIRTY = 8, 422 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 423 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 424 EXT4_INODE_ECOMPR = 11, /* Compression error */ 425 /* End compression flags --- maybe not all used */ 426 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 427 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 428 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 429 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 430 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 431 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 432 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 433 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 434 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 435 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 436 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 437 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 438 }; 439 440 /* 441 * Since it's pretty easy to mix up bit numbers and hex values, we use a 442 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 443 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 444 * any extra space in the compiled kernel image, otherwise, the build will fail. 445 * It's important that these values are the same, since we are using 446 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 447 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 448 * values found in ext2, ext3 and ext4 filesystems, and of course the values 449 * defined in e2fsprogs. 450 * 451 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 452 */ 453 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 454 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 455 456 static inline void ext4_check_flag_values(void) 457 { 458 CHECK_FLAG_VALUE(SECRM); 459 CHECK_FLAG_VALUE(UNRM); 460 CHECK_FLAG_VALUE(COMPR); 461 CHECK_FLAG_VALUE(SYNC); 462 CHECK_FLAG_VALUE(IMMUTABLE); 463 CHECK_FLAG_VALUE(APPEND); 464 CHECK_FLAG_VALUE(NODUMP); 465 CHECK_FLAG_VALUE(NOATIME); 466 CHECK_FLAG_VALUE(DIRTY); 467 CHECK_FLAG_VALUE(COMPRBLK); 468 CHECK_FLAG_VALUE(NOCOMPR); 469 CHECK_FLAG_VALUE(ECOMPR); 470 CHECK_FLAG_VALUE(INDEX); 471 CHECK_FLAG_VALUE(IMAGIC); 472 CHECK_FLAG_VALUE(JOURNAL_DATA); 473 CHECK_FLAG_VALUE(NOTAIL); 474 CHECK_FLAG_VALUE(DIRSYNC); 475 CHECK_FLAG_VALUE(TOPDIR); 476 CHECK_FLAG_VALUE(HUGE_FILE); 477 CHECK_FLAG_VALUE(EXTENTS); 478 CHECK_FLAG_VALUE(EA_INODE); 479 CHECK_FLAG_VALUE(EOFBLOCKS); 480 CHECK_FLAG_VALUE(INLINE_DATA); 481 CHECK_FLAG_VALUE(RESERVED); 482 } 483 484 /* Used to pass group descriptor data when online resize is done */ 485 struct ext4_new_group_input { 486 __u32 group; /* Group number for this data */ 487 __u64 block_bitmap; /* Absolute block number of block bitmap */ 488 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 489 __u64 inode_table; /* Absolute block number of inode table start */ 490 __u32 blocks_count; /* Total number of blocks in this group */ 491 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 492 __u16 unused; 493 }; 494 495 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 496 struct compat_ext4_new_group_input { 497 u32 group; 498 compat_u64 block_bitmap; 499 compat_u64 inode_bitmap; 500 compat_u64 inode_table; 501 u32 blocks_count; 502 u16 reserved_blocks; 503 u16 unused; 504 }; 505 #endif 506 507 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 508 struct ext4_new_group_data { 509 __u32 group; 510 __u64 block_bitmap; 511 __u64 inode_bitmap; 512 __u64 inode_table; 513 __u32 blocks_count; 514 __u16 reserved_blocks; 515 __u16 unused; 516 __u32 free_blocks_count; 517 }; 518 519 /* Indexes used to index group tables in ext4_new_group_data */ 520 enum { 521 BLOCK_BITMAP = 0, /* block bitmap */ 522 INODE_BITMAP, /* inode bitmap */ 523 INODE_TABLE, /* inode tables */ 524 GROUP_TABLE_COUNT, 525 }; 526 527 /* 528 * Flags used by ext4_map_blocks() 529 */ 530 /* Allocate any needed blocks and/or convert an unwritten 531 extent to be an initialized ext4 */ 532 #define EXT4_GET_BLOCKS_CREATE 0x0001 533 /* Request the creation of an unwritten extent */ 534 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 535 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 536 EXT4_GET_BLOCKS_CREATE) 537 /* Caller is from the delayed allocation writeout path 538 * finally doing the actual allocation of delayed blocks */ 539 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 540 /* caller is from the direct IO path, request to creation of an 541 unwritten extents if not allocated, split the unwritten 542 extent if blocks has been preallocated already*/ 543 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 544 #define EXT4_GET_BLOCKS_CONVERT 0x0010 545 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 546 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 547 /* Convert extent to initialized after IO complete */ 548 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 549 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 550 /* Eventual metadata allocation (due to growing extent tree) 551 * should not fail, so try to use reserved blocks for that.*/ 552 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 553 /* Don't normalize allocation size (used for fallocate) */ 554 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 555 /* Request will not result in inode size update (user for fallocate) */ 556 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 557 /* Do not take i_data_sem locking in ext4_map_blocks */ 558 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 559 /* Convert written extents to unwritten */ 560 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200 561 562 /* 563 * The bit position of these flags must not overlap with any of the 564 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 565 * read_extent_tree_block(), ext4_split_extent_at(), 566 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 567 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 568 * caching the extents when reading from the extent tree while a 569 * truncate or punch hole operation is in progress. 570 */ 571 #define EXT4_EX_NOCACHE 0x40000000 572 #define EXT4_EX_FORCE_CACHE 0x20000000 573 574 /* 575 * Flags used by ext4_free_blocks 576 */ 577 #define EXT4_FREE_BLOCKS_METADATA 0x0001 578 #define EXT4_FREE_BLOCKS_FORGET 0x0002 579 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 580 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 581 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 582 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 583 584 /* 585 * ioctl commands 586 */ 587 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 588 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 589 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 590 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 591 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 592 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 593 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 594 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 595 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 596 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 597 #define EXT4_IOC_MIGRATE _IO('f', 9) 598 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 599 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 600 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 601 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 602 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 603 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 604 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 605 606 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 607 /* 608 * ioctl commands in 32 bit emulation 609 */ 610 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 611 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 612 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 613 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 614 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 615 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 616 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 617 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 618 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 619 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 620 #endif 621 622 /* Max physical block we can address w/o extents */ 623 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 624 625 /* 626 * Structure of an inode on the disk 627 */ 628 struct ext4_inode { 629 __le16 i_mode; /* File mode */ 630 __le16 i_uid; /* Low 16 bits of Owner Uid */ 631 __le32 i_size_lo; /* Size in bytes */ 632 __le32 i_atime; /* Access time */ 633 __le32 i_ctime; /* Inode Change time */ 634 __le32 i_mtime; /* Modification time */ 635 __le32 i_dtime; /* Deletion Time */ 636 __le16 i_gid; /* Low 16 bits of Group Id */ 637 __le16 i_links_count; /* Links count */ 638 __le32 i_blocks_lo; /* Blocks count */ 639 __le32 i_flags; /* File flags */ 640 union { 641 struct { 642 __le32 l_i_version; 643 } linux1; 644 struct { 645 __u32 h_i_translator; 646 } hurd1; 647 struct { 648 __u32 m_i_reserved1; 649 } masix1; 650 } osd1; /* OS dependent 1 */ 651 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 652 __le32 i_generation; /* File version (for NFS) */ 653 __le32 i_file_acl_lo; /* File ACL */ 654 __le32 i_size_high; 655 __le32 i_obso_faddr; /* Obsoleted fragment address */ 656 union { 657 struct { 658 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 659 __le16 l_i_file_acl_high; 660 __le16 l_i_uid_high; /* these 2 fields */ 661 __le16 l_i_gid_high; /* were reserved2[0] */ 662 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 663 __le16 l_i_reserved; 664 } linux2; 665 struct { 666 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 667 __u16 h_i_mode_high; 668 __u16 h_i_uid_high; 669 __u16 h_i_gid_high; 670 __u32 h_i_author; 671 } hurd2; 672 struct { 673 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 674 __le16 m_i_file_acl_high; 675 __u32 m_i_reserved2[2]; 676 } masix2; 677 } osd2; /* OS dependent 2 */ 678 __le16 i_extra_isize; 679 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 680 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 681 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 682 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 683 __le32 i_crtime; /* File Creation time */ 684 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 685 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 686 }; 687 688 struct move_extent { 689 __u32 reserved; /* should be zero */ 690 __u32 donor_fd; /* donor file descriptor */ 691 __u64 orig_start; /* logical start offset in block for orig */ 692 __u64 donor_start; /* logical start offset in block for donor */ 693 __u64 len; /* block length to be moved */ 694 __u64 moved_len; /* moved block length */ 695 }; 696 697 #define EXT4_EPOCH_BITS 2 698 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 699 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 700 701 /* 702 * Extended fields will fit into an inode if the filesystem was formatted 703 * with large inodes (-I 256 or larger) and there are not currently any EAs 704 * consuming all of the available space. For new inodes we always reserve 705 * enough space for the kernel's known extended fields, but for inodes 706 * created with an old kernel this might not have been the case. None of 707 * the extended inode fields is critical for correct filesystem operation. 708 * This macro checks if a certain field fits in the inode. Note that 709 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 710 */ 711 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 712 ((offsetof(typeof(*ext4_inode), field) + \ 713 sizeof((ext4_inode)->field)) \ 714 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 715 (einode)->i_extra_isize)) \ 716 717 static inline __le32 ext4_encode_extra_time(struct timespec *time) 718 { 719 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 720 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 721 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 722 } 723 724 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 725 { 726 if (sizeof(time->tv_sec) > 4) 727 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 728 << 32; 729 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 730 } 731 732 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 733 do { \ 734 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 735 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 736 (raw_inode)->xtime ## _extra = \ 737 ext4_encode_extra_time(&(inode)->xtime); \ 738 } while (0) 739 740 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 741 do { \ 742 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 743 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 744 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 745 (raw_inode)->xtime ## _extra = \ 746 ext4_encode_extra_time(&(einode)->xtime); \ 747 } while (0) 748 749 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 750 do { \ 751 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 752 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 753 ext4_decode_extra_time(&(inode)->xtime, \ 754 raw_inode->xtime ## _extra); \ 755 else \ 756 (inode)->xtime.tv_nsec = 0; \ 757 } while (0) 758 759 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 760 do { \ 761 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 762 (einode)->xtime.tv_sec = \ 763 (signed)le32_to_cpu((raw_inode)->xtime); \ 764 else \ 765 (einode)->xtime.tv_sec = 0; \ 766 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 767 ext4_decode_extra_time(&(einode)->xtime, \ 768 raw_inode->xtime ## _extra); \ 769 else \ 770 (einode)->xtime.tv_nsec = 0; \ 771 } while (0) 772 773 #define i_disk_version osd1.linux1.l_i_version 774 775 #if defined(__KERNEL__) || defined(__linux__) 776 #define i_reserved1 osd1.linux1.l_i_reserved1 777 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 778 #define i_blocks_high osd2.linux2.l_i_blocks_high 779 #define i_uid_low i_uid 780 #define i_gid_low i_gid 781 #define i_uid_high osd2.linux2.l_i_uid_high 782 #define i_gid_high osd2.linux2.l_i_gid_high 783 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 784 785 #elif defined(__GNU__) 786 787 #define i_translator osd1.hurd1.h_i_translator 788 #define i_uid_high osd2.hurd2.h_i_uid_high 789 #define i_gid_high osd2.hurd2.h_i_gid_high 790 #define i_author osd2.hurd2.h_i_author 791 792 #elif defined(__masix__) 793 794 #define i_reserved1 osd1.masix1.m_i_reserved1 795 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 796 #define i_reserved2 osd2.masix2.m_i_reserved2 797 798 #endif /* defined(__KERNEL__) || defined(__linux__) */ 799 800 #include "extents_status.h" 801 802 /* 803 * fourth extended file system inode data in memory 804 */ 805 struct ext4_inode_info { 806 __le32 i_data[15]; /* unconverted */ 807 __u32 i_dtime; 808 ext4_fsblk_t i_file_acl; 809 810 /* 811 * i_block_group is the number of the block group which contains 812 * this file's inode. Constant across the lifetime of the inode, 813 * it is ued for making block allocation decisions - we try to 814 * place a file's data blocks near its inode block, and new inodes 815 * near to their parent directory's inode. 816 */ 817 ext4_group_t i_block_group; 818 ext4_lblk_t i_dir_start_lookup; 819 #if (BITS_PER_LONG < 64) 820 unsigned long i_state_flags; /* Dynamic state flags */ 821 #endif 822 unsigned long i_flags; 823 824 /* 825 * Extended attributes can be read independently of the main file 826 * data. Taking i_mutex even when reading would cause contention 827 * between readers of EAs and writers of regular file data, so 828 * instead we synchronize on xattr_sem when reading or changing 829 * EAs. 830 */ 831 struct rw_semaphore xattr_sem; 832 833 struct list_head i_orphan; /* unlinked but open inodes */ 834 835 /* 836 * i_disksize keeps track of what the inode size is ON DISK, not 837 * in memory. During truncate, i_size is set to the new size by 838 * the VFS prior to calling ext4_truncate(), but the filesystem won't 839 * set i_disksize to 0 until the truncate is actually under way. 840 * 841 * The intent is that i_disksize always represents the blocks which 842 * are used by this file. This allows recovery to restart truncate 843 * on orphans if we crash during truncate. We actually write i_disksize 844 * into the on-disk inode when writing inodes out, instead of i_size. 845 * 846 * The only time when i_disksize and i_size may be different is when 847 * a truncate is in progress. The only things which change i_disksize 848 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 849 */ 850 loff_t i_disksize; 851 852 /* 853 * i_data_sem is for serialising ext4_truncate() against 854 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 855 * data tree are chopped off during truncate. We can't do that in 856 * ext4 because whenever we perform intermediate commits during 857 * truncate, the inode and all the metadata blocks *must* be in a 858 * consistent state which allows truncation of the orphans to restart 859 * during recovery. Hence we must fix the get_block-vs-truncate race 860 * by other means, so we have i_data_sem. 861 */ 862 struct rw_semaphore i_data_sem; 863 struct inode vfs_inode; 864 struct jbd2_inode *jinode; 865 866 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 867 868 /* 869 * File creation time. Its function is same as that of 870 * struct timespec i_{a,c,m}time in the generic inode. 871 */ 872 struct timespec i_crtime; 873 874 /* mballoc */ 875 struct list_head i_prealloc_list; 876 spinlock_t i_prealloc_lock; 877 878 /* extents status tree */ 879 struct ext4_es_tree i_es_tree; 880 rwlock_t i_es_lock; 881 struct list_head i_es_list; 882 unsigned int i_es_all_nr; /* protected by i_es_lock */ 883 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 884 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 885 extents to shrink. Protected by 886 i_es_lock */ 887 888 /* ialloc */ 889 ext4_group_t i_last_alloc_group; 890 891 /* allocation reservation info for delalloc */ 892 /* In case of bigalloc, these refer to clusters rather than blocks */ 893 unsigned int i_reserved_data_blocks; 894 unsigned int i_reserved_meta_blocks; 895 unsigned int i_allocated_meta_blocks; 896 ext4_lblk_t i_da_metadata_calc_last_lblock; 897 int i_da_metadata_calc_len; 898 899 /* on-disk additional length */ 900 __u16 i_extra_isize; 901 902 /* Indicate the inline data space. */ 903 u16 i_inline_off; 904 u16 i_inline_size; 905 906 #ifdef CONFIG_QUOTA 907 /* quota space reservation, managed internally by quota code */ 908 qsize_t i_reserved_quota; 909 #endif 910 911 /* Lock protecting lists below */ 912 spinlock_t i_completed_io_lock; 913 /* 914 * Completed IOs that need unwritten extents handling and have 915 * transaction reserved 916 */ 917 struct list_head i_rsv_conversion_list; 918 /* 919 * Completed IOs that need unwritten extents handling and don't have 920 * transaction reserved 921 */ 922 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 923 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 924 struct work_struct i_rsv_conversion_work; 925 926 spinlock_t i_block_reservation_lock; 927 928 /* 929 * Transactions that contain inode's metadata needed to complete 930 * fsync and fdatasync, respectively. 931 */ 932 tid_t i_sync_tid; 933 tid_t i_datasync_tid; 934 935 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 936 __u32 i_csum_seed; 937 }; 938 939 /* 940 * File system states 941 */ 942 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 943 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 944 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 945 946 /* 947 * Misc. filesystem flags 948 */ 949 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 950 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 951 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 952 953 /* 954 * Mount flags set via mount options or defaults 955 */ 956 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 957 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 958 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 959 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 960 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 961 #define EXT4_MOUNT_ERRORS_MASK 0x00070 962 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 963 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 964 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 965 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 966 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 967 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 968 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 969 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 970 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 971 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 972 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 973 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 974 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 975 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 976 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 977 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 978 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 979 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 980 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 981 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 982 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 983 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 984 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 985 986 /* 987 * Mount flags set either automatically (could not be set by mount option) 988 * based on per file system feature or property or in special cases such as 989 * distinguishing between explicit mount option definition and default. 990 */ 991 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 992 specified delalloc */ 993 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 994 size of blocksize * 8 995 blocks */ 996 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 997 file systems */ 998 999 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1000 ~EXT4_MOUNT_##opt 1001 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1002 EXT4_MOUNT_##opt 1003 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1004 EXT4_MOUNT_##opt) 1005 1006 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1007 ~EXT4_MOUNT2_##opt 1008 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1009 EXT4_MOUNT2_##opt 1010 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1011 EXT4_MOUNT2_##opt) 1012 1013 #define ext4_test_and_set_bit __test_and_set_bit_le 1014 #define ext4_set_bit __set_bit_le 1015 #define ext4_set_bit_atomic ext2_set_bit_atomic 1016 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1017 #define ext4_clear_bit __clear_bit_le 1018 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1019 #define ext4_test_bit test_bit_le 1020 #define ext4_find_next_zero_bit find_next_zero_bit_le 1021 #define ext4_find_next_bit find_next_bit_le 1022 1023 extern void ext4_set_bits(void *bm, int cur, int len); 1024 1025 /* 1026 * Maximal mount counts between two filesystem checks 1027 */ 1028 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1029 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1030 1031 /* 1032 * Behaviour when detecting errors 1033 */ 1034 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1035 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1036 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1037 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1038 1039 /* Metadata checksum algorithm codes */ 1040 #define EXT4_CRC32C_CHKSUM 1 1041 1042 /* 1043 * Structure of the super block 1044 */ 1045 struct ext4_super_block { 1046 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1047 __le32 s_blocks_count_lo; /* Blocks count */ 1048 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1049 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1050 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1051 __le32 s_first_data_block; /* First Data Block */ 1052 __le32 s_log_block_size; /* Block size */ 1053 __le32 s_log_cluster_size; /* Allocation cluster size */ 1054 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1055 __le32 s_clusters_per_group; /* # Clusters per group */ 1056 __le32 s_inodes_per_group; /* # Inodes per group */ 1057 __le32 s_mtime; /* Mount time */ 1058 /*30*/ __le32 s_wtime; /* Write time */ 1059 __le16 s_mnt_count; /* Mount count */ 1060 __le16 s_max_mnt_count; /* Maximal mount count */ 1061 __le16 s_magic; /* Magic signature */ 1062 __le16 s_state; /* File system state */ 1063 __le16 s_errors; /* Behaviour when detecting errors */ 1064 __le16 s_minor_rev_level; /* minor revision level */ 1065 /*40*/ __le32 s_lastcheck; /* time of last check */ 1066 __le32 s_checkinterval; /* max. time between checks */ 1067 __le32 s_creator_os; /* OS */ 1068 __le32 s_rev_level; /* Revision level */ 1069 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1070 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1071 /* 1072 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1073 * 1074 * Note: the difference between the compatible feature set and 1075 * the incompatible feature set is that if there is a bit set 1076 * in the incompatible feature set that the kernel doesn't 1077 * know about, it should refuse to mount the filesystem. 1078 * 1079 * e2fsck's requirements are more strict; if it doesn't know 1080 * about a feature in either the compatible or incompatible 1081 * feature set, it must abort and not try to meddle with 1082 * things it doesn't understand... 1083 */ 1084 __le32 s_first_ino; /* First non-reserved inode */ 1085 __le16 s_inode_size; /* size of inode structure */ 1086 __le16 s_block_group_nr; /* block group # of this superblock */ 1087 __le32 s_feature_compat; /* compatible feature set */ 1088 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1089 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1090 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1091 /*78*/ char s_volume_name[16]; /* volume name */ 1092 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1093 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1094 /* 1095 * Performance hints. Directory preallocation should only 1096 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1097 */ 1098 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1099 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1100 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1101 /* 1102 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1103 */ 1104 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1105 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1106 __le32 s_journal_dev; /* device number of journal file */ 1107 __le32 s_last_orphan; /* start of list of inodes to delete */ 1108 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1109 __u8 s_def_hash_version; /* Default hash version to use */ 1110 __u8 s_jnl_backup_type; 1111 __le16 s_desc_size; /* size of group descriptor */ 1112 /*100*/ __le32 s_default_mount_opts; 1113 __le32 s_first_meta_bg; /* First metablock block group */ 1114 __le32 s_mkfs_time; /* When the filesystem was created */ 1115 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1116 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1117 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1118 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1119 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1120 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1121 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1122 __le32 s_flags; /* Miscellaneous flags */ 1123 __le16 s_raid_stride; /* RAID stride */ 1124 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1125 __le64 s_mmp_block; /* Block for multi-mount protection */ 1126 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1127 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1128 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1129 __le16 s_reserved_pad; 1130 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1131 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1132 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1133 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1134 snapshot's future use */ 1135 __le32 s_snapshot_list; /* inode number of the head of the 1136 on-disk snapshot list */ 1137 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1138 __le32 s_error_count; /* number of fs errors */ 1139 __le32 s_first_error_time; /* first time an error happened */ 1140 __le32 s_first_error_ino; /* inode involved in first error */ 1141 __le64 s_first_error_block; /* block involved of first error */ 1142 __u8 s_first_error_func[32]; /* function where the error happened */ 1143 __le32 s_first_error_line; /* line number where error happened */ 1144 __le32 s_last_error_time; /* most recent time of an error */ 1145 __le32 s_last_error_ino; /* inode involved in last error */ 1146 __le32 s_last_error_line; /* line number where error happened */ 1147 __le64 s_last_error_block; /* block involved of last error */ 1148 __u8 s_last_error_func[32]; /* function where the error happened */ 1149 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1150 __u8 s_mount_opts[64]; 1151 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1152 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1153 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1154 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1155 __le32 s_reserved[106]; /* Padding to the end of the block */ 1156 __le32 s_checksum; /* crc32c(superblock) */ 1157 }; 1158 1159 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1160 1161 #ifdef __KERNEL__ 1162 1163 /* 1164 * run-time mount flags 1165 */ 1166 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1167 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1168 1169 /* Number of quota types we support */ 1170 #define EXT4_MAXQUOTAS 2 1171 1172 /* 1173 * fourth extended-fs super-block data in memory 1174 */ 1175 struct ext4_sb_info { 1176 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1177 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1178 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1179 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1180 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1181 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1182 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1183 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1184 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1185 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1186 unsigned long s_overhead; /* # of fs overhead clusters */ 1187 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1188 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1189 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1190 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1191 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1192 struct buffer_head **s_group_desc; 1193 unsigned int s_mount_opt; 1194 unsigned int s_mount_opt2; 1195 unsigned int s_mount_flags; 1196 unsigned int s_def_mount_opt; 1197 ext4_fsblk_t s_sb_block; 1198 atomic64_t s_resv_clusters; 1199 kuid_t s_resuid; 1200 kgid_t s_resgid; 1201 unsigned short s_mount_state; 1202 unsigned short s_pad; 1203 int s_addr_per_block_bits; 1204 int s_desc_per_block_bits; 1205 int s_inode_size; 1206 int s_first_ino; 1207 unsigned int s_inode_readahead_blks; 1208 unsigned int s_inode_goal; 1209 spinlock_t s_next_gen_lock; 1210 u32 s_next_generation; 1211 u32 s_hash_seed[4]; 1212 int s_def_hash_version; 1213 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1214 struct percpu_counter s_freeclusters_counter; 1215 struct percpu_counter s_freeinodes_counter; 1216 struct percpu_counter s_dirs_counter; 1217 struct percpu_counter s_dirtyclusters_counter; 1218 struct blockgroup_lock *s_blockgroup_lock; 1219 struct proc_dir_entry *s_proc; 1220 struct kobject s_kobj; 1221 struct completion s_kobj_unregister; 1222 struct super_block *s_sb; 1223 1224 /* Journaling */ 1225 struct journal_s *s_journal; 1226 struct list_head s_orphan; 1227 struct mutex s_orphan_lock; 1228 unsigned long s_resize_flags; /* Flags indicating if there 1229 is a resizer */ 1230 unsigned long s_commit_interval; 1231 u32 s_max_batch_time; 1232 u32 s_min_batch_time; 1233 struct block_device *journal_bdev; 1234 #ifdef CONFIG_QUOTA 1235 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1236 int s_jquota_fmt; /* Format of quota to use */ 1237 #endif 1238 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1239 struct rb_root system_blks; 1240 1241 #ifdef EXTENTS_STATS 1242 /* ext4 extents stats */ 1243 unsigned long s_ext_min; 1244 unsigned long s_ext_max; 1245 unsigned long s_depth_max; 1246 spinlock_t s_ext_stats_lock; 1247 unsigned long s_ext_blocks; 1248 unsigned long s_ext_extents; 1249 #endif 1250 1251 /* for buddy allocator */ 1252 struct ext4_group_info ***s_group_info; 1253 struct inode *s_buddy_cache; 1254 spinlock_t s_md_lock; 1255 unsigned short *s_mb_offsets; 1256 unsigned int *s_mb_maxs; 1257 unsigned int s_group_info_size; 1258 1259 /* tunables */ 1260 unsigned long s_stripe; 1261 unsigned int s_mb_stream_request; 1262 unsigned int s_mb_max_to_scan; 1263 unsigned int s_mb_min_to_scan; 1264 unsigned int s_mb_stats; 1265 unsigned int s_mb_order2_reqs; 1266 unsigned int s_mb_group_prealloc; 1267 unsigned int s_max_dir_size_kb; 1268 /* where last allocation was done - for stream allocation */ 1269 unsigned long s_mb_last_group; 1270 unsigned long s_mb_last_start; 1271 1272 /* stats for buddy allocator */ 1273 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1274 atomic_t s_bal_success; /* we found long enough chunks */ 1275 atomic_t s_bal_allocated; /* in blocks */ 1276 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1277 atomic_t s_bal_goals; /* goal hits */ 1278 atomic_t s_bal_breaks; /* too long searches */ 1279 atomic_t s_bal_2orders; /* 2^order hits */ 1280 spinlock_t s_bal_lock; 1281 unsigned long s_mb_buddies_generated; 1282 unsigned long long s_mb_generation_time; 1283 atomic_t s_mb_lost_chunks; 1284 atomic_t s_mb_preallocated; 1285 atomic_t s_mb_discarded; 1286 atomic_t s_lock_busy; 1287 1288 /* locality groups */ 1289 struct ext4_locality_group __percpu *s_locality_groups; 1290 1291 /* for write statistics */ 1292 unsigned long s_sectors_written_start; 1293 u64 s_kbytes_written; 1294 1295 /* the size of zero-out chunk */ 1296 unsigned int s_extent_max_zeroout_kb; 1297 1298 unsigned int s_log_groups_per_flex; 1299 struct flex_groups *s_flex_groups; 1300 ext4_group_t s_flex_groups_allocated; 1301 1302 /* workqueue for reserved extent conversions (buffered io) */ 1303 struct workqueue_struct *rsv_conversion_wq; 1304 1305 /* timer for periodic error stats printing */ 1306 struct timer_list s_err_report; 1307 1308 /* Lazy inode table initialization info */ 1309 struct ext4_li_request *s_li_request; 1310 /* Wait multiplier for lazy initialization thread */ 1311 unsigned int s_li_wait_mult; 1312 1313 /* Kernel thread for multiple mount protection */ 1314 struct task_struct *s_mmp_tsk; 1315 1316 /* record the last minlen when FITRIM is called. */ 1317 atomic_t s_last_trim_minblks; 1318 1319 /* Reference to checksum algorithm driver via cryptoapi */ 1320 struct crypto_shash *s_chksum_driver; 1321 1322 /* Precomputed FS UUID checksum for seeding other checksums */ 1323 __u32 s_csum_seed; 1324 1325 /* Reclaim extents from extent status tree */ 1326 struct shrinker s_es_shrinker; 1327 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1328 long s_es_nr_inode; 1329 struct ext4_es_stats s_es_stats; 1330 struct mb_cache *s_mb_cache; 1331 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1332 1333 /* Ratelimit ext4 messages. */ 1334 struct ratelimit_state s_err_ratelimit_state; 1335 struct ratelimit_state s_warning_ratelimit_state; 1336 struct ratelimit_state s_msg_ratelimit_state; 1337 }; 1338 1339 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1340 { 1341 return sb->s_fs_info; 1342 } 1343 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1344 { 1345 return container_of(inode, struct ext4_inode_info, vfs_inode); 1346 } 1347 1348 static inline struct timespec ext4_current_time(struct inode *inode) 1349 { 1350 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1351 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1352 } 1353 1354 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1355 { 1356 return ino == EXT4_ROOT_INO || 1357 ino == EXT4_USR_QUOTA_INO || 1358 ino == EXT4_GRP_QUOTA_INO || 1359 ino == EXT4_BOOT_LOADER_INO || 1360 ino == EXT4_JOURNAL_INO || 1361 ino == EXT4_RESIZE_INO || 1362 (ino >= EXT4_FIRST_INO(sb) && 1363 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1364 } 1365 1366 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1367 struct ext4_io_end *io_end) 1368 { 1369 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1370 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1371 atomic_inc(&EXT4_I(inode)->i_unwritten); 1372 } 1373 } 1374 1375 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1376 { 1377 return inode->i_private; 1378 } 1379 1380 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1381 { 1382 inode->i_private = io; 1383 } 1384 1385 /* 1386 * Inode dynamic state flags 1387 */ 1388 enum { 1389 EXT4_STATE_JDATA, /* journaled data exists */ 1390 EXT4_STATE_NEW, /* inode is newly created */ 1391 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1392 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1393 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1394 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1395 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1396 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1397 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1398 nolocking */ 1399 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1400 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1401 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1402 }; 1403 1404 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1405 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1406 { \ 1407 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1408 } \ 1409 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1410 { \ 1411 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1412 } \ 1413 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1414 { \ 1415 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1416 } 1417 1418 /* Add these declarations here only so that these functions can be 1419 * found by name. Otherwise, they are very hard to locate. */ 1420 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1421 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1422 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1423 EXT4_INODE_BIT_FNS(flag, flags, 0) 1424 1425 /* Add these declarations here only so that these functions can be 1426 * found by name. Otherwise, they are very hard to locate. */ 1427 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1428 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1429 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1430 #if (BITS_PER_LONG < 64) 1431 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1432 1433 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1434 { 1435 (ei)->i_state_flags = 0; 1436 } 1437 #else 1438 EXT4_INODE_BIT_FNS(state, flags, 32) 1439 1440 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1441 { 1442 /* We depend on the fact that callers will set i_flags */ 1443 } 1444 #endif 1445 #else 1446 /* Assume that user mode programs are passing in an ext4fs superblock, not 1447 * a kernel struct super_block. This will allow us to call the feature-test 1448 * macros from user land. */ 1449 #define EXT4_SB(sb) (sb) 1450 #endif 1451 1452 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1453 1454 /* 1455 * Codes for operating systems 1456 */ 1457 #define EXT4_OS_LINUX 0 1458 #define EXT4_OS_HURD 1 1459 #define EXT4_OS_MASIX 2 1460 #define EXT4_OS_FREEBSD 3 1461 #define EXT4_OS_LITES 4 1462 1463 /* 1464 * Revision levels 1465 */ 1466 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1467 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1468 1469 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1470 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1471 1472 #define EXT4_GOOD_OLD_INODE_SIZE 128 1473 1474 /* 1475 * Feature set definitions 1476 */ 1477 1478 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1479 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1480 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1481 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1482 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1483 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1484 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1485 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1486 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1487 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1488 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1489 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1490 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1491 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1492 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1493 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1494 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1495 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1496 1497 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1498 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1499 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1500 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1501 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1502 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1503 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1504 1505 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1506 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1507 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1508 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1509 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1510 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1511 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1512 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1513 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1514 /* 1515 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1516 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1517 * all other data structures' checksums. However, the METADATA_CSUM and 1518 * GDT_CSUM bits are mutually exclusive. 1519 */ 1520 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1521 1522 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1523 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1524 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1525 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1526 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1527 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1528 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1529 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1530 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1531 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1532 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1533 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1534 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1535 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1536 1537 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1538 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1539 EXT4_FEATURE_INCOMPAT_META_BG) 1540 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1541 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1542 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1543 1544 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1545 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1546 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1547 EXT4_FEATURE_INCOMPAT_META_BG) 1548 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1549 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1550 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1551 1552 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1553 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1554 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1555 EXT4_FEATURE_INCOMPAT_META_BG| \ 1556 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1557 EXT4_FEATURE_INCOMPAT_64BIT| \ 1558 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1559 EXT4_FEATURE_INCOMPAT_MMP | \ 1560 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1561 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1562 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1563 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1564 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1565 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1566 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1567 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1568 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1569 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1570 EXT4_FEATURE_RO_COMPAT_QUOTA) 1571 1572 /* 1573 * Default values for user and/or group using reserved blocks 1574 */ 1575 #define EXT4_DEF_RESUID 0 1576 #define EXT4_DEF_RESGID 0 1577 1578 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1579 1580 /* 1581 * Default mount options 1582 */ 1583 #define EXT4_DEFM_DEBUG 0x0001 1584 #define EXT4_DEFM_BSDGROUPS 0x0002 1585 #define EXT4_DEFM_XATTR_USER 0x0004 1586 #define EXT4_DEFM_ACL 0x0008 1587 #define EXT4_DEFM_UID16 0x0010 1588 #define EXT4_DEFM_JMODE 0x0060 1589 #define EXT4_DEFM_JMODE_DATA 0x0020 1590 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1591 #define EXT4_DEFM_JMODE_WBACK 0x0060 1592 #define EXT4_DEFM_NOBARRIER 0x0100 1593 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1594 #define EXT4_DEFM_DISCARD 0x0400 1595 #define EXT4_DEFM_NODELALLOC 0x0800 1596 1597 /* 1598 * Default journal batch times 1599 */ 1600 #define EXT4_DEF_MIN_BATCH_TIME 0 1601 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1602 1603 /* 1604 * Minimum number of groups in a flexgroup before we separate out 1605 * directories into the first block group of a flexgroup 1606 */ 1607 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1608 1609 /* 1610 * Structure of a directory entry 1611 */ 1612 #define EXT4_NAME_LEN 255 1613 1614 struct ext4_dir_entry { 1615 __le32 inode; /* Inode number */ 1616 __le16 rec_len; /* Directory entry length */ 1617 __le16 name_len; /* Name length */ 1618 char name[EXT4_NAME_LEN]; /* File name */ 1619 }; 1620 1621 /* 1622 * The new version of the directory entry. Since EXT4 structures are 1623 * stored in intel byte order, and the name_len field could never be 1624 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1625 * file_type field. 1626 */ 1627 struct ext4_dir_entry_2 { 1628 __le32 inode; /* Inode number */ 1629 __le16 rec_len; /* Directory entry length */ 1630 __u8 name_len; /* Name length */ 1631 __u8 file_type; 1632 char name[EXT4_NAME_LEN]; /* File name */ 1633 }; 1634 1635 /* 1636 * This is a bogus directory entry at the end of each leaf block that 1637 * records checksums. 1638 */ 1639 struct ext4_dir_entry_tail { 1640 __le32 det_reserved_zero1; /* Pretend to be unused */ 1641 __le16 det_rec_len; /* 12 */ 1642 __u8 det_reserved_zero2; /* Zero name length */ 1643 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1644 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1645 }; 1646 1647 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1648 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1649 ((blocksize) - \ 1650 sizeof(struct ext4_dir_entry_tail)))) 1651 1652 /* 1653 * Ext4 directory file types. Only the low 3 bits are used. The 1654 * other bits are reserved for now. 1655 */ 1656 #define EXT4_FT_UNKNOWN 0 1657 #define EXT4_FT_REG_FILE 1 1658 #define EXT4_FT_DIR 2 1659 #define EXT4_FT_CHRDEV 3 1660 #define EXT4_FT_BLKDEV 4 1661 #define EXT4_FT_FIFO 5 1662 #define EXT4_FT_SOCK 6 1663 #define EXT4_FT_SYMLINK 7 1664 1665 #define EXT4_FT_MAX 8 1666 1667 #define EXT4_FT_DIR_CSUM 0xDE 1668 1669 /* 1670 * EXT4_DIR_PAD defines the directory entries boundaries 1671 * 1672 * NOTE: It must be a multiple of 4 1673 */ 1674 #define EXT4_DIR_PAD 4 1675 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1676 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1677 ~EXT4_DIR_ROUND) 1678 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1679 1680 /* 1681 * If we ever get support for fs block sizes > page_size, we'll need 1682 * to remove the #if statements in the next two functions... 1683 */ 1684 static inline unsigned int 1685 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1686 { 1687 unsigned len = le16_to_cpu(dlen); 1688 1689 #if (PAGE_CACHE_SIZE >= 65536) 1690 if (len == EXT4_MAX_REC_LEN || len == 0) 1691 return blocksize; 1692 return (len & 65532) | ((len & 3) << 16); 1693 #else 1694 return len; 1695 #endif 1696 } 1697 1698 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1699 { 1700 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1701 BUG(); 1702 #if (PAGE_CACHE_SIZE >= 65536) 1703 if (len < 65536) 1704 return cpu_to_le16(len); 1705 if (len == blocksize) { 1706 if (blocksize == 65536) 1707 return cpu_to_le16(EXT4_MAX_REC_LEN); 1708 else 1709 return cpu_to_le16(0); 1710 } 1711 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1712 #else 1713 return cpu_to_le16(len); 1714 #endif 1715 } 1716 1717 /* 1718 * Hash Tree Directory indexing 1719 * (c) Daniel Phillips, 2001 1720 */ 1721 1722 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1723 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1724 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1725 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1726 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1727 1728 /* Legal values for the dx_root hash_version field: */ 1729 1730 #define DX_HASH_LEGACY 0 1731 #define DX_HASH_HALF_MD4 1 1732 #define DX_HASH_TEA 2 1733 #define DX_HASH_LEGACY_UNSIGNED 3 1734 #define DX_HASH_HALF_MD4_UNSIGNED 4 1735 #define DX_HASH_TEA_UNSIGNED 5 1736 1737 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1738 const void *address, unsigned int length) 1739 { 1740 struct { 1741 struct shash_desc shash; 1742 char ctx[4]; 1743 } desc; 1744 int err; 1745 1746 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1747 1748 desc.shash.tfm = sbi->s_chksum_driver; 1749 desc.shash.flags = 0; 1750 *(u32 *)desc.ctx = crc; 1751 1752 err = crypto_shash_update(&desc.shash, address, length); 1753 BUG_ON(err); 1754 1755 return *(u32 *)desc.ctx; 1756 } 1757 1758 #ifdef __KERNEL__ 1759 1760 /* hash info structure used by the directory hash */ 1761 struct dx_hash_info 1762 { 1763 u32 hash; 1764 u32 minor_hash; 1765 int hash_version; 1766 u32 *seed; 1767 }; 1768 1769 1770 /* 32 and 64 bit signed EOF for dx directories */ 1771 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1772 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1773 1774 1775 /* 1776 * Control parameters used by ext4_htree_next_block 1777 */ 1778 #define HASH_NB_ALWAYS 1 1779 1780 1781 /* 1782 * Describe an inode's exact location on disk and in memory 1783 */ 1784 struct ext4_iloc 1785 { 1786 struct buffer_head *bh; 1787 unsigned long offset; 1788 ext4_group_t block_group; 1789 }; 1790 1791 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1792 { 1793 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1794 } 1795 1796 /* 1797 * This structure is stuffed into the struct file's private_data field 1798 * for directories. It is where we put information so that we can do 1799 * readdir operations in hash tree order. 1800 */ 1801 struct dir_private_info { 1802 struct rb_root root; 1803 struct rb_node *curr_node; 1804 struct fname *extra_fname; 1805 loff_t last_pos; 1806 __u32 curr_hash; 1807 __u32 curr_minor_hash; 1808 __u32 next_hash; 1809 }; 1810 1811 /* calculate the first block number of the group */ 1812 static inline ext4_fsblk_t 1813 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1814 { 1815 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1816 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1817 } 1818 1819 /* 1820 * Special error return code only used by dx_probe() and its callers. 1821 */ 1822 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 1823 1824 /* 1825 * Timeout and state flag for lazy initialization inode thread. 1826 */ 1827 #define EXT4_DEF_LI_WAIT_MULT 10 1828 #define EXT4_DEF_LI_MAX_START_DELAY 5 1829 #define EXT4_LAZYINIT_QUIT 0x0001 1830 #define EXT4_LAZYINIT_RUNNING 0x0002 1831 1832 /* 1833 * Lazy inode table initialization info 1834 */ 1835 struct ext4_lazy_init { 1836 unsigned long li_state; 1837 struct list_head li_request_list; 1838 struct mutex li_list_mtx; 1839 }; 1840 1841 struct ext4_li_request { 1842 struct super_block *lr_super; 1843 struct ext4_sb_info *lr_sbi; 1844 ext4_group_t lr_next_group; 1845 struct list_head lr_request; 1846 unsigned long lr_next_sched; 1847 unsigned long lr_timeout; 1848 }; 1849 1850 struct ext4_features { 1851 struct kobject f_kobj; 1852 struct completion f_kobj_unregister; 1853 }; 1854 1855 /* 1856 * This structure will be used for multiple mount protection. It will be 1857 * written into the block number saved in the s_mmp_block field in the 1858 * superblock. Programs that check MMP should assume that if 1859 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1860 * to use the filesystem, regardless of how old the timestamp is. 1861 */ 1862 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1863 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1864 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1865 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1866 1867 struct mmp_struct { 1868 __le32 mmp_magic; /* Magic number for MMP */ 1869 __le32 mmp_seq; /* Sequence no. updated periodically */ 1870 1871 /* 1872 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1873 * purposes and do not affect the correctness of the algorithm 1874 */ 1875 __le64 mmp_time; /* Time last updated */ 1876 char mmp_nodename[64]; /* Node which last updated MMP block */ 1877 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1878 1879 /* 1880 * mmp_check_interval is used to verify if the MMP block has been 1881 * updated on the block device. The value is updated based on the 1882 * maximum time to write the MMP block during an update cycle. 1883 */ 1884 __le16 mmp_check_interval; 1885 1886 __le16 mmp_pad1; 1887 __le32 mmp_pad2[226]; 1888 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1889 }; 1890 1891 /* arguments passed to the mmp thread */ 1892 struct mmpd_data { 1893 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1894 struct super_block *sb; /* super block of the fs */ 1895 }; 1896 1897 /* 1898 * Check interval multiplier 1899 * The MMP block is written every update interval and initially checked every 1900 * update interval x the multiplier (the value is then adapted based on the 1901 * write latency). The reason is that writes can be delayed under load and we 1902 * don't want readers to incorrectly assume that the filesystem is no longer 1903 * in use. 1904 */ 1905 #define EXT4_MMP_CHECK_MULT 2UL 1906 1907 /* 1908 * Minimum interval for MMP checking in seconds. 1909 */ 1910 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1911 1912 /* 1913 * Maximum interval for MMP checking in seconds. 1914 */ 1915 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1916 1917 /* 1918 * Function prototypes 1919 */ 1920 1921 /* 1922 * Ok, these declarations are also in <linux/kernel.h> but none of the 1923 * ext4 source programs needs to include it so they are duplicated here. 1924 */ 1925 # define NORET_TYPE /**/ 1926 # define ATTRIB_NORET __attribute__((noreturn)) 1927 # define NORET_AND noreturn, 1928 1929 /* bitmap.c */ 1930 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1931 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1932 struct ext4_group_desc *gdp, 1933 struct buffer_head *bh, int sz); 1934 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1935 struct ext4_group_desc *gdp, 1936 struct buffer_head *bh, int sz); 1937 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1938 struct ext4_group_desc *gdp, 1939 struct buffer_head *bh); 1940 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1941 struct ext4_group_desc *gdp, 1942 struct buffer_head *bh); 1943 1944 /* balloc.c */ 1945 extern void ext4_get_group_no_and_offset(struct super_block *sb, 1946 ext4_fsblk_t blocknr, 1947 ext4_group_t *blockgrpp, 1948 ext4_grpblk_t *offsetp); 1949 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 1950 ext4_fsblk_t block); 1951 1952 extern unsigned int ext4_block_group(struct super_block *sb, 1953 ext4_fsblk_t blocknr); 1954 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1955 ext4_fsblk_t blocknr); 1956 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1957 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1958 ext4_group_t group); 1959 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1960 ext4_fsblk_t goal, 1961 unsigned int flags, 1962 unsigned long *count, 1963 int *errp); 1964 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1965 s64 nclusters, unsigned int flags); 1966 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1967 extern void ext4_check_blocks_bitmap(struct super_block *); 1968 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1969 ext4_group_t block_group, 1970 struct buffer_head ** bh); 1971 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1972 1973 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1974 ext4_group_t block_group); 1975 extern int ext4_wait_block_bitmap(struct super_block *sb, 1976 ext4_group_t block_group, 1977 struct buffer_head *bh); 1978 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1979 ext4_group_t block_group); 1980 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1981 ext4_group_t block_group, 1982 struct ext4_group_desc *gdp); 1983 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1984 1985 /* dir.c */ 1986 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1987 struct file *, 1988 struct ext4_dir_entry_2 *, 1989 struct buffer_head *, char *, int, 1990 unsigned int); 1991 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 1992 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 1993 (de), (bh), (buf), (size), (offset))) 1994 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1995 __u32 minor_hash, 1996 struct ext4_dir_entry_2 *dirent); 1997 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1998 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 1999 struct buffer_head *bh, 2000 void *buf, int buf_size, 2001 const char *name, int namelen, 2002 struct ext4_dir_entry_2 **dest_de); 2003 void ext4_insert_dentry(struct inode *inode, 2004 struct ext4_dir_entry_2 *de, 2005 int buf_size, 2006 const char *name, int namelen); 2007 static inline void ext4_update_dx_flag(struct inode *inode) 2008 { 2009 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 2010 EXT4_FEATURE_COMPAT_DIR_INDEX)) 2011 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2012 } 2013 static unsigned char ext4_filetype_table[] = { 2014 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2015 }; 2016 2017 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2018 { 2019 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 2020 (filetype >= EXT4_FT_MAX)) 2021 return DT_UNKNOWN; 2022 2023 return ext4_filetype_table[filetype]; 2024 } 2025 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2026 void *buf, int buf_size); 2027 2028 /* fsync.c */ 2029 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2030 2031 /* hash.c */ 2032 extern int ext4fs_dirhash(const char *name, int len, struct 2033 dx_hash_info *hinfo); 2034 2035 /* ialloc.c */ 2036 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2037 const struct qstr *qstr, __u32 goal, 2038 uid_t *owner, int handle_type, 2039 unsigned int line_no, int nblocks); 2040 2041 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2042 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2043 0, 0, 0) 2044 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2045 type, nblocks) \ 2046 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2047 (type), __LINE__, (nblocks)) 2048 2049 2050 extern void ext4_free_inode(handle_t *, struct inode *); 2051 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2052 extern unsigned long ext4_count_free_inodes(struct super_block *); 2053 extern unsigned long ext4_count_dirs(struct super_block *); 2054 extern void ext4_check_inodes_bitmap(struct super_block *); 2055 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2056 extern int ext4_init_inode_table(struct super_block *sb, 2057 ext4_group_t group, int barrier); 2058 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2059 2060 /* mballoc.c */ 2061 extern long ext4_mb_stats; 2062 extern long ext4_mb_max_to_scan; 2063 extern int ext4_mb_init(struct super_block *); 2064 extern int ext4_mb_release(struct super_block *); 2065 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2066 struct ext4_allocation_request *, int *); 2067 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2068 extern void ext4_discard_preallocations(struct inode *); 2069 extern int __init ext4_init_mballoc(void); 2070 extern void ext4_exit_mballoc(void); 2071 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2072 struct buffer_head *bh, ext4_fsblk_t block, 2073 unsigned long count, int flags); 2074 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2075 ext4_group_t ngroups); 2076 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2077 ext4_group_t i, struct ext4_group_desc *desc); 2078 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2079 ext4_fsblk_t block, unsigned long count); 2080 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2081 2082 /* inode.c */ 2083 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2084 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2085 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2086 struct buffer_head *bh_result, int create); 2087 int ext4_get_block(struct inode *inode, sector_t iblock, 2088 struct buffer_head *bh_result, int create); 2089 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2090 struct buffer_head *bh, int create); 2091 int ext4_walk_page_buffers(handle_t *handle, 2092 struct buffer_head *head, 2093 unsigned from, 2094 unsigned to, 2095 int *partial, 2096 int (*fn)(handle_t *handle, 2097 struct buffer_head *bh)); 2098 int do_journal_get_write_access(handle_t *handle, 2099 struct buffer_head *bh); 2100 #define FALL_BACK_TO_NONDELALLOC 1 2101 #define CONVERT_INLINE_DATA 2 2102 2103 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2104 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2105 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2106 extern int ext4_setattr(struct dentry *, struct iattr *); 2107 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2108 struct kstat *stat); 2109 extern void ext4_evict_inode(struct inode *); 2110 extern void ext4_clear_inode(struct inode *); 2111 extern int ext4_sync_inode(handle_t *, struct inode *); 2112 extern void ext4_dirty_inode(struct inode *, int); 2113 extern int ext4_change_inode_journal_flag(struct inode *, int); 2114 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2115 extern int ext4_inode_attach_jinode(struct inode *inode); 2116 extern int ext4_can_truncate(struct inode *inode); 2117 extern void ext4_truncate(struct inode *); 2118 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2119 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2120 extern void ext4_set_inode_flags(struct inode *); 2121 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2122 extern int ext4_alloc_da_blocks(struct inode *inode); 2123 extern void ext4_set_aops(struct inode *inode); 2124 extern int ext4_writepage_trans_blocks(struct inode *); 2125 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2126 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2127 loff_t lstart, loff_t lend); 2128 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2129 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2130 extern void ext4_da_update_reserve_space(struct inode *inode, 2131 int used, int quota_claim); 2132 2133 /* indirect.c */ 2134 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2135 struct ext4_map_blocks *map, int flags); 2136 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2137 struct iov_iter *iter, loff_t offset); 2138 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2139 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2140 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2141 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2142 ext4_lblk_t start, ext4_lblk_t end); 2143 2144 /* ioctl.c */ 2145 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2146 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2147 2148 /* migrate.c */ 2149 extern int ext4_ext_migrate(struct inode *); 2150 extern int ext4_ind_migrate(struct inode *inode); 2151 2152 /* namei.c */ 2153 extern int ext4_dirent_csum_verify(struct inode *inode, 2154 struct ext4_dir_entry *dirent); 2155 extern int ext4_orphan_add(handle_t *, struct inode *); 2156 extern int ext4_orphan_del(handle_t *, struct inode *); 2157 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2158 __u32 start_minor_hash, __u32 *next_hash); 2159 extern int search_dir(struct buffer_head *bh, 2160 char *search_buf, 2161 int buf_size, 2162 struct inode *dir, 2163 const struct qstr *d_name, 2164 unsigned int offset, 2165 struct ext4_dir_entry_2 **res_dir); 2166 extern int ext4_generic_delete_entry(handle_t *handle, 2167 struct inode *dir, 2168 struct ext4_dir_entry_2 *de_del, 2169 struct buffer_head *bh, 2170 void *entry_buf, 2171 int buf_size, 2172 int csum_size); 2173 2174 /* resize.c */ 2175 extern int ext4_group_add(struct super_block *sb, 2176 struct ext4_new_group_data *input); 2177 extern int ext4_group_extend(struct super_block *sb, 2178 struct ext4_super_block *es, 2179 ext4_fsblk_t n_blocks_count); 2180 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2181 2182 /* super.c */ 2183 extern int ext4_calculate_overhead(struct super_block *sb); 2184 extern void ext4_superblock_csum_set(struct super_block *sb); 2185 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2186 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2187 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2188 ext4_group_t ngroup); 2189 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2190 char nbuf[16]); 2191 2192 extern __printf(4, 5) 2193 void __ext4_error(struct super_block *, const char *, unsigned int, 2194 const char *, ...); 2195 extern __printf(5, 6) 2196 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2197 const char *, ...); 2198 extern __printf(5, 6) 2199 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2200 const char *, ...); 2201 extern void __ext4_std_error(struct super_block *, const char *, 2202 unsigned int, int); 2203 extern __printf(4, 5) 2204 void __ext4_abort(struct super_block *, const char *, unsigned int, 2205 const char *, ...); 2206 extern __printf(4, 5) 2207 void __ext4_warning(struct super_block *, const char *, unsigned int, 2208 const char *, ...); 2209 extern __printf(3, 4) 2210 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2211 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2212 const char *, unsigned int, const char *); 2213 extern __printf(7, 8) 2214 void __ext4_grp_locked_error(const char *, unsigned int, 2215 struct super_block *, ext4_group_t, 2216 unsigned long, ext4_fsblk_t, 2217 const char *, ...); 2218 2219 #ifdef CONFIG_PRINTK 2220 2221 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2222 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2223 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2224 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2225 #define ext4_error(sb, fmt, ...) \ 2226 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2227 #define ext4_abort(sb, fmt, ...) \ 2228 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2229 #define ext4_warning(sb, fmt, ...) \ 2230 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2231 #define ext4_msg(sb, level, fmt, ...) \ 2232 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2233 #define dump_mmp_msg(sb, mmp, msg) \ 2234 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2235 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2236 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2237 fmt, ##__VA_ARGS__) 2238 2239 #else 2240 2241 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2242 do { \ 2243 no_printk(fmt, ##__VA_ARGS__); \ 2244 __ext4_error_inode(inode, "", 0, block, " "); \ 2245 } while (0) 2246 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2247 do { \ 2248 no_printk(fmt, ##__VA_ARGS__); \ 2249 __ext4_error_file(file, "", 0, block, " "); \ 2250 } while (0) 2251 #define ext4_error(sb, fmt, ...) \ 2252 do { \ 2253 no_printk(fmt, ##__VA_ARGS__); \ 2254 __ext4_error(sb, "", 0, " "); \ 2255 } while (0) 2256 #define ext4_abort(sb, fmt, ...) \ 2257 do { \ 2258 no_printk(fmt, ##__VA_ARGS__); \ 2259 __ext4_abort(sb, "", 0, " "); \ 2260 } while (0) 2261 #define ext4_warning(sb, fmt, ...) \ 2262 do { \ 2263 no_printk(fmt, ##__VA_ARGS__); \ 2264 __ext4_warning(sb, "", 0, " "); \ 2265 } while (0) 2266 #define ext4_msg(sb, level, fmt, ...) \ 2267 do { \ 2268 no_printk(fmt, ##__VA_ARGS__); \ 2269 __ext4_msg(sb, "", " "); \ 2270 } while (0) 2271 #define dump_mmp_msg(sb, mmp, msg) \ 2272 __dump_mmp_msg(sb, mmp, "", 0, "") 2273 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2274 do { \ 2275 no_printk(fmt, ##__VA_ARGS__); \ 2276 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2277 } while (0) 2278 2279 #endif 2280 2281 extern void ext4_update_dynamic_rev(struct super_block *sb); 2282 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2283 __u32 compat); 2284 extern int ext4_update_rocompat_feature(handle_t *handle, 2285 struct super_block *sb, __u32 rocompat); 2286 extern int ext4_update_incompat_feature(handle_t *handle, 2287 struct super_block *sb, __u32 incompat); 2288 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2289 struct ext4_group_desc *bg); 2290 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2291 struct ext4_group_desc *bg); 2292 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2293 struct ext4_group_desc *bg); 2294 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2295 struct ext4_group_desc *bg); 2296 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2297 struct ext4_group_desc *bg); 2298 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2299 struct ext4_group_desc *bg); 2300 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2301 struct ext4_group_desc *bg); 2302 extern void ext4_block_bitmap_set(struct super_block *sb, 2303 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2304 extern void ext4_inode_bitmap_set(struct super_block *sb, 2305 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2306 extern void ext4_inode_table_set(struct super_block *sb, 2307 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2308 extern void ext4_free_group_clusters_set(struct super_block *sb, 2309 struct ext4_group_desc *bg, 2310 __u32 count); 2311 extern void ext4_free_inodes_set(struct super_block *sb, 2312 struct ext4_group_desc *bg, __u32 count); 2313 extern void ext4_used_dirs_set(struct super_block *sb, 2314 struct ext4_group_desc *bg, __u32 count); 2315 extern void ext4_itable_unused_set(struct super_block *sb, 2316 struct ext4_group_desc *bg, __u32 count); 2317 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2318 struct ext4_group_desc *gdp); 2319 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2320 struct ext4_group_desc *gdp); 2321 extern int ext4_register_li_request(struct super_block *sb, 2322 ext4_group_t first_not_zeroed); 2323 2324 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2325 { 2326 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2327 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) || 2328 (EXT4_SB(sb)->s_chksum_driver != NULL); 2329 } 2330 2331 static inline int ext4_has_metadata_csum(struct super_block *sb) 2332 { 2333 WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb, 2334 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) && 2335 !EXT4_SB(sb)->s_chksum_driver); 2336 2337 return (EXT4_SB(sb)->s_chksum_driver != NULL); 2338 } 2339 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2340 { 2341 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2342 le32_to_cpu(es->s_blocks_count_lo); 2343 } 2344 2345 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2346 { 2347 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2348 le32_to_cpu(es->s_r_blocks_count_lo); 2349 } 2350 2351 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2352 { 2353 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2354 le32_to_cpu(es->s_free_blocks_count_lo); 2355 } 2356 2357 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2358 ext4_fsblk_t blk) 2359 { 2360 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2361 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2362 } 2363 2364 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2365 ext4_fsblk_t blk) 2366 { 2367 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2368 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2369 } 2370 2371 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2372 ext4_fsblk_t blk) 2373 { 2374 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2375 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2376 } 2377 2378 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2379 { 2380 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2381 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2382 le32_to_cpu(raw_inode->i_size_lo); 2383 else 2384 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2385 } 2386 2387 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2388 { 2389 raw_inode->i_size_lo = cpu_to_le32(i_size); 2390 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2391 } 2392 2393 static inline 2394 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2395 ext4_group_t group) 2396 { 2397 struct ext4_group_info ***grp_info; 2398 long indexv, indexh; 2399 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2400 grp_info = EXT4_SB(sb)->s_group_info; 2401 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2402 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2403 return grp_info[indexv][indexh]; 2404 } 2405 2406 /* 2407 * Reading s_groups_count requires using smp_rmb() afterwards. See 2408 * the locking protocol documented in the comments of ext4_group_add() 2409 * in resize.c 2410 */ 2411 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2412 { 2413 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2414 2415 smp_rmb(); 2416 return ngroups; 2417 } 2418 2419 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2420 ext4_group_t block_group) 2421 { 2422 return block_group >> sbi->s_log_groups_per_flex; 2423 } 2424 2425 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2426 { 2427 return 1 << sbi->s_log_groups_per_flex; 2428 } 2429 2430 #define ext4_std_error(sb, errno) \ 2431 do { \ 2432 if ((errno)) \ 2433 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2434 } while (0) 2435 2436 #ifdef CONFIG_SMP 2437 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2438 * counters. So we need to make sure we have free clusters more 2439 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2440 */ 2441 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2442 #else 2443 #define EXT4_FREECLUSTERS_WATERMARK 0 2444 #endif 2445 2446 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2447 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2448 { 2449 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2450 !mutex_is_locked(&inode->i_mutex)); 2451 down_write(&EXT4_I(inode)->i_data_sem); 2452 if (newsize > EXT4_I(inode)->i_disksize) 2453 EXT4_I(inode)->i_disksize = newsize; 2454 up_write(&EXT4_I(inode)->i_data_sem); 2455 } 2456 2457 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2458 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2459 { 2460 int changed = 0; 2461 2462 if (newsize > inode->i_size) { 2463 i_size_write(inode, newsize); 2464 changed = 1; 2465 } 2466 if (newsize > EXT4_I(inode)->i_disksize) { 2467 ext4_update_i_disksize(inode, newsize); 2468 changed |= 2; 2469 } 2470 return changed; 2471 } 2472 2473 struct ext4_group_info { 2474 unsigned long bb_state; 2475 struct rb_root bb_free_root; 2476 ext4_grpblk_t bb_first_free; /* first free block */ 2477 ext4_grpblk_t bb_free; /* total free blocks */ 2478 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2479 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2480 struct list_head bb_prealloc_list; 2481 #ifdef DOUBLE_CHECK 2482 void *bb_bitmap; 2483 #endif 2484 struct rw_semaphore alloc_sem; 2485 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2486 * regions, index is order. 2487 * bb_counters[3] = 5 means 2488 * 5 free 8-block regions. */ 2489 }; 2490 2491 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2492 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2493 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2494 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2495 2496 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2497 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2498 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2499 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2500 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2501 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2502 2503 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2504 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2505 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2506 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2507 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2508 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2509 2510 #define EXT4_MAX_CONTENTION 8 2511 #define EXT4_CONTENTION_THRESHOLD 2 2512 2513 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2514 ext4_group_t group) 2515 { 2516 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2517 } 2518 2519 /* 2520 * Returns true if the filesystem is busy enough that attempts to 2521 * access the block group locks has run into contention. 2522 */ 2523 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2524 { 2525 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2526 } 2527 2528 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2529 { 2530 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2531 if (spin_trylock(lock)) 2532 /* 2533 * We're able to grab the lock right away, so drop the 2534 * lock contention counter. 2535 */ 2536 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2537 else { 2538 /* 2539 * The lock is busy, so bump the contention counter, 2540 * and then wait on the spin lock. 2541 */ 2542 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2543 EXT4_MAX_CONTENTION); 2544 spin_lock(lock); 2545 } 2546 } 2547 2548 static inline void ext4_unlock_group(struct super_block *sb, 2549 ext4_group_t group) 2550 { 2551 spin_unlock(ext4_group_lock_ptr(sb, group)); 2552 } 2553 2554 /* 2555 * Block validity checking 2556 */ 2557 #define ext4_check_indirect_blockref(inode, bh) \ 2558 ext4_check_blockref(__func__, __LINE__, inode, \ 2559 (__le32 *)(bh)->b_data, \ 2560 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2561 2562 #define ext4_ind_check_inode(inode) \ 2563 ext4_check_blockref(__func__, __LINE__, inode, \ 2564 EXT4_I(inode)->i_data, \ 2565 EXT4_NDIR_BLOCKS) 2566 2567 /* 2568 * Inodes and files operations 2569 */ 2570 2571 /* dir.c */ 2572 extern const struct file_operations ext4_dir_operations; 2573 2574 /* file.c */ 2575 extern const struct inode_operations ext4_file_inode_operations; 2576 extern const struct file_operations ext4_file_operations; 2577 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2578 2579 /* inline.c */ 2580 extern int ext4_get_max_inline_size(struct inode *inode); 2581 extern int ext4_find_inline_data_nolock(struct inode *inode); 2582 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2583 unsigned int len); 2584 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2585 2586 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2587 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2588 struct inode *inode, 2589 loff_t pos, unsigned len, 2590 unsigned flags, 2591 struct page **pagep); 2592 extern int ext4_write_inline_data_end(struct inode *inode, 2593 loff_t pos, unsigned len, 2594 unsigned copied, 2595 struct page *page); 2596 extern struct buffer_head * 2597 ext4_journalled_write_inline_data(struct inode *inode, 2598 unsigned len, 2599 struct page *page); 2600 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2601 struct inode *inode, 2602 loff_t pos, unsigned len, 2603 unsigned flags, 2604 struct page **pagep, 2605 void **fsdata); 2606 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2607 unsigned len, unsigned copied, 2608 struct page *page); 2609 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry, 2610 struct inode *inode); 2611 extern int ext4_try_create_inline_dir(handle_t *handle, 2612 struct inode *parent, 2613 struct inode *inode); 2614 extern int ext4_read_inline_dir(struct file *filp, 2615 struct dir_context *ctx, 2616 int *has_inline_data); 2617 extern int htree_inlinedir_to_tree(struct file *dir_file, 2618 struct inode *dir, ext4_lblk_t block, 2619 struct dx_hash_info *hinfo, 2620 __u32 start_hash, __u32 start_minor_hash, 2621 int *has_inline_data); 2622 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2623 const struct qstr *d_name, 2624 struct ext4_dir_entry_2 **res_dir, 2625 int *has_inline_data); 2626 extern int ext4_delete_inline_entry(handle_t *handle, 2627 struct inode *dir, 2628 struct ext4_dir_entry_2 *de_del, 2629 struct buffer_head *bh, 2630 int *has_inline_data); 2631 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 2632 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 2633 struct ext4_dir_entry_2 **parent_de, 2634 int *retval); 2635 extern int ext4_inline_data_fiemap(struct inode *inode, 2636 struct fiemap_extent_info *fieinfo, 2637 int *has_inline, __u64 start, __u64 len); 2638 extern int ext4_try_to_evict_inline_data(handle_t *handle, 2639 struct inode *inode, 2640 int needed); 2641 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 2642 2643 extern int ext4_convert_inline_data(struct inode *inode); 2644 2645 static inline int ext4_has_inline_data(struct inode *inode) 2646 { 2647 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 2648 EXT4_I(inode)->i_inline_off; 2649 } 2650 2651 /* namei.c */ 2652 extern const struct inode_operations ext4_dir_inode_operations; 2653 extern const struct inode_operations ext4_special_inode_operations; 2654 extern struct dentry *ext4_get_parent(struct dentry *child); 2655 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2656 struct ext4_dir_entry_2 *de, 2657 int blocksize, int csum_size, 2658 unsigned int parent_ino, int dotdot_real_len); 2659 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2660 unsigned int blocksize); 2661 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2662 struct inode *inode, 2663 struct buffer_head *bh); 2664 #define S_SHIFT 12 2665 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 2666 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 2667 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 2668 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 2669 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 2670 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 2671 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 2672 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 2673 }; 2674 2675 static inline void ext4_set_de_type(struct super_block *sb, 2676 struct ext4_dir_entry_2 *de, 2677 umode_t mode) { 2678 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE)) 2679 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 2680 } 2681 2682 2683 /* symlink.c */ 2684 extern const struct inode_operations ext4_symlink_inode_operations; 2685 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2686 2687 /* block_validity */ 2688 extern void ext4_release_system_zone(struct super_block *sb); 2689 extern int ext4_setup_system_zone(struct super_block *sb); 2690 extern int __init ext4_init_system_zone(void); 2691 extern void ext4_exit_system_zone(void); 2692 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2693 ext4_fsblk_t start_blk, 2694 unsigned int count); 2695 extern int ext4_check_blockref(const char *, unsigned int, 2696 struct inode *, __le32 *, unsigned int); 2697 2698 /* extents.c */ 2699 struct ext4_ext_path; 2700 struct ext4_extent; 2701 2702 /* 2703 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 2704 * __le32. 2705 */ 2706 #define EXT_MAX_BLOCKS 0xffffffff 2707 2708 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2709 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2710 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 2711 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2712 struct ext4_map_blocks *map, int flags); 2713 extern void ext4_ext_truncate(handle_t *, struct inode *); 2714 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 2715 ext4_lblk_t end); 2716 extern void ext4_ext_init(struct super_block *); 2717 extern void ext4_ext_release(struct super_block *); 2718 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2719 loff_t len); 2720 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 2721 loff_t offset, ssize_t len); 2722 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2723 struct ext4_map_blocks *map, int flags); 2724 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2725 ext4_lblk_t lblocks); 2726 extern int ext4_extent_tree_init(handle_t *, struct inode *); 2727 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2728 int num, 2729 struct ext4_ext_path *path); 2730 extern int ext4_can_extents_be_merged(struct inode *inode, 2731 struct ext4_extent *ex1, 2732 struct ext4_extent *ex2); 2733 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2734 struct ext4_ext_path **, 2735 struct ext4_extent *, int); 2736 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 2737 struct ext4_ext_path **, 2738 int flags); 2739 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2740 extern int ext4_ext_check_inode(struct inode *inode); 2741 extern int ext4_find_delalloc_range(struct inode *inode, 2742 ext4_lblk_t lblk_start, 2743 ext4_lblk_t lblk_end); 2744 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2745 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 2746 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2747 __u64 start, __u64 len); 2748 extern int ext4_ext_precache(struct inode *inode); 2749 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 2750 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 2751 struct inode *inode2, ext4_lblk_t lblk1, 2752 ext4_lblk_t lblk2, ext4_lblk_t count, 2753 int mark_unwritten,int *err); 2754 2755 /* move_extent.c */ 2756 extern void ext4_double_down_write_data_sem(struct inode *first, 2757 struct inode *second); 2758 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 2759 struct inode *donor_inode); 2760 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2761 __u64 start_orig, __u64 start_donor, 2762 __u64 len, __u64 *moved_len); 2763 2764 /* page-io.c */ 2765 extern int __init ext4_init_pageio(void); 2766 extern void ext4_exit_pageio(void); 2767 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2768 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 2769 extern int ext4_put_io_end(ext4_io_end_t *io_end); 2770 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 2771 extern void ext4_io_submit_init(struct ext4_io_submit *io, 2772 struct writeback_control *wbc); 2773 extern void ext4_end_io_rsv_work(struct work_struct *work); 2774 extern void ext4_io_submit(struct ext4_io_submit *io); 2775 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2776 struct page *page, 2777 int len, 2778 struct writeback_control *wbc, 2779 bool keep_towrite); 2780 2781 /* mmp.c */ 2782 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2783 2784 /* 2785 * Add new method to test whether block and inode bitmaps are properly 2786 * initialized. With uninit_bg reading the block from disk is not enough 2787 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2788 */ 2789 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2790 2791 static inline int bitmap_uptodate(struct buffer_head *bh) 2792 { 2793 return (buffer_uptodate(bh) && 2794 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2795 } 2796 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2797 { 2798 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2799 } 2800 2801 /* 2802 * Disable DIO read nolock optimization, so new dioreaders will be forced 2803 * to grab i_mutex 2804 */ 2805 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2806 { 2807 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2808 smp_mb(); 2809 } 2810 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2811 { 2812 smp_mb(); 2813 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2814 } 2815 2816 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2817 2818 /* For ioend & aio unwritten conversion wait queues */ 2819 #define EXT4_WQ_HASH_SZ 37 2820 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2821 EXT4_WQ_HASH_SZ]) 2822 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2823 EXT4_WQ_HASH_SZ]) 2824 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2825 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2826 2827 #define EXT4_RESIZING 0 2828 extern int ext4_resize_begin(struct super_block *sb); 2829 extern void ext4_resize_end(struct super_block *sb); 2830 2831 #endif /* __KERNEL__ */ 2832 2833 #endif /* _EXT4_H */ 2834