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_UNINIT (1 << BH_Uninit) 162 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of 163 * ext4_map_blocks wants to know whether or not the underlying cluster has 164 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that 165 * the requested mapping was from previously mapped (or delayed allocated) 166 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster 167 * should never appear on buffer_head's state flags. 168 */ 169 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) 170 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 171 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 172 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) 173 174 struct ext4_map_blocks { 175 ext4_fsblk_t m_pblk; 176 ext4_lblk_t m_lblk; 177 unsigned int m_len; 178 unsigned int m_flags; 179 }; 180 181 /* 182 * Flags for ext4_io_end->flags 183 */ 184 #define EXT4_IO_END_UNWRITTEN 0x0001 185 186 /* 187 * For converting uninitialized extents on a work queue. 'handle' is used for 188 * buffered writeback. 189 */ 190 typedef struct ext4_io_end { 191 struct list_head list; /* per-file finished IO list */ 192 handle_t *handle; /* handle reserved for extent 193 * conversion */ 194 struct inode *inode; /* file being written to */ 195 struct bio *bio; /* Linked list of completed 196 * bios covering the extent */ 197 unsigned int flag; /* unwritten or not */ 198 loff_t offset; /* offset in the file */ 199 ssize_t size; /* size of the extent */ 200 atomic_t count; /* reference counter */ 201 } ext4_io_end_t; 202 203 struct ext4_io_submit { 204 int io_op; 205 struct bio *io_bio; 206 ext4_io_end_t *io_end; 207 sector_t io_next_block; 208 }; 209 210 /* 211 * Special inodes numbers 212 */ 213 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 214 #define EXT4_ROOT_INO 2 /* Root inode */ 215 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 216 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 217 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 218 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 219 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 220 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 221 222 /* First non-reserved inode for old ext4 filesystems */ 223 #define EXT4_GOOD_OLD_FIRST_INO 11 224 225 /* 226 * Maximal count of links to a file 227 */ 228 #define EXT4_LINK_MAX 65000 229 230 /* 231 * Macro-instructions used to manage several block sizes 232 */ 233 #define EXT4_MIN_BLOCK_SIZE 1024 234 #define EXT4_MAX_BLOCK_SIZE 65536 235 #define EXT4_MIN_BLOCK_LOG_SIZE 10 236 #define EXT4_MAX_BLOCK_LOG_SIZE 16 237 #ifdef __KERNEL__ 238 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 239 #else 240 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 241 #endif 242 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 243 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 244 EXT4_SB(s)->s_cluster_bits) 245 #ifdef __KERNEL__ 246 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 247 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 248 #else 249 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 250 #endif 251 #ifdef __KERNEL__ 252 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 253 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 254 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 255 #else 256 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 257 EXT4_GOOD_OLD_INODE_SIZE : \ 258 (s)->s_inode_size) 259 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 260 EXT4_GOOD_OLD_FIRST_INO : \ 261 (s)->s_first_ino) 262 #endif 263 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 264 265 /* Translate a block number to a cluster number */ 266 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 267 /* Translate a cluster number to a block number */ 268 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 269 /* Translate # of blks to # of clusters */ 270 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 271 (sbi)->s_cluster_bits) 272 /* Mask out the low bits to get the starting block of the cluster */ 273 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 274 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 275 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 276 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 277 /* Get the cluster offset */ 278 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 279 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 280 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 281 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 282 283 /* 284 * Structure of a blocks group descriptor 285 */ 286 struct ext4_group_desc 287 { 288 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 289 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 290 __le32 bg_inode_table_lo; /* Inodes table block */ 291 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 292 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 293 __le16 bg_used_dirs_count_lo; /* Directories count */ 294 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 295 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 296 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 297 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 298 __le16 bg_itable_unused_lo; /* Unused inodes count */ 299 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 300 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 301 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 302 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 303 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 304 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 305 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 306 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 307 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 308 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 309 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 310 __u32 bg_reserved; 311 }; 312 313 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 314 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 315 sizeof(__le16)) 316 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 317 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 318 sizeof(__le16)) 319 320 /* 321 * Structure of a flex block group info 322 */ 323 324 struct flex_groups { 325 atomic64_t free_clusters; 326 atomic_t free_inodes; 327 atomic_t used_dirs; 328 }; 329 330 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 331 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 332 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 333 334 /* 335 * Macro-instructions used to manage group descriptors 336 */ 337 #define EXT4_MIN_DESC_SIZE 32 338 #define EXT4_MIN_DESC_SIZE_64BIT 64 339 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 340 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 341 #ifdef __KERNEL__ 342 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 343 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 344 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 345 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 346 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 347 #else 348 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 349 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 350 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 351 #endif 352 353 /* 354 * Constants relative to the data blocks 355 */ 356 #define EXT4_NDIR_BLOCKS 12 357 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 358 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 359 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 360 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 361 362 /* 363 * Inode flags 364 */ 365 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 366 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 367 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 368 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 369 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 370 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 371 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 372 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 373 /* Reserved for compression usage... */ 374 #define EXT4_DIRTY_FL 0x00000100 375 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 376 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 377 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 378 /* End compression flags --- maybe not all used */ 379 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 380 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 381 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 382 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 383 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 384 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 385 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 386 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 387 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 388 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 389 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 390 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 391 392 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 393 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 394 395 /* Flags that should be inherited by new inodes from their parent. */ 396 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 397 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 398 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 399 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 400 401 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 402 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 403 404 /* Flags that are appropriate for non-directories/regular files. */ 405 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 406 407 /* Mask out flags that are inappropriate for the given type of inode. */ 408 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 409 { 410 if (S_ISDIR(mode)) 411 return flags; 412 else if (S_ISREG(mode)) 413 return flags & EXT4_REG_FLMASK; 414 else 415 return flags & EXT4_OTHER_FLMASK; 416 } 417 418 /* 419 * Inode flags used for atomic set/get 420 */ 421 enum { 422 EXT4_INODE_SECRM = 0, /* Secure deletion */ 423 EXT4_INODE_UNRM = 1, /* Undelete */ 424 EXT4_INODE_COMPR = 2, /* Compress file */ 425 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 426 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 427 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 428 EXT4_INODE_NODUMP = 6, /* do not dump file */ 429 EXT4_INODE_NOATIME = 7, /* do not update atime */ 430 /* Reserved for compression usage... */ 431 EXT4_INODE_DIRTY = 8, 432 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 433 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 434 EXT4_INODE_ECOMPR = 11, /* Compression error */ 435 /* End compression flags --- maybe not all used */ 436 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 437 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 438 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 439 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 440 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 441 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 442 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 443 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 444 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 445 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 446 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 447 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 448 }; 449 450 /* 451 * Since it's pretty easy to mix up bit numbers and hex values, we use a 452 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 453 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 454 * any extra space in the compiled kernel image, otherwise, the build will fail. 455 * It's important that these values are the same, since we are using 456 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 457 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 458 * values found in ext2, ext3 and ext4 filesystems, and of course the values 459 * defined in e2fsprogs. 460 * 461 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 462 */ 463 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 464 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 465 466 static inline void ext4_check_flag_values(void) 467 { 468 CHECK_FLAG_VALUE(SECRM); 469 CHECK_FLAG_VALUE(UNRM); 470 CHECK_FLAG_VALUE(COMPR); 471 CHECK_FLAG_VALUE(SYNC); 472 CHECK_FLAG_VALUE(IMMUTABLE); 473 CHECK_FLAG_VALUE(APPEND); 474 CHECK_FLAG_VALUE(NODUMP); 475 CHECK_FLAG_VALUE(NOATIME); 476 CHECK_FLAG_VALUE(DIRTY); 477 CHECK_FLAG_VALUE(COMPRBLK); 478 CHECK_FLAG_VALUE(NOCOMPR); 479 CHECK_FLAG_VALUE(ECOMPR); 480 CHECK_FLAG_VALUE(INDEX); 481 CHECK_FLAG_VALUE(IMAGIC); 482 CHECK_FLAG_VALUE(JOURNAL_DATA); 483 CHECK_FLAG_VALUE(NOTAIL); 484 CHECK_FLAG_VALUE(DIRSYNC); 485 CHECK_FLAG_VALUE(TOPDIR); 486 CHECK_FLAG_VALUE(HUGE_FILE); 487 CHECK_FLAG_VALUE(EXTENTS); 488 CHECK_FLAG_VALUE(EA_INODE); 489 CHECK_FLAG_VALUE(EOFBLOCKS); 490 CHECK_FLAG_VALUE(INLINE_DATA); 491 CHECK_FLAG_VALUE(RESERVED); 492 } 493 494 /* Used to pass group descriptor data when online resize is done */ 495 struct ext4_new_group_input { 496 __u32 group; /* Group number for this data */ 497 __u64 block_bitmap; /* Absolute block number of block bitmap */ 498 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 499 __u64 inode_table; /* Absolute block number of inode table start */ 500 __u32 blocks_count; /* Total number of blocks in this group */ 501 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 502 __u16 unused; 503 }; 504 505 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 506 struct compat_ext4_new_group_input { 507 u32 group; 508 compat_u64 block_bitmap; 509 compat_u64 inode_bitmap; 510 compat_u64 inode_table; 511 u32 blocks_count; 512 u16 reserved_blocks; 513 u16 unused; 514 }; 515 #endif 516 517 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 518 struct ext4_new_group_data { 519 __u32 group; 520 __u64 block_bitmap; 521 __u64 inode_bitmap; 522 __u64 inode_table; 523 __u32 blocks_count; 524 __u16 reserved_blocks; 525 __u16 unused; 526 __u32 free_blocks_count; 527 }; 528 529 /* Indexes used to index group tables in ext4_new_group_data */ 530 enum { 531 BLOCK_BITMAP = 0, /* block bitmap */ 532 INODE_BITMAP, /* inode bitmap */ 533 INODE_TABLE, /* inode tables */ 534 GROUP_TABLE_COUNT, 535 }; 536 537 /* 538 * Flags used by ext4_map_blocks() 539 */ 540 /* Allocate any needed blocks and/or convert an unitialized 541 extent to be an initialized ext4 */ 542 #define EXT4_GET_BLOCKS_CREATE 0x0001 543 /* Request the creation of an unitialized extent */ 544 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 545 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 546 EXT4_GET_BLOCKS_CREATE) 547 /* Caller is from the delayed allocation writeout path 548 * finally doing the actual allocation of delayed blocks */ 549 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 550 /* caller is from the direct IO path, request to creation of an 551 unitialized extents if not allocated, split the uninitialized 552 extent if blocks has been preallocated already*/ 553 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 554 #define EXT4_GET_BLOCKS_CONVERT 0x0010 555 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 556 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 557 /* Convert extent to initialized after IO complete */ 558 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 559 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 560 /* Eventual metadata allocation (due to growing extent tree) 561 * should not fail, so try to use reserved blocks for that.*/ 562 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 563 /* Don't normalize allocation size (used for fallocate) */ 564 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 565 /* Request will not result in inode size update (user for fallocate) */ 566 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 567 /* Do not take i_data_sem locking in ext4_map_blocks */ 568 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100 569 /* Do not put hole in extent cache */ 570 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200 571 572 /* 573 * The bit position of these flags must not overlap with any of the 574 * EXT4_GET_BLOCKS_*. They are used by ext4_ext_find_extent(), 575 * read_extent_tree_block(), ext4_split_extent_at(), 576 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 577 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 578 * caching the extents when reading from the extent tree while a 579 * truncate or punch hole operation is in progress. 580 */ 581 #define EXT4_EX_NOCACHE 0x0400 582 #define EXT4_EX_FORCE_CACHE 0x0800 583 584 /* 585 * Flags used by ext4_free_blocks 586 */ 587 #define EXT4_FREE_BLOCKS_METADATA 0x0001 588 #define EXT4_FREE_BLOCKS_FORGET 0x0002 589 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 590 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 591 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 592 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 593 #define EXT4_FREE_BLOCKS_RESERVE 0x0040 594 595 /* 596 * ioctl commands 597 */ 598 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 599 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 600 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 601 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 602 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 603 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 604 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 605 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 606 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 607 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 608 #define EXT4_IOC_MIGRATE _IO('f', 9) 609 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 610 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 611 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 612 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 613 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 614 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 615 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 616 617 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 618 /* 619 * ioctl commands in 32 bit emulation 620 */ 621 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 622 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 623 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 624 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 625 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 626 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 627 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 628 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 629 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 630 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 631 #endif 632 633 /* Max physical block we can address w/o extents */ 634 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 635 636 /* 637 * Structure of an inode on the disk 638 */ 639 struct ext4_inode { 640 __le16 i_mode; /* File mode */ 641 __le16 i_uid; /* Low 16 bits of Owner Uid */ 642 __le32 i_size_lo; /* Size in bytes */ 643 __le32 i_atime; /* Access time */ 644 __le32 i_ctime; /* Inode Change time */ 645 __le32 i_mtime; /* Modification time */ 646 __le32 i_dtime; /* Deletion Time */ 647 __le16 i_gid; /* Low 16 bits of Group Id */ 648 __le16 i_links_count; /* Links count */ 649 __le32 i_blocks_lo; /* Blocks count */ 650 __le32 i_flags; /* File flags */ 651 union { 652 struct { 653 __le32 l_i_version; 654 } linux1; 655 struct { 656 __u32 h_i_translator; 657 } hurd1; 658 struct { 659 __u32 m_i_reserved1; 660 } masix1; 661 } osd1; /* OS dependent 1 */ 662 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 663 __le32 i_generation; /* File version (for NFS) */ 664 __le32 i_file_acl_lo; /* File ACL */ 665 __le32 i_size_high; 666 __le32 i_obso_faddr; /* Obsoleted fragment address */ 667 union { 668 struct { 669 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 670 __le16 l_i_file_acl_high; 671 __le16 l_i_uid_high; /* these 2 fields */ 672 __le16 l_i_gid_high; /* were reserved2[0] */ 673 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 674 __le16 l_i_reserved; 675 } linux2; 676 struct { 677 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 678 __u16 h_i_mode_high; 679 __u16 h_i_uid_high; 680 __u16 h_i_gid_high; 681 __u32 h_i_author; 682 } hurd2; 683 struct { 684 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 685 __le16 m_i_file_acl_high; 686 __u32 m_i_reserved2[2]; 687 } masix2; 688 } osd2; /* OS dependent 2 */ 689 __le16 i_extra_isize; 690 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 691 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 692 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 693 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 694 __le32 i_crtime; /* File Creation time */ 695 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 696 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 697 }; 698 699 struct move_extent { 700 __u32 reserved; /* should be zero */ 701 __u32 donor_fd; /* donor file descriptor */ 702 __u64 orig_start; /* logical start offset in block for orig */ 703 __u64 donor_start; /* logical start offset in block for donor */ 704 __u64 len; /* block length to be moved */ 705 __u64 moved_len; /* moved block length */ 706 }; 707 708 #define EXT4_EPOCH_BITS 2 709 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 710 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 711 712 /* 713 * Extended fields will fit into an inode if the filesystem was formatted 714 * with large inodes (-I 256 or larger) and there are not currently any EAs 715 * consuming all of the available space. For new inodes we always reserve 716 * enough space for the kernel's known extended fields, but for inodes 717 * created with an old kernel this might not have been the case. None of 718 * the extended inode fields is critical for correct filesystem operation. 719 * This macro checks if a certain field fits in the inode. Note that 720 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 721 */ 722 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 723 ((offsetof(typeof(*ext4_inode), field) + \ 724 sizeof((ext4_inode)->field)) \ 725 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 726 (einode)->i_extra_isize)) \ 727 728 static inline __le32 ext4_encode_extra_time(struct timespec *time) 729 { 730 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 731 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 732 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 733 } 734 735 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 736 { 737 if (sizeof(time->tv_sec) > 4) 738 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 739 << 32; 740 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 741 } 742 743 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 744 do { \ 745 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 746 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 747 (raw_inode)->xtime ## _extra = \ 748 ext4_encode_extra_time(&(inode)->xtime); \ 749 } while (0) 750 751 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 752 do { \ 753 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 754 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 755 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 756 (raw_inode)->xtime ## _extra = \ 757 ext4_encode_extra_time(&(einode)->xtime); \ 758 } while (0) 759 760 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 761 do { \ 762 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 763 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 764 ext4_decode_extra_time(&(inode)->xtime, \ 765 raw_inode->xtime ## _extra); \ 766 else \ 767 (inode)->xtime.tv_nsec = 0; \ 768 } while (0) 769 770 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 771 do { \ 772 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 773 (einode)->xtime.tv_sec = \ 774 (signed)le32_to_cpu((raw_inode)->xtime); \ 775 else \ 776 (einode)->xtime.tv_sec = 0; \ 777 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 778 ext4_decode_extra_time(&(einode)->xtime, \ 779 raw_inode->xtime ## _extra); \ 780 else \ 781 (einode)->xtime.tv_nsec = 0; \ 782 } while (0) 783 784 #define i_disk_version osd1.linux1.l_i_version 785 786 #if defined(__KERNEL__) || defined(__linux__) 787 #define i_reserved1 osd1.linux1.l_i_reserved1 788 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 789 #define i_blocks_high osd2.linux2.l_i_blocks_high 790 #define i_uid_low i_uid 791 #define i_gid_low i_gid 792 #define i_uid_high osd2.linux2.l_i_uid_high 793 #define i_gid_high osd2.linux2.l_i_gid_high 794 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 795 796 #elif defined(__GNU__) 797 798 #define i_translator osd1.hurd1.h_i_translator 799 #define i_uid_high osd2.hurd2.h_i_uid_high 800 #define i_gid_high osd2.hurd2.h_i_gid_high 801 #define i_author osd2.hurd2.h_i_author 802 803 #elif defined(__masix__) 804 805 #define i_reserved1 osd1.masix1.m_i_reserved1 806 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 807 #define i_reserved2 osd2.masix2.m_i_reserved2 808 809 #endif /* defined(__KERNEL__) || defined(__linux__) */ 810 811 #include "extents_status.h" 812 813 /* 814 * fourth extended file system inode data in memory 815 */ 816 struct ext4_inode_info { 817 __le32 i_data[15]; /* unconverted */ 818 __u32 i_dtime; 819 ext4_fsblk_t i_file_acl; 820 821 /* 822 * i_block_group is the number of the block group which contains 823 * this file's inode. Constant across the lifetime of the inode, 824 * it is ued for making block allocation decisions - we try to 825 * place a file's data blocks near its inode block, and new inodes 826 * near to their parent directory's inode. 827 */ 828 ext4_group_t i_block_group; 829 ext4_lblk_t i_dir_start_lookup; 830 #if (BITS_PER_LONG < 64) 831 unsigned long i_state_flags; /* Dynamic state flags */ 832 #endif 833 unsigned long i_flags; 834 835 /* 836 * Extended attributes can be read independently of the main file 837 * data. Taking i_mutex even when reading would cause contention 838 * between readers of EAs and writers of regular file data, so 839 * instead we synchronize on xattr_sem when reading or changing 840 * EAs. 841 */ 842 struct rw_semaphore xattr_sem; 843 844 struct list_head i_orphan; /* unlinked but open inodes */ 845 846 /* 847 * i_disksize keeps track of what the inode size is ON DISK, not 848 * in memory. During truncate, i_size is set to the new size by 849 * the VFS prior to calling ext4_truncate(), but the filesystem won't 850 * set i_disksize to 0 until the truncate is actually under way. 851 * 852 * The intent is that i_disksize always represents the blocks which 853 * are used by this file. This allows recovery to restart truncate 854 * on orphans if we crash during truncate. We actually write i_disksize 855 * into the on-disk inode when writing inodes out, instead of i_size. 856 * 857 * The only time when i_disksize and i_size may be different is when 858 * a truncate is in progress. The only things which change i_disksize 859 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 860 */ 861 loff_t i_disksize; 862 863 /* 864 * i_data_sem is for serialising ext4_truncate() against 865 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 866 * data tree are chopped off during truncate. We can't do that in 867 * ext4 because whenever we perform intermediate commits during 868 * truncate, the inode and all the metadata blocks *must* be in a 869 * consistent state which allows truncation of the orphans to restart 870 * during recovery. Hence we must fix the get_block-vs-truncate race 871 * by other means, so we have i_data_sem. 872 */ 873 struct rw_semaphore i_data_sem; 874 struct inode vfs_inode; 875 struct jbd2_inode *jinode; 876 877 /* 878 * File creation time. Its function is same as that of 879 * struct timespec i_{a,c,m}time in the generic inode. 880 */ 881 struct timespec i_crtime; 882 883 /* mballoc */ 884 struct list_head i_prealloc_list; 885 spinlock_t i_prealloc_lock; 886 887 /* extents status tree */ 888 struct ext4_es_tree i_es_tree; 889 rwlock_t i_es_lock; 890 struct list_head i_es_lru; 891 unsigned int i_es_lru_nr; /* protected by i_es_lock */ 892 unsigned long i_touch_when; /* jiffies of last accessing */ 893 894 /* ialloc */ 895 ext4_group_t i_last_alloc_group; 896 897 /* allocation reservation info for delalloc */ 898 /* In case of bigalloc, these refer to clusters rather than blocks */ 899 unsigned int i_reserved_data_blocks; 900 unsigned int i_reserved_meta_blocks; 901 unsigned int i_allocated_meta_blocks; 902 ext4_lblk_t i_da_metadata_calc_last_lblock; 903 int i_da_metadata_calc_len; 904 905 /* on-disk additional length */ 906 __u16 i_extra_isize; 907 908 /* Indicate the inline data space. */ 909 u16 i_inline_off; 910 u16 i_inline_size; 911 912 #ifdef CONFIG_QUOTA 913 /* quota space reservation, managed internally by quota code */ 914 qsize_t i_reserved_quota; 915 #endif 916 917 /* Lock protecting lists below */ 918 spinlock_t i_completed_io_lock; 919 /* 920 * Completed IOs that need unwritten extents handling and have 921 * transaction reserved 922 */ 923 struct list_head i_rsv_conversion_list; 924 /* 925 * Completed IOs that need unwritten extents handling and don't have 926 * transaction reserved 927 */ 928 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 929 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 930 struct work_struct i_rsv_conversion_work; 931 932 spinlock_t i_block_reservation_lock; 933 934 /* 935 * Transactions that contain inode's metadata needed to complete 936 * fsync and fdatasync, respectively. 937 */ 938 tid_t i_sync_tid; 939 tid_t i_datasync_tid; 940 941 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 942 __u32 i_csum_seed; 943 }; 944 945 /* 946 * File system states 947 */ 948 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 949 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 950 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 951 952 /* 953 * Misc. filesystem flags 954 */ 955 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 956 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 957 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 958 959 /* 960 * Mount flags set via mount options or defaults 961 */ 962 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 963 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 964 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 965 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 966 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 967 #define EXT4_MOUNT_ERRORS_MASK 0x00070 968 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 969 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 970 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 971 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 972 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 973 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 974 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 975 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 976 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 977 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 978 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 979 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 980 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 981 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 982 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 983 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 984 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 985 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 986 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 987 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 988 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 989 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 990 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 991 992 /* 993 * Mount flags set either automatically (could not be set by mount option) 994 * based on per file system feature or property or in special cases such as 995 * distinguishing between explicit mount option definition and default. 996 */ 997 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 998 specified delalloc */ 999 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1000 size of blocksize * 8 1001 blocks */ 1002 1003 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1004 ~EXT4_MOUNT_##opt 1005 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1006 EXT4_MOUNT_##opt 1007 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1008 EXT4_MOUNT_##opt) 1009 1010 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1011 ~EXT4_MOUNT2_##opt 1012 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1013 EXT4_MOUNT2_##opt 1014 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1015 EXT4_MOUNT2_##opt) 1016 1017 #define ext4_test_and_set_bit __test_and_set_bit_le 1018 #define ext4_set_bit __set_bit_le 1019 #define ext4_set_bit_atomic ext2_set_bit_atomic 1020 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1021 #define ext4_clear_bit __clear_bit_le 1022 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1023 #define ext4_test_bit test_bit_le 1024 #define ext4_find_next_zero_bit find_next_zero_bit_le 1025 #define ext4_find_next_bit find_next_bit_le 1026 1027 extern void ext4_set_bits(void *bm, int cur, int len); 1028 1029 /* 1030 * Maximal mount counts between two filesystem checks 1031 */ 1032 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1033 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1034 1035 /* 1036 * Behaviour when detecting errors 1037 */ 1038 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1039 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1040 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1041 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1042 1043 /* Metadata checksum algorithm codes */ 1044 #define EXT4_CRC32C_CHKSUM 1 1045 1046 /* 1047 * Structure of the super block 1048 */ 1049 struct ext4_super_block { 1050 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1051 __le32 s_blocks_count_lo; /* Blocks count */ 1052 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1053 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1054 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1055 __le32 s_first_data_block; /* First Data Block */ 1056 __le32 s_log_block_size; /* Block size */ 1057 __le32 s_log_cluster_size; /* Allocation cluster size */ 1058 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1059 __le32 s_clusters_per_group; /* # Clusters per group */ 1060 __le32 s_inodes_per_group; /* # Inodes per group */ 1061 __le32 s_mtime; /* Mount time */ 1062 /*30*/ __le32 s_wtime; /* Write time */ 1063 __le16 s_mnt_count; /* Mount count */ 1064 __le16 s_max_mnt_count; /* Maximal mount count */ 1065 __le16 s_magic; /* Magic signature */ 1066 __le16 s_state; /* File system state */ 1067 __le16 s_errors; /* Behaviour when detecting errors */ 1068 __le16 s_minor_rev_level; /* minor revision level */ 1069 /*40*/ __le32 s_lastcheck; /* time of last check */ 1070 __le32 s_checkinterval; /* max. time between checks */ 1071 __le32 s_creator_os; /* OS */ 1072 __le32 s_rev_level; /* Revision level */ 1073 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1074 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1075 /* 1076 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1077 * 1078 * Note: the difference between the compatible feature set and 1079 * the incompatible feature set is that if there is a bit set 1080 * in the incompatible feature set that the kernel doesn't 1081 * know about, it should refuse to mount the filesystem. 1082 * 1083 * e2fsck's requirements are more strict; if it doesn't know 1084 * about a feature in either the compatible or incompatible 1085 * feature set, it must abort and not try to meddle with 1086 * things it doesn't understand... 1087 */ 1088 __le32 s_first_ino; /* First non-reserved inode */ 1089 __le16 s_inode_size; /* size of inode structure */ 1090 __le16 s_block_group_nr; /* block group # of this superblock */ 1091 __le32 s_feature_compat; /* compatible feature set */ 1092 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1093 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1094 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1095 /*78*/ char s_volume_name[16]; /* volume name */ 1096 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1097 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1098 /* 1099 * Performance hints. Directory preallocation should only 1100 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1101 */ 1102 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1103 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1104 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1105 /* 1106 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1107 */ 1108 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1109 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1110 __le32 s_journal_dev; /* device number of journal file */ 1111 __le32 s_last_orphan; /* start of list of inodes to delete */ 1112 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1113 __u8 s_def_hash_version; /* Default hash version to use */ 1114 __u8 s_jnl_backup_type; 1115 __le16 s_desc_size; /* size of group descriptor */ 1116 /*100*/ __le32 s_default_mount_opts; 1117 __le32 s_first_meta_bg; /* First metablock block group */ 1118 __le32 s_mkfs_time; /* When the filesystem was created */ 1119 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1120 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1121 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1122 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1123 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1124 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1125 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1126 __le32 s_flags; /* Miscellaneous flags */ 1127 __le16 s_raid_stride; /* RAID stride */ 1128 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1129 __le64 s_mmp_block; /* Block for multi-mount protection */ 1130 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1131 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1132 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1133 __le16 s_reserved_pad; 1134 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1135 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1136 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1137 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1138 snapshot's future use */ 1139 __le32 s_snapshot_list; /* inode number of the head of the 1140 on-disk snapshot list */ 1141 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1142 __le32 s_error_count; /* number of fs errors */ 1143 __le32 s_first_error_time; /* first time an error happened */ 1144 __le32 s_first_error_ino; /* inode involved in first error */ 1145 __le64 s_first_error_block; /* block involved of first error */ 1146 __u8 s_first_error_func[32]; /* function where the error happened */ 1147 __le32 s_first_error_line; /* line number where error happened */ 1148 __le32 s_last_error_time; /* most recent time of an error */ 1149 __le32 s_last_error_ino; /* inode involved in last error */ 1150 __le32 s_last_error_line; /* line number where error happened */ 1151 __le64 s_last_error_block; /* block involved of last error */ 1152 __u8 s_last_error_func[32]; /* function where the error happened */ 1153 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1154 __u8 s_mount_opts[64]; 1155 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1156 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1157 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1158 __le32 s_reserved[108]; /* Padding to the end of the block */ 1159 __le32 s_checksum; /* crc32c(superblock) */ 1160 }; 1161 1162 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1163 1164 #ifdef __KERNEL__ 1165 1166 /* 1167 * run-time mount flags 1168 */ 1169 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1170 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 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[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_lru; 1328 unsigned long s_es_last_sorted; 1329 struct percpu_counter s_extent_cache_cnt; 1330 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp; 1331 1332 /* Ratelimit ext4 messages. */ 1333 struct ratelimit_state s_err_ratelimit_state; 1334 struct ratelimit_state s_warning_ratelimit_state; 1335 struct ratelimit_state s_msg_ratelimit_state; 1336 }; 1337 1338 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1339 { 1340 return sb->s_fs_info; 1341 } 1342 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1343 { 1344 return container_of(inode, struct ext4_inode_info, vfs_inode); 1345 } 1346 1347 static inline struct timespec ext4_current_time(struct inode *inode) 1348 { 1349 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1350 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1351 } 1352 1353 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1354 { 1355 return ino == EXT4_ROOT_INO || 1356 ino == EXT4_USR_QUOTA_INO || 1357 ino == EXT4_GRP_QUOTA_INO || 1358 ino == EXT4_BOOT_LOADER_INO || 1359 ino == EXT4_JOURNAL_INO || 1360 ino == EXT4_RESIZE_INO || 1361 (ino >= EXT4_FIRST_INO(sb) && 1362 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1363 } 1364 1365 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1366 struct ext4_io_end *io_end) 1367 { 1368 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1369 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1370 atomic_inc(&EXT4_I(inode)->i_unwritten); 1371 } 1372 } 1373 1374 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1375 { 1376 return inode->i_private; 1377 } 1378 1379 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1380 { 1381 inode->i_private = io; 1382 } 1383 1384 /* 1385 * Inode dynamic state flags 1386 */ 1387 enum { 1388 EXT4_STATE_JDATA, /* journaled data exists */ 1389 EXT4_STATE_NEW, /* inode is newly created */ 1390 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1391 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1392 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1393 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1394 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1395 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1396 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 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 1504 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1505 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1506 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1507 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1508 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1509 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1510 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1511 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1512 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1513 /* 1514 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1515 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1516 * all other data structures' checksums. However, the METADATA_CSUM and 1517 * GDT_CSUM bits are mutually exclusive. 1518 */ 1519 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1520 1521 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1522 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1523 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1524 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1525 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1526 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1527 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1528 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1529 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1530 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1531 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1532 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1533 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1534 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1535 1536 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1537 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1538 EXT4_FEATURE_INCOMPAT_META_BG) 1539 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1540 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1541 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1542 1543 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1544 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1545 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1546 EXT4_FEATURE_INCOMPAT_META_BG) 1547 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1548 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1549 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1550 1551 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1552 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1553 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1554 EXT4_FEATURE_INCOMPAT_META_BG| \ 1555 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1556 EXT4_FEATURE_INCOMPAT_64BIT| \ 1557 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1558 EXT4_FEATURE_INCOMPAT_MMP | \ 1559 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1560 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1561 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1562 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1563 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1564 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1565 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1566 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1567 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1568 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1569 EXT4_FEATURE_RO_COMPAT_QUOTA) 1570 1571 /* 1572 * Default values for user and/or group using reserved blocks 1573 */ 1574 #define EXT4_DEF_RESUID 0 1575 #define EXT4_DEF_RESGID 0 1576 1577 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1578 1579 /* 1580 * Default mount options 1581 */ 1582 #define EXT4_DEFM_DEBUG 0x0001 1583 #define EXT4_DEFM_BSDGROUPS 0x0002 1584 #define EXT4_DEFM_XATTR_USER 0x0004 1585 #define EXT4_DEFM_ACL 0x0008 1586 #define EXT4_DEFM_UID16 0x0010 1587 #define EXT4_DEFM_JMODE 0x0060 1588 #define EXT4_DEFM_JMODE_DATA 0x0020 1589 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1590 #define EXT4_DEFM_JMODE_WBACK 0x0060 1591 #define EXT4_DEFM_NOBARRIER 0x0100 1592 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1593 #define EXT4_DEFM_DISCARD 0x0400 1594 #define EXT4_DEFM_NODELALLOC 0x0800 1595 1596 /* 1597 * Default journal batch times 1598 */ 1599 #define EXT4_DEF_MIN_BATCH_TIME 0 1600 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1601 1602 /* 1603 * Minimum number of groups in a flexgroup before we separate out 1604 * directories into the first block group of a flexgroup 1605 */ 1606 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1607 1608 /* 1609 * Structure of a directory entry 1610 */ 1611 #define EXT4_NAME_LEN 255 1612 1613 struct ext4_dir_entry { 1614 __le32 inode; /* Inode number */ 1615 __le16 rec_len; /* Directory entry length */ 1616 __le16 name_len; /* Name length */ 1617 char name[EXT4_NAME_LEN]; /* File name */ 1618 }; 1619 1620 /* 1621 * The new version of the directory entry. Since EXT4 structures are 1622 * stored in intel byte order, and the name_len field could never be 1623 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1624 * file_type field. 1625 */ 1626 struct ext4_dir_entry_2 { 1627 __le32 inode; /* Inode number */ 1628 __le16 rec_len; /* Directory entry length */ 1629 __u8 name_len; /* Name length */ 1630 __u8 file_type; 1631 char name[EXT4_NAME_LEN]; /* File name */ 1632 }; 1633 1634 /* 1635 * This is a bogus directory entry at the end of each leaf block that 1636 * records checksums. 1637 */ 1638 struct ext4_dir_entry_tail { 1639 __le32 det_reserved_zero1; /* Pretend to be unused */ 1640 __le16 det_rec_len; /* 12 */ 1641 __u8 det_reserved_zero2; /* Zero name length */ 1642 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1643 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1644 }; 1645 1646 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1647 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1648 ((blocksize) - \ 1649 sizeof(struct ext4_dir_entry_tail)))) 1650 1651 /* 1652 * Ext4 directory file types. Only the low 3 bits are used. The 1653 * other bits are reserved for now. 1654 */ 1655 #define EXT4_FT_UNKNOWN 0 1656 #define EXT4_FT_REG_FILE 1 1657 #define EXT4_FT_DIR 2 1658 #define EXT4_FT_CHRDEV 3 1659 #define EXT4_FT_BLKDEV 4 1660 #define EXT4_FT_FIFO 5 1661 #define EXT4_FT_SOCK 6 1662 #define EXT4_FT_SYMLINK 7 1663 1664 #define EXT4_FT_MAX 8 1665 1666 #define EXT4_FT_DIR_CSUM 0xDE 1667 1668 /* 1669 * EXT4_DIR_PAD defines the directory entries boundaries 1670 * 1671 * NOTE: It must be a multiple of 4 1672 */ 1673 #define EXT4_DIR_PAD 4 1674 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1675 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1676 ~EXT4_DIR_ROUND) 1677 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1678 1679 /* 1680 * If we ever get support for fs block sizes > page_size, we'll need 1681 * to remove the #if statements in the next two functions... 1682 */ 1683 static inline unsigned int 1684 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1685 { 1686 unsigned len = le16_to_cpu(dlen); 1687 1688 #if (PAGE_CACHE_SIZE >= 65536) 1689 if (len == EXT4_MAX_REC_LEN || len == 0) 1690 return blocksize; 1691 return (len & 65532) | ((len & 3) << 16); 1692 #else 1693 return len; 1694 #endif 1695 } 1696 1697 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1698 { 1699 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1700 BUG(); 1701 #if (PAGE_CACHE_SIZE >= 65536) 1702 if (len < 65536) 1703 return cpu_to_le16(len); 1704 if (len == blocksize) { 1705 if (blocksize == 65536) 1706 return cpu_to_le16(EXT4_MAX_REC_LEN); 1707 else 1708 return cpu_to_le16(0); 1709 } 1710 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1711 #else 1712 return cpu_to_le16(len); 1713 #endif 1714 } 1715 1716 /* 1717 * Hash Tree Directory indexing 1718 * (c) Daniel Phillips, 2001 1719 */ 1720 1721 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1722 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1723 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1724 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1725 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1726 1727 /* Legal values for the dx_root hash_version field: */ 1728 1729 #define DX_HASH_LEGACY 0 1730 #define DX_HASH_HALF_MD4 1 1731 #define DX_HASH_TEA 2 1732 #define DX_HASH_LEGACY_UNSIGNED 3 1733 #define DX_HASH_HALF_MD4_UNSIGNED 4 1734 #define DX_HASH_TEA_UNSIGNED 5 1735 1736 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1737 const void *address, unsigned int length) 1738 { 1739 struct { 1740 struct shash_desc shash; 1741 char ctx[4]; 1742 } desc; 1743 int err; 1744 1745 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1746 1747 desc.shash.tfm = sbi->s_chksum_driver; 1748 desc.shash.flags = 0; 1749 *(u32 *)desc.ctx = crc; 1750 1751 err = crypto_shash_update(&desc.shash, address, length); 1752 BUG_ON(err); 1753 1754 return *(u32 *)desc.ctx; 1755 } 1756 1757 #ifdef __KERNEL__ 1758 1759 /* hash info structure used by the directory hash */ 1760 struct dx_hash_info 1761 { 1762 u32 hash; 1763 u32 minor_hash; 1764 int hash_version; 1765 u32 *seed; 1766 }; 1767 1768 1769 /* 32 and 64 bit signed EOF for dx directories */ 1770 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1771 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1772 1773 1774 /* 1775 * Control parameters used by ext4_htree_next_block 1776 */ 1777 #define HASH_NB_ALWAYS 1 1778 1779 1780 /* 1781 * Describe an inode's exact location on disk and in memory 1782 */ 1783 struct ext4_iloc 1784 { 1785 struct buffer_head *bh; 1786 unsigned long offset; 1787 ext4_group_t block_group; 1788 }; 1789 1790 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1791 { 1792 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1793 } 1794 1795 /* 1796 * This structure is stuffed into the struct file's private_data field 1797 * for directories. It is where we put information so that we can do 1798 * readdir operations in hash tree order. 1799 */ 1800 struct dir_private_info { 1801 struct rb_root root; 1802 struct rb_node *curr_node; 1803 struct fname *extra_fname; 1804 loff_t last_pos; 1805 __u32 curr_hash; 1806 __u32 curr_minor_hash; 1807 __u32 next_hash; 1808 }; 1809 1810 /* calculate the first block number of the group */ 1811 static inline ext4_fsblk_t 1812 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1813 { 1814 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1815 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1816 } 1817 1818 /* 1819 * Special error return code only used by dx_probe() and its callers. 1820 */ 1821 #define ERR_BAD_DX_DIR -75000 1822 1823 /* 1824 * Timeout and state flag for lazy initialization inode thread. 1825 */ 1826 #define EXT4_DEF_LI_WAIT_MULT 10 1827 #define EXT4_DEF_LI_MAX_START_DELAY 5 1828 #define EXT4_LAZYINIT_QUIT 0x0001 1829 #define EXT4_LAZYINIT_RUNNING 0x0002 1830 1831 /* 1832 * Lazy inode table initialization info 1833 */ 1834 struct ext4_lazy_init { 1835 unsigned long li_state; 1836 struct list_head li_request_list; 1837 struct mutex li_list_mtx; 1838 }; 1839 1840 struct ext4_li_request { 1841 struct super_block *lr_super; 1842 struct ext4_sb_info *lr_sbi; 1843 ext4_group_t lr_next_group; 1844 struct list_head lr_request; 1845 unsigned long lr_next_sched; 1846 unsigned long lr_timeout; 1847 }; 1848 1849 struct ext4_features { 1850 struct kobject f_kobj; 1851 struct completion f_kobj_unregister; 1852 }; 1853 1854 /* 1855 * This structure will be used for multiple mount protection. It will be 1856 * written into the block number saved in the s_mmp_block field in the 1857 * superblock. Programs that check MMP should assume that if 1858 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1859 * to use the filesystem, regardless of how old the timestamp is. 1860 */ 1861 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1862 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1863 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1864 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1865 1866 struct mmp_struct { 1867 __le32 mmp_magic; /* Magic number for MMP */ 1868 __le32 mmp_seq; /* Sequence no. updated periodically */ 1869 1870 /* 1871 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1872 * purposes and do not affect the correctness of the algorithm 1873 */ 1874 __le64 mmp_time; /* Time last updated */ 1875 char mmp_nodename[64]; /* Node which last updated MMP block */ 1876 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1877 1878 /* 1879 * mmp_check_interval is used to verify if the MMP block has been 1880 * updated on the block device. The value is updated based on the 1881 * maximum time to write the MMP block during an update cycle. 1882 */ 1883 __le16 mmp_check_interval; 1884 1885 __le16 mmp_pad1; 1886 __le32 mmp_pad2[226]; 1887 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1888 }; 1889 1890 /* arguments passed to the mmp thread */ 1891 struct mmpd_data { 1892 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1893 struct super_block *sb; /* super block of the fs */ 1894 }; 1895 1896 /* 1897 * Check interval multiplier 1898 * The MMP block is written every update interval and initially checked every 1899 * update interval x the multiplier (the value is then adapted based on the 1900 * write latency). The reason is that writes can be delayed under load and we 1901 * don't want readers to incorrectly assume that the filesystem is no longer 1902 * in use. 1903 */ 1904 #define EXT4_MMP_CHECK_MULT 2UL 1905 1906 /* 1907 * Minimum interval for MMP checking in seconds. 1908 */ 1909 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1910 1911 /* 1912 * Maximum interval for MMP checking in seconds. 1913 */ 1914 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1915 1916 /* 1917 * Function prototypes 1918 */ 1919 1920 /* 1921 * Ok, these declarations are also in <linux/kernel.h> but none of the 1922 * ext4 source programs needs to include it so they are duplicated here. 1923 */ 1924 # define NORET_TYPE /**/ 1925 # define ATTRIB_NORET __attribute__((noreturn)) 1926 # define NORET_AND noreturn, 1927 1928 /* bitmap.c */ 1929 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1930 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1931 struct ext4_group_desc *gdp, 1932 struct buffer_head *bh, int sz); 1933 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1934 struct ext4_group_desc *gdp, 1935 struct buffer_head *bh, int sz); 1936 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1937 struct ext4_group_desc *gdp, 1938 struct buffer_head *bh); 1939 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1940 struct ext4_group_desc *gdp, 1941 struct buffer_head *bh); 1942 1943 /* balloc.c */ 1944 extern void ext4_get_group_no_and_offset(struct super_block *sb, 1945 ext4_fsblk_t blocknr, 1946 ext4_group_t *blockgrpp, 1947 ext4_grpblk_t *offsetp); 1948 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 1949 ext4_fsblk_t block); 1950 1951 extern void ext4_validate_block_bitmap(struct super_block *sb, 1952 struct ext4_group_desc *desc, 1953 ext4_group_t block_group, 1954 struct buffer_head *bh); 1955 extern unsigned int ext4_block_group(struct super_block *sb, 1956 ext4_fsblk_t blocknr); 1957 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1958 ext4_fsblk_t blocknr); 1959 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1960 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1961 ext4_group_t group); 1962 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1963 ext4_fsblk_t goal, 1964 unsigned int flags, 1965 unsigned long *count, 1966 int *errp); 1967 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1968 s64 nclusters, unsigned int flags); 1969 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1970 extern void ext4_check_blocks_bitmap(struct super_block *); 1971 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1972 ext4_group_t block_group, 1973 struct buffer_head ** bh); 1974 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1975 1976 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1977 ext4_group_t block_group); 1978 extern int ext4_wait_block_bitmap(struct super_block *sb, 1979 ext4_group_t block_group, 1980 struct buffer_head *bh); 1981 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1982 ext4_group_t block_group); 1983 extern void ext4_init_block_bitmap(struct super_block *sb, 1984 struct buffer_head *bh, 1985 ext4_group_t group, 1986 struct ext4_group_desc *desc); 1987 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1988 ext4_group_t block_group, 1989 struct ext4_group_desc *gdp); 1990 extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1991 ext4_group_t block_group, 1992 struct ext4_group_desc *gdp); 1993 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1994 1995 /* dir.c */ 1996 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1997 struct file *, 1998 struct ext4_dir_entry_2 *, 1999 struct buffer_head *, char *, int, 2000 unsigned int); 2001 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2002 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2003 (de), (bh), (buf), (size), (offset))) 2004 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2005 __u32 minor_hash, 2006 struct ext4_dir_entry_2 *dirent); 2007 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2008 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2009 struct buffer_head *bh, 2010 void *buf, int buf_size, 2011 const char *name, int namelen, 2012 struct ext4_dir_entry_2 **dest_de); 2013 void ext4_insert_dentry(struct inode *inode, 2014 struct ext4_dir_entry_2 *de, 2015 int buf_size, 2016 const char *name, int namelen); 2017 static inline void ext4_update_dx_flag(struct inode *inode) 2018 { 2019 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 2020 EXT4_FEATURE_COMPAT_DIR_INDEX)) 2021 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2022 } 2023 static unsigned char ext4_filetype_table[] = { 2024 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2025 }; 2026 2027 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2028 { 2029 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 2030 (filetype >= EXT4_FT_MAX)) 2031 return DT_UNKNOWN; 2032 2033 return ext4_filetype_table[filetype]; 2034 } 2035 2036 /* fsync.c */ 2037 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2038 2039 /* hash.c */ 2040 extern int ext4fs_dirhash(const char *name, int len, struct 2041 dx_hash_info *hinfo); 2042 2043 /* ialloc.c */ 2044 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2045 const struct qstr *qstr, __u32 goal, 2046 uid_t *owner, int handle_type, 2047 unsigned int line_no, int nblocks); 2048 2049 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2050 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2051 0, 0, 0) 2052 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2053 type, nblocks) \ 2054 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2055 (type), __LINE__, (nblocks)) 2056 2057 2058 extern void ext4_free_inode(handle_t *, struct inode *); 2059 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2060 extern unsigned long ext4_count_free_inodes(struct super_block *); 2061 extern unsigned long ext4_count_dirs(struct super_block *); 2062 extern void ext4_check_inodes_bitmap(struct super_block *); 2063 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2064 extern int ext4_init_inode_table(struct super_block *sb, 2065 ext4_group_t group, int barrier); 2066 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2067 2068 /* mballoc.c */ 2069 extern long ext4_mb_stats; 2070 extern long ext4_mb_max_to_scan; 2071 extern int ext4_mb_init(struct super_block *); 2072 extern int ext4_mb_release(struct super_block *); 2073 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2074 struct ext4_allocation_request *, int *); 2075 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2076 extern void ext4_discard_preallocations(struct inode *); 2077 extern int __init ext4_init_mballoc(void); 2078 extern void ext4_exit_mballoc(void); 2079 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2080 struct buffer_head *bh, ext4_fsblk_t block, 2081 unsigned long count, int flags); 2082 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2083 ext4_group_t ngroups); 2084 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2085 ext4_group_t i, struct ext4_group_desc *desc); 2086 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2087 ext4_fsblk_t block, unsigned long count); 2088 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2089 2090 /* inode.c */ 2091 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 2092 ext4_lblk_t, int, int *); 2093 struct buffer_head *ext4_bread(handle_t *, struct inode *, 2094 ext4_lblk_t, int, int *); 2095 int ext4_get_block_write(struct inode *inode, sector_t iblock, 2096 struct buffer_head *bh_result, int create); 2097 int ext4_get_block(struct inode *inode, sector_t iblock, 2098 struct buffer_head *bh_result, int create); 2099 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2100 struct buffer_head *bh, int create); 2101 int ext4_walk_page_buffers(handle_t *handle, 2102 struct buffer_head *head, 2103 unsigned from, 2104 unsigned to, 2105 int *partial, 2106 int (*fn)(handle_t *handle, 2107 struct buffer_head *bh)); 2108 int do_journal_get_write_access(handle_t *handle, 2109 struct buffer_head *bh); 2110 #define FALL_BACK_TO_NONDELALLOC 1 2111 #define CONVERT_INLINE_DATA 2 2112 2113 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2114 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2115 extern int ext4_setattr(struct dentry *, struct iattr *); 2116 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2117 struct kstat *stat); 2118 extern void ext4_evict_inode(struct inode *); 2119 extern void ext4_clear_inode(struct inode *); 2120 extern int ext4_sync_inode(handle_t *, struct inode *); 2121 extern void ext4_dirty_inode(struct inode *, int); 2122 extern int ext4_change_inode_journal_flag(struct inode *, int); 2123 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2124 extern int ext4_inode_attach_jinode(struct inode *inode); 2125 extern int ext4_can_truncate(struct inode *inode); 2126 extern void ext4_truncate(struct inode *); 2127 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2128 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2129 extern void ext4_set_inode_flags(struct inode *); 2130 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2131 extern int ext4_alloc_da_blocks(struct inode *inode); 2132 extern void ext4_set_aops(struct inode *inode); 2133 extern int ext4_writepage_trans_blocks(struct inode *); 2134 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2135 extern int ext4_block_truncate_page(handle_t *handle, 2136 struct address_space *mapping, loff_t from); 2137 extern int ext4_block_zero_page_range(handle_t *handle, 2138 struct address_space *mapping, loff_t from, loff_t length); 2139 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2140 loff_t lstart, loff_t lend); 2141 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2142 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2143 extern void ext4_da_update_reserve_space(struct inode *inode, 2144 int used, int quota_claim); 2145 2146 /* indirect.c */ 2147 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2148 struct ext4_map_blocks *map, int flags); 2149 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2150 const struct iovec *iov, loff_t offset, 2151 unsigned long nr_segs); 2152 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2153 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2154 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2155 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode, 2156 ext4_lblk_t first, ext4_lblk_t stop); 2157 2158 /* ioctl.c */ 2159 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2160 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2161 2162 /* migrate.c */ 2163 extern int ext4_ext_migrate(struct inode *); 2164 extern int ext4_ind_migrate(struct inode *inode); 2165 2166 /* namei.c */ 2167 extern int ext4_dirent_csum_verify(struct inode *inode, 2168 struct ext4_dir_entry *dirent); 2169 extern int ext4_orphan_add(handle_t *, struct inode *); 2170 extern int ext4_orphan_del(handle_t *, struct inode *); 2171 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2172 __u32 start_minor_hash, __u32 *next_hash); 2173 extern int search_dir(struct buffer_head *bh, 2174 char *search_buf, 2175 int buf_size, 2176 struct inode *dir, 2177 const struct qstr *d_name, 2178 unsigned int offset, 2179 struct ext4_dir_entry_2 **res_dir); 2180 extern int ext4_generic_delete_entry(handle_t *handle, 2181 struct inode *dir, 2182 struct ext4_dir_entry_2 *de_del, 2183 struct buffer_head *bh, 2184 void *entry_buf, 2185 int buf_size, 2186 int csum_size); 2187 2188 /* resize.c */ 2189 extern int ext4_group_add(struct super_block *sb, 2190 struct ext4_new_group_data *input); 2191 extern int ext4_group_extend(struct super_block *sb, 2192 struct ext4_super_block *es, 2193 ext4_fsblk_t n_blocks_count); 2194 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2195 2196 /* super.c */ 2197 extern int ext4_calculate_overhead(struct super_block *sb); 2198 extern int ext4_superblock_csum_verify(struct super_block *sb, 2199 struct ext4_super_block *es); 2200 extern void ext4_superblock_csum_set(struct super_block *sb); 2201 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2202 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2203 extern void ext4_kvfree(void *ptr); 2204 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2205 ext4_group_t ngroup); 2206 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2207 char nbuf[16]); 2208 2209 extern __printf(4, 5) 2210 void __ext4_error(struct super_block *, const char *, unsigned int, 2211 const char *, ...); 2212 extern __printf(5, 6) 2213 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2214 const char *, ...); 2215 extern __printf(5, 6) 2216 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2217 const char *, ...); 2218 extern void __ext4_std_error(struct super_block *, const char *, 2219 unsigned int, int); 2220 extern __printf(4, 5) 2221 void __ext4_abort(struct super_block *, const char *, unsigned int, 2222 const char *, ...); 2223 extern __printf(4, 5) 2224 void __ext4_warning(struct super_block *, const char *, unsigned int, 2225 const char *, ...); 2226 extern __printf(3, 4) 2227 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2228 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2229 const char *, unsigned int, const char *); 2230 extern __printf(7, 8) 2231 void __ext4_grp_locked_error(const char *, unsigned int, 2232 struct super_block *, ext4_group_t, 2233 unsigned long, ext4_fsblk_t, 2234 const char *, ...); 2235 2236 #ifdef CONFIG_PRINTK 2237 2238 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2239 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2240 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2241 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2242 #define ext4_error(sb, fmt, ...) \ 2243 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2244 #define ext4_abort(sb, fmt, ...) \ 2245 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2246 #define ext4_warning(sb, fmt, ...) \ 2247 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2248 #define ext4_msg(sb, level, fmt, ...) \ 2249 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2250 #define dump_mmp_msg(sb, mmp, msg) \ 2251 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2252 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2253 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2254 fmt, ##__VA_ARGS__) 2255 2256 #else 2257 2258 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2259 do { \ 2260 no_printk(fmt, ##__VA_ARGS__); \ 2261 __ext4_error_inode(inode, "", 0, block, " "); \ 2262 } while (0) 2263 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2264 do { \ 2265 no_printk(fmt, ##__VA_ARGS__); \ 2266 __ext4_error_file(file, "", 0, block, " "); \ 2267 } while (0) 2268 #define ext4_error(sb, fmt, ...) \ 2269 do { \ 2270 no_printk(fmt, ##__VA_ARGS__); \ 2271 __ext4_error(sb, "", 0, " "); \ 2272 } while (0) 2273 #define ext4_abort(sb, fmt, ...) \ 2274 do { \ 2275 no_printk(fmt, ##__VA_ARGS__); \ 2276 __ext4_abort(sb, "", 0, " "); \ 2277 } while (0) 2278 #define ext4_warning(sb, fmt, ...) \ 2279 do { \ 2280 no_printk(fmt, ##__VA_ARGS__); \ 2281 __ext4_warning(sb, "", 0, " "); \ 2282 } while (0) 2283 #define ext4_msg(sb, level, fmt, ...) \ 2284 do { \ 2285 no_printk(fmt, ##__VA_ARGS__); \ 2286 __ext4_msg(sb, "", " "); \ 2287 } while (0) 2288 #define dump_mmp_msg(sb, mmp, msg) \ 2289 __dump_mmp_msg(sb, mmp, "", 0, "") 2290 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2291 do { \ 2292 no_printk(fmt, ##__VA_ARGS__); \ 2293 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2294 } while (0) 2295 2296 #endif 2297 2298 extern void ext4_update_dynamic_rev(struct super_block *sb); 2299 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2300 __u32 compat); 2301 extern int ext4_update_rocompat_feature(handle_t *handle, 2302 struct super_block *sb, __u32 rocompat); 2303 extern int ext4_update_incompat_feature(handle_t *handle, 2304 struct super_block *sb, __u32 incompat); 2305 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2306 struct ext4_group_desc *bg); 2307 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2308 struct ext4_group_desc *bg); 2309 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2310 struct ext4_group_desc *bg); 2311 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2312 struct ext4_group_desc *bg); 2313 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2314 struct ext4_group_desc *bg); 2315 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2316 struct ext4_group_desc *bg); 2317 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2318 struct ext4_group_desc *bg); 2319 extern void ext4_block_bitmap_set(struct super_block *sb, 2320 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2321 extern void ext4_inode_bitmap_set(struct super_block *sb, 2322 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2323 extern void ext4_inode_table_set(struct super_block *sb, 2324 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2325 extern void ext4_free_group_clusters_set(struct super_block *sb, 2326 struct ext4_group_desc *bg, 2327 __u32 count); 2328 extern void ext4_free_inodes_set(struct super_block *sb, 2329 struct ext4_group_desc *bg, __u32 count); 2330 extern void ext4_used_dirs_set(struct super_block *sb, 2331 struct ext4_group_desc *bg, __u32 count); 2332 extern void ext4_itable_unused_set(struct super_block *sb, 2333 struct ext4_group_desc *bg, __u32 count); 2334 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2335 struct ext4_group_desc *gdp); 2336 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2337 struct ext4_group_desc *gdp); 2338 extern int ext4_register_li_request(struct super_block *sb, 2339 ext4_group_t first_not_zeroed); 2340 2341 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2342 { 2343 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2344 EXT4_FEATURE_RO_COMPAT_GDT_CSUM | 2345 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM); 2346 } 2347 2348 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2349 { 2350 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2351 le32_to_cpu(es->s_blocks_count_lo); 2352 } 2353 2354 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2355 { 2356 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2357 le32_to_cpu(es->s_r_blocks_count_lo); 2358 } 2359 2360 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2361 { 2362 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2363 le32_to_cpu(es->s_free_blocks_count_lo); 2364 } 2365 2366 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2367 ext4_fsblk_t blk) 2368 { 2369 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2370 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2371 } 2372 2373 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2374 ext4_fsblk_t blk) 2375 { 2376 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2377 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2378 } 2379 2380 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2381 ext4_fsblk_t blk) 2382 { 2383 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2384 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2385 } 2386 2387 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2388 { 2389 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2390 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2391 le32_to_cpu(raw_inode->i_size_lo); 2392 else 2393 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2394 } 2395 2396 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2397 { 2398 raw_inode->i_size_lo = cpu_to_le32(i_size); 2399 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2400 } 2401 2402 static inline 2403 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2404 ext4_group_t group) 2405 { 2406 struct ext4_group_info ***grp_info; 2407 long indexv, indexh; 2408 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2409 grp_info = EXT4_SB(sb)->s_group_info; 2410 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2411 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2412 return grp_info[indexv][indexh]; 2413 } 2414 2415 /* 2416 * Reading s_groups_count requires using smp_rmb() afterwards. See 2417 * the locking protocol documented in the comments of ext4_group_add() 2418 * in resize.c 2419 */ 2420 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2421 { 2422 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2423 2424 smp_rmb(); 2425 return ngroups; 2426 } 2427 2428 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2429 ext4_group_t block_group) 2430 { 2431 return block_group >> sbi->s_log_groups_per_flex; 2432 } 2433 2434 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2435 { 2436 return 1 << sbi->s_log_groups_per_flex; 2437 } 2438 2439 #define ext4_std_error(sb, errno) \ 2440 do { \ 2441 if ((errno)) \ 2442 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2443 } while (0) 2444 2445 #ifdef CONFIG_SMP 2446 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2447 * counters. So we need to make sure we have free clusters more 2448 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2449 */ 2450 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2451 #else 2452 #define EXT4_FREECLUSTERS_WATERMARK 0 2453 #endif 2454 2455 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2456 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2457 { 2458 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2459 !mutex_is_locked(&inode->i_mutex)); 2460 down_write(&EXT4_I(inode)->i_data_sem); 2461 if (newsize > EXT4_I(inode)->i_disksize) 2462 EXT4_I(inode)->i_disksize = newsize; 2463 up_write(&EXT4_I(inode)->i_data_sem); 2464 } 2465 2466 /* 2467 * Update i_disksize after writeback has been started. Races with truncate 2468 * are avoided by checking i_size under i_data_sem. 2469 */ 2470 static inline void ext4_wb_update_i_disksize(struct inode *inode, loff_t newsize) 2471 { 2472 loff_t i_size; 2473 2474 down_write(&EXT4_I(inode)->i_data_sem); 2475 i_size = i_size_read(inode); 2476 if (newsize > i_size) 2477 newsize = i_size; 2478 if (newsize > EXT4_I(inode)->i_disksize) 2479 EXT4_I(inode)->i_disksize = newsize; 2480 up_write(&EXT4_I(inode)->i_data_sem); 2481 } 2482 2483 struct ext4_group_info { 2484 unsigned long bb_state; 2485 struct rb_root bb_free_root; 2486 ext4_grpblk_t bb_first_free; /* first free block */ 2487 ext4_grpblk_t bb_free; /* total free blocks */ 2488 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2489 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2490 struct list_head bb_prealloc_list; 2491 #ifdef DOUBLE_CHECK 2492 void *bb_bitmap; 2493 #endif 2494 struct rw_semaphore alloc_sem; 2495 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2496 * regions, index is order. 2497 * bb_counters[3] = 5 means 2498 * 5 free 8-block regions. */ 2499 }; 2500 2501 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2502 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2503 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2504 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2505 2506 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2507 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2508 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2509 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2510 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2511 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2512 2513 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2514 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2515 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2516 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2517 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2518 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2519 2520 #define EXT4_MAX_CONTENTION 8 2521 #define EXT4_CONTENTION_THRESHOLD 2 2522 2523 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2524 ext4_group_t group) 2525 { 2526 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2527 } 2528 2529 /* 2530 * Returns true if the filesystem is busy enough that attempts to 2531 * access the block group locks has run into contention. 2532 */ 2533 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2534 { 2535 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2536 } 2537 2538 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2539 { 2540 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2541 if (spin_trylock(lock)) 2542 /* 2543 * We're able to grab the lock right away, so drop the 2544 * lock contention counter. 2545 */ 2546 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2547 else { 2548 /* 2549 * The lock is busy, so bump the contention counter, 2550 * and then wait on the spin lock. 2551 */ 2552 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2553 EXT4_MAX_CONTENTION); 2554 spin_lock(lock); 2555 } 2556 } 2557 2558 static inline void ext4_unlock_group(struct super_block *sb, 2559 ext4_group_t group) 2560 { 2561 spin_unlock(ext4_group_lock_ptr(sb, group)); 2562 } 2563 2564 /* 2565 * Block validity checking 2566 */ 2567 #define ext4_check_indirect_blockref(inode, bh) \ 2568 ext4_check_blockref(__func__, __LINE__, inode, \ 2569 (__le32 *)(bh)->b_data, \ 2570 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2571 2572 #define ext4_ind_check_inode(inode) \ 2573 ext4_check_blockref(__func__, __LINE__, inode, \ 2574 EXT4_I(inode)->i_data, \ 2575 EXT4_NDIR_BLOCKS) 2576 2577 /* 2578 * Inodes and files operations 2579 */ 2580 2581 /* dir.c */ 2582 extern const struct file_operations ext4_dir_operations; 2583 2584 /* file.c */ 2585 extern const struct inode_operations ext4_file_inode_operations; 2586 extern const struct file_operations ext4_file_operations; 2587 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2588 extern void ext4_unwritten_wait(struct inode *inode); 2589 2590 /* inline.c */ 2591 extern int ext4_has_inline_data(struct inode *inode); 2592 extern int ext4_get_inline_size(struct inode *inode); 2593 extern int ext4_get_max_inline_size(struct inode *inode); 2594 extern int ext4_find_inline_data_nolock(struct inode *inode); 2595 extern void ext4_write_inline_data(struct inode *inode, 2596 struct ext4_iloc *iloc, 2597 void *buffer, loff_t pos, 2598 unsigned int len); 2599 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode, 2600 unsigned int len); 2601 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2602 unsigned int len); 2603 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2604 2605 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2606 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2607 struct inode *inode, 2608 loff_t pos, unsigned len, 2609 unsigned flags, 2610 struct page **pagep); 2611 extern int ext4_write_inline_data_end(struct inode *inode, 2612 loff_t pos, unsigned len, 2613 unsigned copied, 2614 struct page *page); 2615 extern struct buffer_head * 2616 ext4_journalled_write_inline_data(struct inode *inode, 2617 unsigned len, 2618 struct page *page); 2619 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2620 struct inode *inode, 2621 loff_t pos, unsigned len, 2622 unsigned flags, 2623 struct page **pagep, 2624 void **fsdata); 2625 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2626 unsigned len, unsigned copied, 2627 struct page *page); 2628 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry, 2629 struct inode *inode); 2630 extern int ext4_try_create_inline_dir(handle_t *handle, 2631 struct inode *parent, 2632 struct inode *inode); 2633 extern int ext4_read_inline_dir(struct file *filp, 2634 struct dir_context *ctx, 2635 int *has_inline_data); 2636 extern int htree_inlinedir_to_tree(struct file *dir_file, 2637 struct inode *dir, ext4_lblk_t block, 2638 struct dx_hash_info *hinfo, 2639 __u32 start_hash, __u32 start_minor_hash, 2640 int *has_inline_data); 2641 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2642 const struct qstr *d_name, 2643 struct ext4_dir_entry_2 **res_dir, 2644 int *has_inline_data); 2645 extern int ext4_delete_inline_entry(handle_t *handle, 2646 struct inode *dir, 2647 struct ext4_dir_entry_2 *de_del, 2648 struct buffer_head *bh, 2649 int *has_inline_data); 2650 extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 2651 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 2652 struct ext4_dir_entry_2 **parent_de, 2653 int *retval); 2654 extern int ext4_inline_data_fiemap(struct inode *inode, 2655 struct fiemap_extent_info *fieinfo, 2656 int *has_inline); 2657 extern int ext4_try_to_evict_inline_data(handle_t *handle, 2658 struct inode *inode, 2659 int needed); 2660 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 2661 2662 extern int ext4_convert_inline_data(struct inode *inode); 2663 2664 /* namei.c */ 2665 extern const struct inode_operations ext4_dir_inode_operations; 2666 extern const struct inode_operations ext4_special_inode_operations; 2667 extern struct dentry *ext4_get_parent(struct dentry *child); 2668 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2669 struct ext4_dir_entry_2 *de, 2670 int blocksize, int csum_size, 2671 unsigned int parent_ino, int dotdot_real_len); 2672 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2673 unsigned int blocksize); 2674 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2675 struct inode *inode, 2676 struct buffer_head *bh); 2677 #define S_SHIFT 12 2678 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 2679 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 2680 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 2681 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 2682 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 2683 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 2684 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 2685 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 2686 }; 2687 2688 static inline void ext4_set_de_type(struct super_block *sb, 2689 struct ext4_dir_entry_2 *de, 2690 umode_t mode) { 2691 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE)) 2692 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 2693 } 2694 2695 2696 /* symlink.c */ 2697 extern const struct inode_operations ext4_symlink_inode_operations; 2698 extern const struct inode_operations ext4_fast_symlink_inode_operations; 2699 2700 /* block_validity */ 2701 extern void ext4_release_system_zone(struct super_block *sb); 2702 extern int ext4_setup_system_zone(struct super_block *sb); 2703 extern int __init ext4_init_system_zone(void); 2704 extern void ext4_exit_system_zone(void); 2705 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2706 ext4_fsblk_t start_blk, 2707 unsigned int count); 2708 extern int ext4_check_blockref(const char *, unsigned int, 2709 struct inode *, __le32 *, unsigned int); 2710 2711 /* extents.c */ 2712 struct ext4_ext_path; 2713 struct ext4_extent; 2714 2715 /* 2716 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 2717 * __le32. 2718 */ 2719 #define EXT_MAX_BLOCKS 0xffffffff 2720 2721 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2722 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2723 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 2724 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2725 struct ext4_map_blocks *map, int flags); 2726 extern void ext4_ext_truncate(handle_t *, struct inode *); 2727 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 2728 ext4_lblk_t end); 2729 extern void ext4_ext_init(struct super_block *); 2730 extern void ext4_ext_release(struct super_block *); 2731 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2732 loff_t len); 2733 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 2734 loff_t offset, ssize_t len); 2735 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2736 struct ext4_map_blocks *map, int flags); 2737 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2738 ext4_lblk_t lblocks); 2739 extern int ext4_extent_tree_init(handle_t *, struct inode *); 2740 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2741 int num, 2742 struct ext4_ext_path *path); 2743 extern int ext4_can_extents_be_merged(struct inode *inode, 2744 struct ext4_extent *ex1, 2745 struct ext4_extent *ex2); 2746 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2747 struct ext4_ext_path *, 2748 struct ext4_extent *, int); 2749 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t, 2750 struct ext4_ext_path *, 2751 int flags); 2752 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2753 extern int ext4_ext_check_inode(struct inode *inode); 2754 extern int ext4_find_delalloc_range(struct inode *inode, 2755 ext4_lblk_t lblk_start, 2756 ext4_lblk_t lblk_end); 2757 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2758 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2759 __u64 start, __u64 len); 2760 extern int ext4_ext_precache(struct inode *inode); 2761 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 2762 2763 /* move_extent.c */ 2764 extern void ext4_double_down_write_data_sem(struct inode *first, 2765 struct inode *second); 2766 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 2767 struct inode *donor_inode); 2768 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2769 __u64 start_orig, __u64 start_donor, 2770 __u64 len, __u64 *moved_len); 2771 extern int mext_next_extent(struct inode *inode, struct ext4_ext_path *path, 2772 struct ext4_extent **extent); 2773 2774 /* page-io.c */ 2775 extern int __init ext4_init_pageio(void); 2776 extern void ext4_exit_pageio(void); 2777 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2778 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 2779 extern int ext4_put_io_end(ext4_io_end_t *io_end); 2780 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 2781 extern void ext4_io_submit_init(struct ext4_io_submit *io, 2782 struct writeback_control *wbc); 2783 extern void ext4_end_io_rsv_work(struct work_struct *work); 2784 extern void ext4_io_submit(struct ext4_io_submit *io); 2785 extern int ext4_bio_write_page(struct ext4_io_submit *io, 2786 struct page *page, 2787 int len, 2788 struct writeback_control *wbc); 2789 2790 /* mmp.c */ 2791 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2792 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp); 2793 extern int ext4_mmp_csum_verify(struct super_block *sb, 2794 struct mmp_struct *mmp); 2795 2796 /* 2797 * Note that these flags will never ever appear in a buffer_head's state flag. 2798 * See EXT4_MAP_... to see where this is used. 2799 */ 2800 enum ext4_state_bits { 2801 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2802 = BH_JBDPrivateStart, 2803 BH_AllocFromCluster, /* allocated blocks were part of already 2804 * allocated cluster. */ 2805 }; 2806 2807 /* 2808 * Add new method to test whether block and inode bitmaps are properly 2809 * initialized. With uninit_bg reading the block from disk is not enough 2810 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2811 */ 2812 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2813 2814 static inline int bitmap_uptodate(struct buffer_head *bh) 2815 { 2816 return (buffer_uptodate(bh) && 2817 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2818 } 2819 static inline void set_bitmap_uptodate(struct buffer_head *bh) 2820 { 2821 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2822 } 2823 2824 /* 2825 * Disable DIO read nolock optimization, so new dioreaders will be forced 2826 * to grab i_mutex 2827 */ 2828 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2829 { 2830 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2831 smp_mb(); 2832 } 2833 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2834 { 2835 smp_mb(); 2836 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2837 } 2838 2839 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2840 2841 /* For ioend & aio unwritten conversion wait queues */ 2842 #define EXT4_WQ_HASH_SZ 37 2843 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2844 EXT4_WQ_HASH_SZ]) 2845 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2846 EXT4_WQ_HASH_SZ]) 2847 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2848 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2849 2850 #define EXT4_RESIZING 0 2851 extern int ext4_resize_begin(struct super_block *sb); 2852 extern void ext4_resize_end(struct super_block *sb); 2853 2854 #endif /* __KERNEL__ */ 2855 2856 #endif /* _EXT4_H */ 2857