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/version.h> 30 #include <linux/wait.h> 31 #include <linux/blockgroup_lock.h> 32 #include <linux/percpu_counter.h> 33 #include <linux/ratelimit.h> 34 #include <crypto/hash.h> 35 #include <linux/fscrypto.h> 36 #include <linux/falloc.h> 37 #include <linux/percpu-rwsem.h> 38 #ifdef __KERNEL__ 39 #include <linux/compat.h> 40 #endif 41 42 /* 43 * The fourth extended filesystem constants/structures 44 */ 45 46 /* 47 * with AGGRESSIVE_CHECK allocator runs consistency checks over 48 * structures. these checks slow things down a lot 49 */ 50 #define AGGRESSIVE_CHECK__ 51 52 /* 53 * with DOUBLE_CHECK defined mballoc creates persistent in-core 54 * bitmaps, maintains and uses them to check for double allocations 55 */ 56 #define DOUBLE_CHECK__ 57 58 /* 59 * Define EXT4FS_DEBUG to produce debug messages 60 */ 61 #undef EXT4FS_DEBUG 62 63 /* 64 * Debug code 65 */ 66 #ifdef EXT4FS_DEBUG 67 #define ext4_debug(f, a...) \ 68 do { \ 69 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 70 __FILE__, __LINE__, __func__); \ 71 printk(KERN_DEBUG f, ## a); \ 72 } while (0) 73 #else 74 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 75 #endif 76 77 /* 78 * Turn on EXT_DEBUG to get lots of info about extents operations. 79 */ 80 #define EXT_DEBUG__ 81 #ifdef EXT_DEBUG 82 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 83 #else 84 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 85 #endif 86 87 /* data type for block offset of block group */ 88 typedef int ext4_grpblk_t; 89 90 /* data type for filesystem-wide blocks number */ 91 typedef unsigned long long ext4_fsblk_t; 92 93 /* data type for file logical block number */ 94 typedef __u32 ext4_lblk_t; 95 96 /* data type for block group number */ 97 typedef unsigned int ext4_group_t; 98 99 enum SHIFT_DIRECTION { 100 SHIFT_LEFT = 0, 101 SHIFT_RIGHT, 102 }; 103 104 /* 105 * Flags used in mballoc's allocation_context flags field. 106 * 107 * Also used to show what's going on for debugging purposes when the 108 * flag field is exported via the traceport interface 109 */ 110 111 /* prefer goal again. length */ 112 #define EXT4_MB_HINT_MERGE 0x0001 113 /* blocks already reserved */ 114 #define EXT4_MB_HINT_RESERVED 0x0002 115 /* metadata is being allocated */ 116 #define EXT4_MB_HINT_METADATA 0x0004 117 /* first blocks in the file */ 118 #define EXT4_MB_HINT_FIRST 0x0008 119 /* search for the best chunk */ 120 #define EXT4_MB_HINT_BEST 0x0010 121 /* data is being allocated */ 122 #define EXT4_MB_HINT_DATA 0x0020 123 /* don't preallocate (for tails) */ 124 #define EXT4_MB_HINT_NOPREALLOC 0x0040 125 /* allocate for locality group */ 126 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 127 /* allocate goal blocks or none */ 128 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 129 /* goal is meaningful */ 130 #define EXT4_MB_HINT_TRY_GOAL 0x0200 131 /* blocks already pre-reserved by delayed allocation */ 132 #define EXT4_MB_DELALLOC_RESERVED 0x0400 133 /* We are doing stream allocation */ 134 #define EXT4_MB_STREAM_ALLOC 0x0800 135 /* Use reserved root blocks if needed */ 136 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 137 /* Use blocks from reserved pool */ 138 #define EXT4_MB_USE_RESERVED 0x2000 139 140 struct ext4_allocation_request { 141 /* target inode for block we're allocating */ 142 struct inode *inode; 143 /* how many blocks we want to allocate */ 144 unsigned int len; 145 /* logical block in target inode */ 146 ext4_lblk_t logical; 147 /* the closest logical allocated block to the left */ 148 ext4_lblk_t lleft; 149 /* the closest logical allocated block to the right */ 150 ext4_lblk_t lright; 151 /* phys. target (a hint) */ 152 ext4_fsblk_t goal; 153 /* phys. block for the closest logical allocated block to the left */ 154 ext4_fsblk_t pleft; 155 /* phys. block for the closest logical allocated block to the right */ 156 ext4_fsblk_t pright; 157 /* flags. see above EXT4_MB_HINT_* */ 158 unsigned int flags; 159 }; 160 161 /* 162 * Logical to physical block mapping, used by ext4_map_blocks() 163 * 164 * This structure is used to pass requests into ext4_map_blocks() as 165 * well as to store the information returned by ext4_map_blocks(). It 166 * takes less room on the stack than a struct buffer_head. 167 */ 168 #define EXT4_MAP_NEW (1 << BH_New) 169 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 170 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 171 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 172 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 173 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 174 175 struct ext4_map_blocks { 176 ext4_fsblk_t m_pblk; 177 ext4_lblk_t m_lblk; 178 unsigned int m_len; 179 unsigned int m_flags; 180 }; 181 182 /* 183 * Flags for ext4_io_end->flags 184 */ 185 #define EXT4_IO_END_UNWRITTEN 0x0001 186 187 /* 188 * For converting unwritten extents on a work queue. 'handle' is used for 189 * buffered writeback. 190 */ 191 typedef struct ext4_io_end { 192 struct list_head list; /* per-file finished IO list */ 193 handle_t *handle; /* handle reserved for extent 194 * conversion */ 195 struct inode *inode; /* file being written to */ 196 struct bio *bio; /* Linked list of completed 197 * bios covering the extent */ 198 unsigned int flag; /* unwritten or not */ 199 atomic_t count; /* reference counter */ 200 loff_t offset; /* offset in the file */ 201 ssize_t size; /* size of the extent */ 202 } ext4_io_end_t; 203 204 struct ext4_io_submit { 205 struct writeback_control *io_wbc; 206 struct bio *io_bio; 207 ext4_io_end_t *io_end; 208 sector_t io_next_block; 209 }; 210 211 /* 212 * Special inodes numbers 213 */ 214 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 215 #define EXT4_ROOT_INO 2 /* Root inode */ 216 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 217 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 218 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 219 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 220 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 221 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 222 223 /* First non-reserved inode for old ext4 filesystems */ 224 #define EXT4_GOOD_OLD_FIRST_INO 11 225 226 /* 227 * Maximal count of links to a file 228 */ 229 #define EXT4_LINK_MAX 65000 230 231 /* 232 * Macro-instructions used to manage several block sizes 233 */ 234 #define EXT4_MIN_BLOCK_SIZE 1024 235 #define EXT4_MAX_BLOCK_SIZE 65536 236 #define EXT4_MIN_BLOCK_LOG_SIZE 10 237 #define EXT4_MAX_BLOCK_LOG_SIZE 16 238 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 239 #ifdef __KERNEL__ 240 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 241 #else 242 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 243 #endif 244 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 245 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 246 EXT4_SB(s)->s_cluster_bits) 247 #ifdef __KERNEL__ 248 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 249 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 250 #else 251 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 252 #endif 253 #ifdef __KERNEL__ 254 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 255 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 256 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 257 #else 258 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 259 EXT4_GOOD_OLD_INODE_SIZE : \ 260 (s)->s_inode_size) 261 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 262 EXT4_GOOD_OLD_FIRST_INO : \ 263 (s)->s_first_ino) 264 #endif 265 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 266 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 267 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 268 blkbits)) 269 270 /* Translate a block number to a cluster number */ 271 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 272 /* Translate a cluster number to a block number */ 273 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 274 /* Translate # of blks to # of clusters */ 275 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 276 (sbi)->s_cluster_bits) 277 /* Mask out the low bits to get the starting block of the cluster */ 278 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 279 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 280 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 281 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 282 /* Get the cluster offset */ 283 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 284 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 285 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 286 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 287 288 /* 289 * Structure of a blocks group descriptor 290 */ 291 struct ext4_group_desc 292 { 293 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 294 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 295 __le32 bg_inode_table_lo; /* Inodes table block */ 296 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 297 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 298 __le16 bg_used_dirs_count_lo; /* Directories count */ 299 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 300 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 301 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 302 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 303 __le16 bg_itable_unused_lo; /* Unused inodes count */ 304 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 305 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 306 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 307 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 308 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 309 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 310 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 311 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 312 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 313 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 314 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 315 __u32 bg_reserved; 316 }; 317 318 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 319 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 320 sizeof(__le16)) 321 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 322 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 323 sizeof(__le16)) 324 325 /* 326 * Structure of a flex block group info 327 */ 328 329 struct flex_groups { 330 atomic64_t free_clusters; 331 atomic_t free_inodes; 332 atomic_t used_dirs; 333 }; 334 335 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 336 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 337 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 338 339 /* 340 * Macro-instructions used to manage group descriptors 341 */ 342 #define EXT4_MIN_DESC_SIZE 32 343 #define EXT4_MIN_DESC_SIZE_64BIT 64 344 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 345 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 346 #ifdef __KERNEL__ 347 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 348 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 349 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 350 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 351 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 352 #else 353 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 355 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 356 #endif 357 358 /* 359 * Constants relative to the data blocks 360 */ 361 #define EXT4_NDIR_BLOCKS 12 362 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 363 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 364 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 365 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 366 367 /* 368 * Inode flags 369 */ 370 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 371 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 372 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 373 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 374 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 375 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 376 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 377 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 378 /* Reserved for compression usage... */ 379 #define EXT4_DIRTY_FL 0x00000100 380 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 381 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 382 /* nb: was previously EXT2_ECOMPR_FL */ 383 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 384 /* End compression flags --- maybe not all used */ 385 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 386 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 387 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 388 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 389 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 390 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 391 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 392 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 393 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 394 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 395 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 396 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 397 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 398 399 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 400 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 401 402 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 403 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 404 EXT4_IMMUTABLE_FL | \ 405 EXT4_APPEND_FL | \ 406 EXT4_NODUMP_FL | \ 407 EXT4_NOATIME_FL | \ 408 EXT4_PROJINHERIT_FL) 409 410 /* Flags that should be inherited by new inodes from their parent. */ 411 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 412 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 413 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 414 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 415 EXT4_PROJINHERIT_FL) 416 417 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 418 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 419 420 /* Flags that are appropriate for non-directories/regular files. */ 421 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 422 423 /* Mask out flags that are inappropriate for the given type of inode. */ 424 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 425 { 426 if (S_ISDIR(mode)) 427 return flags; 428 else if (S_ISREG(mode)) 429 return flags & EXT4_REG_FLMASK; 430 else 431 return flags & EXT4_OTHER_FLMASK; 432 } 433 434 /* 435 * Inode flags used for atomic set/get 436 */ 437 enum { 438 EXT4_INODE_SECRM = 0, /* Secure deletion */ 439 EXT4_INODE_UNRM = 1, /* Undelete */ 440 EXT4_INODE_COMPR = 2, /* Compress file */ 441 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 442 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 443 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 444 EXT4_INODE_NODUMP = 6, /* do not dump file */ 445 EXT4_INODE_NOATIME = 7, /* do not update atime */ 446 /* Reserved for compression usage... */ 447 EXT4_INODE_DIRTY = 8, 448 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 449 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 450 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 451 /* End compression flags --- maybe not all used */ 452 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 453 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 454 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 455 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 456 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 457 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 458 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 459 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 460 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 461 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 462 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 463 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 464 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 465 }; 466 467 /* 468 * Since it's pretty easy to mix up bit numbers and hex values, we use a 469 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 470 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 471 * any extra space in the compiled kernel image, otherwise, the build will fail. 472 * It's important that these values are the same, since we are using 473 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 474 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 475 * values found in ext2, ext3 and ext4 filesystems, and of course the values 476 * defined in e2fsprogs. 477 * 478 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 479 */ 480 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 481 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 482 483 static inline void ext4_check_flag_values(void) 484 { 485 CHECK_FLAG_VALUE(SECRM); 486 CHECK_FLAG_VALUE(UNRM); 487 CHECK_FLAG_VALUE(COMPR); 488 CHECK_FLAG_VALUE(SYNC); 489 CHECK_FLAG_VALUE(IMMUTABLE); 490 CHECK_FLAG_VALUE(APPEND); 491 CHECK_FLAG_VALUE(NODUMP); 492 CHECK_FLAG_VALUE(NOATIME); 493 CHECK_FLAG_VALUE(DIRTY); 494 CHECK_FLAG_VALUE(COMPRBLK); 495 CHECK_FLAG_VALUE(NOCOMPR); 496 CHECK_FLAG_VALUE(ENCRYPT); 497 CHECK_FLAG_VALUE(INDEX); 498 CHECK_FLAG_VALUE(IMAGIC); 499 CHECK_FLAG_VALUE(JOURNAL_DATA); 500 CHECK_FLAG_VALUE(NOTAIL); 501 CHECK_FLAG_VALUE(DIRSYNC); 502 CHECK_FLAG_VALUE(TOPDIR); 503 CHECK_FLAG_VALUE(HUGE_FILE); 504 CHECK_FLAG_VALUE(EXTENTS); 505 CHECK_FLAG_VALUE(EA_INODE); 506 CHECK_FLAG_VALUE(EOFBLOCKS); 507 CHECK_FLAG_VALUE(INLINE_DATA); 508 CHECK_FLAG_VALUE(PROJINHERIT); 509 CHECK_FLAG_VALUE(RESERVED); 510 } 511 512 /* Used to pass group descriptor data when online resize is done */ 513 struct ext4_new_group_input { 514 __u32 group; /* Group number for this data */ 515 __u64 block_bitmap; /* Absolute block number of block bitmap */ 516 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 517 __u64 inode_table; /* Absolute block number of inode table start */ 518 __u32 blocks_count; /* Total number of blocks in this group */ 519 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 520 __u16 unused; 521 }; 522 523 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 524 struct compat_ext4_new_group_input { 525 u32 group; 526 compat_u64 block_bitmap; 527 compat_u64 inode_bitmap; 528 compat_u64 inode_table; 529 u32 blocks_count; 530 u16 reserved_blocks; 531 u16 unused; 532 }; 533 #endif 534 535 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 536 struct ext4_new_group_data { 537 __u32 group; 538 __u64 block_bitmap; 539 __u64 inode_bitmap; 540 __u64 inode_table; 541 __u32 blocks_count; 542 __u16 reserved_blocks; 543 __u16 unused; 544 __u32 free_blocks_count; 545 }; 546 547 /* Indexes used to index group tables in ext4_new_group_data */ 548 enum { 549 BLOCK_BITMAP = 0, /* block bitmap */ 550 INODE_BITMAP, /* inode bitmap */ 551 INODE_TABLE, /* inode tables */ 552 GROUP_TABLE_COUNT, 553 }; 554 555 /* 556 * Flags used by ext4_map_blocks() 557 */ 558 /* Allocate any needed blocks and/or convert an unwritten 559 extent to be an initialized ext4 */ 560 #define EXT4_GET_BLOCKS_CREATE 0x0001 561 /* Request the creation of an unwritten extent */ 562 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 563 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 564 EXT4_GET_BLOCKS_CREATE) 565 /* Caller is from the delayed allocation writeout path 566 * finally doing the actual allocation of delayed blocks */ 567 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 568 /* caller is from the direct IO path, request to creation of an 569 unwritten extents if not allocated, split the unwritten 570 extent if blocks has been preallocated already*/ 571 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 572 #define EXT4_GET_BLOCKS_CONVERT 0x0010 573 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 574 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 575 /* Convert extent to initialized after IO complete */ 576 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 577 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 578 /* Eventual metadata allocation (due to growing extent tree) 579 * should not fail, so try to use reserved blocks for that.*/ 580 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 581 /* Don't normalize allocation size (used for fallocate) */ 582 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 583 /* Request will not result in inode size update (user for fallocate) */ 584 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 585 /* Convert written extents to unwritten */ 586 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 587 /* Write zeros to newly created written extents */ 588 #define EXT4_GET_BLOCKS_ZERO 0x0200 589 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 590 EXT4_GET_BLOCKS_ZERO) 591 /* Caller will submit data before dropping transaction handle. This 592 * allows jbd2 to avoid submitting data before commit. */ 593 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 594 595 /* 596 * The bit position of these flags must not overlap with any of the 597 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 598 * read_extent_tree_block(), ext4_split_extent_at(), 599 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 600 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 601 * caching the extents when reading from the extent tree while a 602 * truncate or punch hole operation is in progress. 603 */ 604 #define EXT4_EX_NOCACHE 0x40000000 605 #define EXT4_EX_FORCE_CACHE 0x20000000 606 607 /* 608 * Flags used by ext4_free_blocks 609 */ 610 #define EXT4_FREE_BLOCKS_METADATA 0x0001 611 #define EXT4_FREE_BLOCKS_FORGET 0x0002 612 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 613 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 614 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 615 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 616 617 /* 618 * ioctl commands 619 */ 620 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 621 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 622 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 623 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 624 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 625 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 626 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 627 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 628 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 629 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 630 #define EXT4_IOC_MIGRATE _IO('f', 9) 631 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 632 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 633 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 634 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 635 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 636 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 637 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 638 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 639 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 640 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 641 642 #ifndef FS_IOC_FSGETXATTR 643 /* Until the uapi changes get merged for project quota... */ 644 645 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr) 646 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr) 647 648 /* 649 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR. 650 */ 651 struct fsxattr { 652 __u32 fsx_xflags; /* xflags field value (get/set) */ 653 __u32 fsx_extsize; /* extsize field value (get/set)*/ 654 __u32 fsx_nextents; /* nextents field value (get) */ 655 __u32 fsx_projid; /* project identifier (get/set) */ 656 unsigned char fsx_pad[12]; 657 }; 658 659 /* 660 * Flags for the fsx_xflags field 661 */ 662 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */ 663 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */ 664 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */ 665 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */ 666 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */ 667 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */ 668 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */ 669 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */ 670 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */ 671 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */ 672 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */ 673 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */ 674 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */ 675 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */ 676 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */ 677 #endif /* !defined(FS_IOC_FSGETXATTR) */ 678 679 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 680 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 681 682 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 683 /* 684 * ioctl commands in 32 bit emulation 685 */ 686 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 687 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 688 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 689 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 690 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 691 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 692 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 693 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 694 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 695 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 696 #endif 697 698 /* Max physical block we can address w/o extents */ 699 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 700 701 /* 702 * Structure of an inode on the disk 703 */ 704 struct ext4_inode { 705 __le16 i_mode; /* File mode */ 706 __le16 i_uid; /* Low 16 bits of Owner Uid */ 707 __le32 i_size_lo; /* Size in bytes */ 708 __le32 i_atime; /* Access time */ 709 __le32 i_ctime; /* Inode Change time */ 710 __le32 i_mtime; /* Modification time */ 711 __le32 i_dtime; /* Deletion Time */ 712 __le16 i_gid; /* Low 16 bits of Group Id */ 713 __le16 i_links_count; /* Links count */ 714 __le32 i_blocks_lo; /* Blocks count */ 715 __le32 i_flags; /* File flags */ 716 union { 717 struct { 718 __le32 l_i_version; 719 } linux1; 720 struct { 721 __u32 h_i_translator; 722 } hurd1; 723 struct { 724 __u32 m_i_reserved1; 725 } masix1; 726 } osd1; /* OS dependent 1 */ 727 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 728 __le32 i_generation; /* File version (for NFS) */ 729 __le32 i_file_acl_lo; /* File ACL */ 730 __le32 i_size_high; 731 __le32 i_obso_faddr; /* Obsoleted fragment address */ 732 union { 733 struct { 734 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 735 __le16 l_i_file_acl_high; 736 __le16 l_i_uid_high; /* these 2 fields */ 737 __le16 l_i_gid_high; /* were reserved2[0] */ 738 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 739 __le16 l_i_reserved; 740 } linux2; 741 struct { 742 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 743 __u16 h_i_mode_high; 744 __u16 h_i_uid_high; 745 __u16 h_i_gid_high; 746 __u32 h_i_author; 747 } hurd2; 748 struct { 749 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 750 __le16 m_i_file_acl_high; 751 __u32 m_i_reserved2[2]; 752 } masix2; 753 } osd2; /* OS dependent 2 */ 754 __le16 i_extra_isize; 755 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 756 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 757 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 758 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 759 __le32 i_crtime; /* File Creation time */ 760 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 761 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 762 __le32 i_projid; /* Project ID */ 763 }; 764 765 struct move_extent { 766 __u32 reserved; /* should be zero */ 767 __u32 donor_fd; /* donor file descriptor */ 768 __u64 orig_start; /* logical start offset in block for orig */ 769 __u64 donor_start; /* logical start offset in block for donor */ 770 __u64 len; /* block length to be moved */ 771 __u64 moved_len; /* moved block length */ 772 }; 773 774 #define EXT4_EPOCH_BITS 2 775 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 776 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 777 778 /* 779 * Extended fields will fit into an inode if the filesystem was formatted 780 * with large inodes (-I 256 or larger) and there are not currently any EAs 781 * consuming all of the available space. For new inodes we always reserve 782 * enough space for the kernel's known extended fields, but for inodes 783 * created with an old kernel this might not have been the case. None of 784 * the extended inode fields is critical for correct filesystem operation. 785 * This macro checks if a certain field fits in the inode. Note that 786 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 787 */ 788 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 789 ((offsetof(typeof(*ext4_inode), field) + \ 790 sizeof((ext4_inode)->field)) \ 791 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 792 (einode)->i_extra_isize)) \ 793 794 /* 795 * We use an encoding that preserves the times for extra epoch "00": 796 * 797 * extra msb of adjust for signed 798 * epoch 32-bit 32-bit tv_sec to 799 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 800 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 801 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 802 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 803 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 804 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 805 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 806 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 807 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 808 * 809 * Note that previous versions of the kernel on 64-bit systems would 810 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 811 * 1970. e2fsck will correct this, assuming that it is run on the 812 * affected filesystem before 2242. 813 */ 814 815 static inline __le32 ext4_encode_extra_time(struct timespec *time) 816 { 817 u32 extra = sizeof(time->tv_sec) > 4 ? 818 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0; 819 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 820 } 821 822 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 823 { 824 if (unlikely(sizeof(time->tv_sec) > 4 && 825 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) { 826 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0) 827 /* Handle legacy encoding of pre-1970 dates with epoch 828 * bits 1,1. We assume that by kernel version 4.20, 829 * everyone will have run fsck over the affected 830 * filesystems to correct the problem. (This 831 * backwards compatibility may be removed before this 832 * time, at the discretion of the ext4 developers.) 833 */ 834 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 835 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 836 extra_bits = 0; 837 time->tv_sec += extra_bits << 32; 838 #else 839 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 840 #endif 841 } 842 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 843 } 844 845 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 846 do { \ 847 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 848 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 849 (raw_inode)->xtime ## _extra = \ 850 ext4_encode_extra_time(&(inode)->xtime); \ 851 } while (0) 852 853 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 854 do { \ 855 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 856 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 857 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 858 (raw_inode)->xtime ## _extra = \ 859 ext4_encode_extra_time(&(einode)->xtime); \ 860 } while (0) 861 862 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 863 do { \ 864 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 865 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 866 ext4_decode_extra_time(&(inode)->xtime, \ 867 raw_inode->xtime ## _extra); \ 868 else \ 869 (inode)->xtime.tv_nsec = 0; \ 870 } while (0) 871 872 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 873 do { \ 874 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 875 (einode)->xtime.tv_sec = \ 876 (signed)le32_to_cpu((raw_inode)->xtime); \ 877 else \ 878 (einode)->xtime.tv_sec = 0; \ 879 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 880 ext4_decode_extra_time(&(einode)->xtime, \ 881 raw_inode->xtime ## _extra); \ 882 else \ 883 (einode)->xtime.tv_nsec = 0; \ 884 } while (0) 885 886 #define i_disk_version osd1.linux1.l_i_version 887 888 #if defined(__KERNEL__) || defined(__linux__) 889 #define i_reserved1 osd1.linux1.l_i_reserved1 890 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 891 #define i_blocks_high osd2.linux2.l_i_blocks_high 892 #define i_uid_low i_uid 893 #define i_gid_low i_gid 894 #define i_uid_high osd2.linux2.l_i_uid_high 895 #define i_gid_high osd2.linux2.l_i_gid_high 896 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 897 898 #elif defined(__GNU__) 899 900 #define i_translator osd1.hurd1.h_i_translator 901 #define i_uid_high osd2.hurd2.h_i_uid_high 902 #define i_gid_high osd2.hurd2.h_i_gid_high 903 #define i_author osd2.hurd2.h_i_author 904 905 #elif defined(__masix__) 906 907 #define i_reserved1 osd1.masix1.m_i_reserved1 908 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 909 #define i_reserved2 osd2.masix2.m_i_reserved2 910 911 #endif /* defined(__KERNEL__) || defined(__linux__) */ 912 913 #include "extents_status.h" 914 915 /* 916 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 917 * 918 * These are needed to avoid lockdep false positives when we need to 919 * allocate blocks to the quota inode during ext4_map_blocks(), while 920 * holding i_data_sem for a normal (non-quota) inode. Since we don't 921 * do quota tracking for the quota inode, this avoids deadlock (as 922 * well as infinite recursion, since it isn't turtles all the way 923 * down...) 924 * 925 * I_DATA_SEM_NORMAL - Used for most inodes 926 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 927 * where the second inode has larger inode number 928 * than the first 929 * I_DATA_SEM_QUOTA - Used for quota inodes only 930 */ 931 enum { 932 I_DATA_SEM_NORMAL = 0, 933 I_DATA_SEM_OTHER, 934 I_DATA_SEM_QUOTA, 935 }; 936 937 938 /* 939 * fourth extended file system inode data in memory 940 */ 941 struct ext4_inode_info { 942 __le32 i_data[15]; /* unconverted */ 943 __u32 i_dtime; 944 ext4_fsblk_t i_file_acl; 945 946 /* 947 * i_block_group is the number of the block group which contains 948 * this file's inode. Constant across the lifetime of the inode, 949 * it is ued for making block allocation decisions - we try to 950 * place a file's data blocks near its inode block, and new inodes 951 * near to their parent directory's inode. 952 */ 953 ext4_group_t i_block_group; 954 ext4_lblk_t i_dir_start_lookup; 955 #if (BITS_PER_LONG < 64) 956 unsigned long i_state_flags; /* Dynamic state flags */ 957 #endif 958 unsigned long i_flags; 959 960 /* 961 * Extended attributes can be read independently of the main file 962 * data. Taking i_mutex even when reading would cause contention 963 * between readers of EAs and writers of regular file data, so 964 * instead we synchronize on xattr_sem when reading or changing 965 * EAs. 966 */ 967 struct rw_semaphore xattr_sem; 968 969 struct list_head i_orphan; /* unlinked but open inodes */ 970 971 /* 972 * i_disksize keeps track of what the inode size is ON DISK, not 973 * in memory. During truncate, i_size is set to the new size by 974 * the VFS prior to calling ext4_truncate(), but the filesystem won't 975 * set i_disksize to 0 until the truncate is actually under way. 976 * 977 * The intent is that i_disksize always represents the blocks which 978 * are used by this file. This allows recovery to restart truncate 979 * on orphans if we crash during truncate. We actually write i_disksize 980 * into the on-disk inode when writing inodes out, instead of i_size. 981 * 982 * The only time when i_disksize and i_size may be different is when 983 * a truncate is in progress. The only things which change i_disksize 984 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 985 */ 986 loff_t i_disksize; 987 988 /* 989 * i_data_sem is for serialising ext4_truncate() against 990 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 991 * data tree are chopped off during truncate. We can't do that in 992 * ext4 because whenever we perform intermediate commits during 993 * truncate, the inode and all the metadata blocks *must* be in a 994 * consistent state which allows truncation of the orphans to restart 995 * during recovery. Hence we must fix the get_block-vs-truncate race 996 * by other means, so we have i_data_sem. 997 */ 998 struct rw_semaphore i_data_sem; 999 /* 1000 * i_mmap_sem is for serializing page faults with truncate / punch hole 1001 * operations. We have to make sure that new page cannot be faulted in 1002 * a section of the inode that is being punched. We cannot easily use 1003 * i_data_sem for this since we need protection for the whole punch 1004 * operation and i_data_sem ranks below transaction start so we have 1005 * to occasionally drop it. 1006 */ 1007 struct rw_semaphore i_mmap_sem; 1008 struct inode vfs_inode; 1009 struct jbd2_inode *jinode; 1010 1011 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1012 1013 /* 1014 * File creation time. Its function is same as that of 1015 * struct timespec i_{a,c,m}time in the generic inode. 1016 */ 1017 struct timespec i_crtime; 1018 1019 /* mballoc */ 1020 struct list_head i_prealloc_list; 1021 spinlock_t i_prealloc_lock; 1022 1023 /* extents status tree */ 1024 struct ext4_es_tree i_es_tree; 1025 rwlock_t i_es_lock; 1026 struct list_head i_es_list; 1027 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1028 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1029 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1030 extents to shrink. Protected by 1031 i_es_lock */ 1032 1033 /* ialloc */ 1034 ext4_group_t i_last_alloc_group; 1035 1036 /* allocation reservation info for delalloc */ 1037 /* In case of bigalloc, these refer to clusters rather than blocks */ 1038 unsigned int i_reserved_data_blocks; 1039 unsigned int i_reserved_meta_blocks; 1040 unsigned int i_allocated_meta_blocks; 1041 ext4_lblk_t i_da_metadata_calc_last_lblock; 1042 int i_da_metadata_calc_len; 1043 1044 /* on-disk additional length */ 1045 __u16 i_extra_isize; 1046 1047 /* Indicate the inline data space. */ 1048 u16 i_inline_off; 1049 u16 i_inline_size; 1050 1051 #ifdef CONFIG_QUOTA 1052 /* quota space reservation, managed internally by quota code */ 1053 qsize_t i_reserved_quota; 1054 #endif 1055 1056 /* Lock protecting lists below */ 1057 spinlock_t i_completed_io_lock; 1058 /* 1059 * Completed IOs that need unwritten extents handling and have 1060 * transaction reserved 1061 */ 1062 struct list_head i_rsv_conversion_list; 1063 struct work_struct i_rsv_conversion_work; 1064 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1065 1066 spinlock_t i_block_reservation_lock; 1067 1068 /* 1069 * Transactions that contain inode's metadata needed to complete 1070 * fsync and fdatasync, respectively. 1071 */ 1072 tid_t i_sync_tid; 1073 tid_t i_datasync_tid; 1074 1075 #ifdef CONFIG_QUOTA 1076 struct dquot *i_dquot[MAXQUOTAS]; 1077 #endif 1078 1079 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1080 __u32 i_csum_seed; 1081 1082 kprojid_t i_projid; 1083 }; 1084 1085 /* 1086 * File system states 1087 */ 1088 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1089 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1090 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1091 1092 /* 1093 * Misc. filesystem flags 1094 */ 1095 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1096 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1097 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1098 1099 /* 1100 * Mount flags set via mount options or defaults 1101 */ 1102 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1103 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1104 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1105 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1106 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1107 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1108 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1109 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1110 #ifdef CONFIG_FS_DAX 1111 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1112 #else 1113 #define EXT4_MOUNT_DAX 0 1114 #endif 1115 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1116 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1117 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1118 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1119 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1120 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1121 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1122 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1123 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1124 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1125 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1126 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1127 * enable enforcement for hidden 1128 * quota files */ 1129 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1130 * enforcement for hidden quota 1131 * files */ 1132 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1133 * enforcement */ 1134 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1135 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1136 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1137 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1138 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1139 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1140 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1141 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1142 1143 /* 1144 * Mount flags set either automatically (could not be set by mount option) 1145 * based on per file system feature or property or in special cases such as 1146 * distinguishing between explicit mount option definition and default. 1147 */ 1148 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1149 specified delalloc */ 1150 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1151 size of blocksize * 8 1152 blocks */ 1153 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1154 file systems */ 1155 1156 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1157 specified journal checksum */ 1158 1159 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1160 ~EXT4_MOUNT_##opt 1161 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1162 EXT4_MOUNT_##opt 1163 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1164 EXT4_MOUNT_##opt) 1165 1166 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1167 ~EXT4_MOUNT2_##opt 1168 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1169 EXT4_MOUNT2_##opt 1170 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1171 EXT4_MOUNT2_##opt) 1172 1173 #define ext4_test_and_set_bit __test_and_set_bit_le 1174 #define ext4_set_bit __set_bit_le 1175 #define ext4_set_bit_atomic ext2_set_bit_atomic 1176 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1177 #define ext4_clear_bit __clear_bit_le 1178 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1179 #define ext4_test_bit test_bit_le 1180 #define ext4_find_next_zero_bit find_next_zero_bit_le 1181 #define ext4_find_next_bit find_next_bit_le 1182 1183 extern void ext4_set_bits(void *bm, int cur, int len); 1184 1185 /* 1186 * Maximal mount counts between two filesystem checks 1187 */ 1188 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1189 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1190 1191 /* 1192 * Behaviour when detecting errors 1193 */ 1194 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1195 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1196 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1197 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1198 1199 /* Metadata checksum algorithm codes */ 1200 #define EXT4_CRC32C_CHKSUM 1 1201 1202 /* 1203 * Structure of the super block 1204 */ 1205 struct ext4_super_block { 1206 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1207 __le32 s_blocks_count_lo; /* Blocks count */ 1208 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1209 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1210 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1211 __le32 s_first_data_block; /* First Data Block */ 1212 __le32 s_log_block_size; /* Block size */ 1213 __le32 s_log_cluster_size; /* Allocation cluster size */ 1214 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1215 __le32 s_clusters_per_group; /* # Clusters per group */ 1216 __le32 s_inodes_per_group; /* # Inodes per group */ 1217 __le32 s_mtime; /* Mount time */ 1218 /*30*/ __le32 s_wtime; /* Write time */ 1219 __le16 s_mnt_count; /* Mount count */ 1220 __le16 s_max_mnt_count; /* Maximal mount count */ 1221 __le16 s_magic; /* Magic signature */ 1222 __le16 s_state; /* File system state */ 1223 __le16 s_errors; /* Behaviour when detecting errors */ 1224 __le16 s_minor_rev_level; /* minor revision level */ 1225 /*40*/ __le32 s_lastcheck; /* time of last check */ 1226 __le32 s_checkinterval; /* max. time between checks */ 1227 __le32 s_creator_os; /* OS */ 1228 __le32 s_rev_level; /* Revision level */ 1229 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1230 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1231 /* 1232 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1233 * 1234 * Note: the difference between the compatible feature set and 1235 * the incompatible feature set is that if there is a bit set 1236 * in the incompatible feature set that the kernel doesn't 1237 * know about, it should refuse to mount the filesystem. 1238 * 1239 * e2fsck's requirements are more strict; if it doesn't know 1240 * about a feature in either the compatible or incompatible 1241 * feature set, it must abort and not try to meddle with 1242 * things it doesn't understand... 1243 */ 1244 __le32 s_first_ino; /* First non-reserved inode */ 1245 __le16 s_inode_size; /* size of inode structure */ 1246 __le16 s_block_group_nr; /* block group # of this superblock */ 1247 __le32 s_feature_compat; /* compatible feature set */ 1248 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1249 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1250 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1251 /*78*/ char s_volume_name[16]; /* volume name */ 1252 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1253 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1254 /* 1255 * Performance hints. Directory preallocation should only 1256 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1257 */ 1258 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1259 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1260 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1261 /* 1262 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1263 */ 1264 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1265 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1266 __le32 s_journal_dev; /* device number of journal file */ 1267 __le32 s_last_orphan; /* start of list of inodes to delete */ 1268 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1269 __u8 s_def_hash_version; /* Default hash version to use */ 1270 __u8 s_jnl_backup_type; 1271 __le16 s_desc_size; /* size of group descriptor */ 1272 /*100*/ __le32 s_default_mount_opts; 1273 __le32 s_first_meta_bg; /* First metablock block group */ 1274 __le32 s_mkfs_time; /* When the filesystem was created */ 1275 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1276 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1277 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1278 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1279 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1280 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1281 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1282 __le32 s_flags; /* Miscellaneous flags */ 1283 __le16 s_raid_stride; /* RAID stride */ 1284 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1285 __le64 s_mmp_block; /* Block for multi-mount protection */ 1286 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1287 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1288 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1289 __u8 s_encryption_level; /* versioning level for encryption */ 1290 __u8 s_reserved_pad; /* Padding to next 32bits */ 1291 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1292 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1293 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1294 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1295 snapshot's future use */ 1296 __le32 s_snapshot_list; /* inode number of the head of the 1297 on-disk snapshot list */ 1298 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1299 __le32 s_error_count; /* number of fs errors */ 1300 __le32 s_first_error_time; /* first time an error happened */ 1301 __le32 s_first_error_ino; /* inode involved in first error */ 1302 __le64 s_first_error_block; /* block involved of first error */ 1303 __u8 s_first_error_func[32]; /* function where the error happened */ 1304 __le32 s_first_error_line; /* line number where error happened */ 1305 __le32 s_last_error_time; /* most recent time of an error */ 1306 __le32 s_last_error_ino; /* inode involved in last error */ 1307 __le32 s_last_error_line; /* line number where error happened */ 1308 __le64 s_last_error_block; /* block involved of last error */ 1309 __u8 s_last_error_func[32]; /* function where the error happened */ 1310 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1311 __u8 s_mount_opts[64]; 1312 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1313 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1314 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1315 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1316 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1317 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1318 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1319 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1320 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1321 __le32 s_reserved[98]; /* Padding to the end of the block */ 1322 __le32 s_checksum; /* crc32c(superblock) */ 1323 }; 1324 1325 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1326 1327 #ifdef __KERNEL__ 1328 1329 /* 1330 * run-time mount flags 1331 */ 1332 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1333 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1334 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1335 1336 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1337 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1338 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1339 #else 1340 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1341 #endif 1342 1343 /* Number of quota types we support */ 1344 #define EXT4_MAXQUOTAS 3 1345 1346 /* 1347 * fourth extended-fs super-block data in memory 1348 */ 1349 struct ext4_sb_info { 1350 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1351 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1352 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1353 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1354 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1355 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1356 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1357 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1358 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1359 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1360 unsigned long s_overhead; /* # of fs overhead clusters */ 1361 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1362 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1363 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1364 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1365 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1366 struct buffer_head **s_group_desc; 1367 unsigned int s_mount_opt; 1368 unsigned int s_mount_opt2; 1369 unsigned int s_mount_flags; 1370 unsigned int s_def_mount_opt; 1371 ext4_fsblk_t s_sb_block; 1372 atomic64_t s_resv_clusters; 1373 kuid_t s_resuid; 1374 kgid_t s_resgid; 1375 unsigned short s_mount_state; 1376 unsigned short s_pad; 1377 int s_addr_per_block_bits; 1378 int s_desc_per_block_bits; 1379 int s_inode_size; 1380 int s_first_ino; 1381 unsigned int s_inode_readahead_blks; 1382 unsigned int s_inode_goal; 1383 spinlock_t s_next_gen_lock; 1384 u32 s_next_generation; 1385 u32 s_hash_seed[4]; 1386 int s_def_hash_version; 1387 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1388 struct percpu_counter s_freeclusters_counter; 1389 struct percpu_counter s_freeinodes_counter; 1390 struct percpu_counter s_dirs_counter; 1391 struct percpu_counter s_dirtyclusters_counter; 1392 struct blockgroup_lock *s_blockgroup_lock; 1393 struct proc_dir_entry *s_proc; 1394 struct kobject s_kobj; 1395 struct completion s_kobj_unregister; 1396 struct super_block *s_sb; 1397 1398 /* Journaling */ 1399 struct journal_s *s_journal; 1400 struct list_head s_orphan; 1401 struct mutex s_orphan_lock; 1402 unsigned long s_resize_flags; /* Flags indicating if there 1403 is a resizer */ 1404 unsigned long s_commit_interval; 1405 u32 s_max_batch_time; 1406 u32 s_min_batch_time; 1407 struct block_device *journal_bdev; 1408 #ifdef CONFIG_QUOTA 1409 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1410 int s_jquota_fmt; /* Format of quota to use */ 1411 #endif 1412 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1413 struct rb_root system_blks; 1414 1415 #ifdef EXTENTS_STATS 1416 /* ext4 extents stats */ 1417 unsigned long s_ext_min; 1418 unsigned long s_ext_max; 1419 unsigned long s_depth_max; 1420 spinlock_t s_ext_stats_lock; 1421 unsigned long s_ext_blocks; 1422 unsigned long s_ext_extents; 1423 #endif 1424 1425 /* for buddy allocator */ 1426 struct ext4_group_info ***s_group_info; 1427 struct inode *s_buddy_cache; 1428 spinlock_t s_md_lock; 1429 unsigned short *s_mb_offsets; 1430 unsigned int *s_mb_maxs; 1431 unsigned int s_group_info_size; 1432 unsigned int s_mb_free_pending; 1433 1434 /* tunables */ 1435 unsigned long s_stripe; 1436 unsigned int s_mb_stream_request; 1437 unsigned int s_mb_max_to_scan; 1438 unsigned int s_mb_min_to_scan; 1439 unsigned int s_mb_stats; 1440 unsigned int s_mb_order2_reqs; 1441 unsigned int s_mb_group_prealloc; 1442 unsigned int s_max_dir_size_kb; 1443 /* where last allocation was done - for stream allocation */ 1444 unsigned long s_mb_last_group; 1445 unsigned long s_mb_last_start; 1446 1447 /* stats for buddy allocator */ 1448 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1449 atomic_t s_bal_success; /* we found long enough chunks */ 1450 atomic_t s_bal_allocated; /* in blocks */ 1451 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1452 atomic_t s_bal_goals; /* goal hits */ 1453 atomic_t s_bal_breaks; /* too long searches */ 1454 atomic_t s_bal_2orders; /* 2^order hits */ 1455 spinlock_t s_bal_lock; 1456 unsigned long s_mb_buddies_generated; 1457 unsigned long long s_mb_generation_time; 1458 atomic_t s_mb_lost_chunks; 1459 atomic_t s_mb_preallocated; 1460 atomic_t s_mb_discarded; 1461 atomic_t s_lock_busy; 1462 1463 /* locality groups */ 1464 struct ext4_locality_group __percpu *s_locality_groups; 1465 1466 /* for write statistics */ 1467 unsigned long s_sectors_written_start; 1468 u64 s_kbytes_written; 1469 1470 /* the size of zero-out chunk */ 1471 unsigned int s_extent_max_zeroout_kb; 1472 1473 unsigned int s_log_groups_per_flex; 1474 struct flex_groups *s_flex_groups; 1475 ext4_group_t s_flex_groups_allocated; 1476 1477 /* workqueue for reserved extent conversions (buffered io) */ 1478 struct workqueue_struct *rsv_conversion_wq; 1479 1480 /* timer for periodic error stats printing */ 1481 struct timer_list s_err_report; 1482 1483 /* Lazy inode table initialization info */ 1484 struct ext4_li_request *s_li_request; 1485 /* Wait multiplier for lazy initialization thread */ 1486 unsigned int s_li_wait_mult; 1487 1488 /* Kernel thread for multiple mount protection */ 1489 struct task_struct *s_mmp_tsk; 1490 1491 /* record the last minlen when FITRIM is called. */ 1492 atomic_t s_last_trim_minblks; 1493 1494 /* Reference to checksum algorithm driver via cryptoapi */ 1495 struct crypto_shash *s_chksum_driver; 1496 1497 /* Precomputed FS UUID checksum for seeding other checksums */ 1498 __u32 s_csum_seed; 1499 1500 /* Reclaim extents from extent status tree */ 1501 struct shrinker s_es_shrinker; 1502 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1503 long s_es_nr_inode; 1504 struct ext4_es_stats s_es_stats; 1505 struct mb_cache *s_mb_cache; 1506 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1507 1508 /* Ratelimit ext4 messages. */ 1509 struct ratelimit_state s_err_ratelimit_state; 1510 struct ratelimit_state s_warning_ratelimit_state; 1511 struct ratelimit_state s_msg_ratelimit_state; 1512 1513 /* Barrier between changing inodes' journal flags and writepages ops. */ 1514 struct percpu_rw_semaphore s_journal_flag_rwsem; 1515 }; 1516 1517 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1518 { 1519 return sb->s_fs_info; 1520 } 1521 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1522 { 1523 return container_of(inode, struct ext4_inode_info, vfs_inode); 1524 } 1525 1526 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1527 { 1528 return ino == EXT4_ROOT_INO || 1529 ino == EXT4_USR_QUOTA_INO || 1530 ino == EXT4_GRP_QUOTA_INO || 1531 ino == EXT4_BOOT_LOADER_INO || 1532 ino == EXT4_JOURNAL_INO || 1533 ino == EXT4_RESIZE_INO || 1534 (ino >= EXT4_FIRST_INO(sb) && 1535 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1536 } 1537 1538 /* 1539 * Inode dynamic state flags 1540 */ 1541 enum { 1542 EXT4_STATE_JDATA, /* journaled data exists */ 1543 EXT4_STATE_NEW, /* inode is newly created */ 1544 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1545 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1546 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1547 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1548 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1549 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1550 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1551 nolocking */ 1552 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1553 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1554 }; 1555 1556 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1557 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1558 { \ 1559 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1560 } \ 1561 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1562 { \ 1563 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1564 } \ 1565 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1566 { \ 1567 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1568 } 1569 1570 /* Add these declarations here only so that these functions can be 1571 * found by name. Otherwise, they are very hard to locate. */ 1572 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1573 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1574 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1575 EXT4_INODE_BIT_FNS(flag, flags, 0) 1576 1577 /* Add these declarations here only so that these functions can be 1578 * found by name. Otherwise, they are very hard to locate. */ 1579 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1580 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1581 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1582 #if (BITS_PER_LONG < 64) 1583 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1584 1585 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1586 { 1587 (ei)->i_state_flags = 0; 1588 } 1589 #else 1590 EXT4_INODE_BIT_FNS(state, flags, 32) 1591 1592 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1593 { 1594 /* We depend on the fact that callers will set i_flags */ 1595 } 1596 #endif 1597 #else 1598 /* Assume that user mode programs are passing in an ext4fs superblock, not 1599 * a kernel struct super_block. This will allow us to call the feature-test 1600 * macros from user land. */ 1601 #define EXT4_SB(sb) (sb) 1602 #endif 1603 1604 /* 1605 * Returns true if the inode is inode is encrypted 1606 */ 1607 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1608 1609 /* 1610 * Codes for operating systems 1611 */ 1612 #define EXT4_OS_LINUX 0 1613 #define EXT4_OS_HURD 1 1614 #define EXT4_OS_MASIX 2 1615 #define EXT4_OS_FREEBSD 3 1616 #define EXT4_OS_LITES 4 1617 1618 /* 1619 * Revision levels 1620 */ 1621 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1622 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1623 1624 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1625 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1626 1627 #define EXT4_GOOD_OLD_INODE_SIZE 128 1628 1629 /* 1630 * Feature set definitions 1631 */ 1632 1633 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1634 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1635 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1636 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1637 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1638 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1639 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1640 1641 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1642 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1643 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1644 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1645 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1646 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1647 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1648 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1649 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1650 /* 1651 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1652 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1653 * all other data structures' checksums. However, the METADATA_CSUM and 1654 * GDT_CSUM bits are mutually exclusive. 1655 */ 1656 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1657 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1658 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1659 1660 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1661 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1662 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1663 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1664 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1665 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1666 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1667 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1668 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1669 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1670 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1671 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1672 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1673 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1674 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1675 1676 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1677 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1678 { \ 1679 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1680 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1681 } \ 1682 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1683 { \ 1684 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1685 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1686 } \ 1687 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1688 { \ 1689 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1690 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1691 } 1692 1693 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1694 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1695 { \ 1696 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1697 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1698 } \ 1699 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1700 { \ 1701 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1702 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1703 } \ 1704 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1705 { \ 1706 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1707 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1708 } 1709 1710 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1711 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1712 { \ 1713 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1714 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1715 } \ 1716 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1717 { \ 1718 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1719 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1720 } \ 1721 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1722 { \ 1723 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1724 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1725 } 1726 1727 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1728 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1729 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1730 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1731 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1732 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1733 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1734 1735 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1736 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1737 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1738 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1739 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1740 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1741 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1742 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1743 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1744 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1745 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1746 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1747 1748 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1749 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1750 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1751 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1752 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1753 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1754 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1755 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1756 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1757 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1758 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1759 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1760 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1761 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1762 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1763 1764 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1765 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1766 EXT4_FEATURE_INCOMPAT_META_BG) 1767 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1768 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1769 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1770 1771 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1772 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1773 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1774 EXT4_FEATURE_INCOMPAT_META_BG) 1775 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1776 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1777 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1778 1779 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1780 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1781 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1782 EXT4_FEATURE_INCOMPAT_META_BG| \ 1783 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1784 EXT4_FEATURE_INCOMPAT_64BIT| \ 1785 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1786 EXT4_FEATURE_INCOMPAT_MMP | \ 1787 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1788 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1789 EXT4_FEATURE_INCOMPAT_CSUM_SEED) 1790 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1791 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1792 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1793 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1794 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1795 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1796 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1797 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1798 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1799 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1800 EXT4_FEATURE_RO_COMPAT_PROJECT) 1801 1802 #define EXTN_FEATURE_FUNCS(ver) \ 1803 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1804 { \ 1805 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1806 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1807 } \ 1808 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1809 { \ 1810 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1811 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1812 } \ 1813 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1814 { \ 1815 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1816 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1817 } 1818 1819 EXTN_FEATURE_FUNCS(2) 1820 EXTN_FEATURE_FUNCS(3) 1821 EXTN_FEATURE_FUNCS(4) 1822 1823 static inline bool ext4_has_compat_features(struct super_block *sb) 1824 { 1825 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1826 } 1827 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1828 { 1829 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1830 } 1831 static inline bool ext4_has_incompat_features(struct super_block *sb) 1832 { 1833 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1834 } 1835 1836 /* 1837 * Default values for user and/or group using reserved blocks 1838 */ 1839 #define EXT4_DEF_RESUID 0 1840 #define EXT4_DEF_RESGID 0 1841 1842 /* 1843 * Default project ID 1844 */ 1845 #define EXT4_DEF_PROJID 0 1846 1847 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1848 1849 /* 1850 * Default mount options 1851 */ 1852 #define EXT4_DEFM_DEBUG 0x0001 1853 #define EXT4_DEFM_BSDGROUPS 0x0002 1854 #define EXT4_DEFM_XATTR_USER 0x0004 1855 #define EXT4_DEFM_ACL 0x0008 1856 #define EXT4_DEFM_UID16 0x0010 1857 #define EXT4_DEFM_JMODE 0x0060 1858 #define EXT4_DEFM_JMODE_DATA 0x0020 1859 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1860 #define EXT4_DEFM_JMODE_WBACK 0x0060 1861 #define EXT4_DEFM_NOBARRIER 0x0100 1862 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1863 #define EXT4_DEFM_DISCARD 0x0400 1864 #define EXT4_DEFM_NODELALLOC 0x0800 1865 1866 /* 1867 * Default journal batch times 1868 */ 1869 #define EXT4_DEF_MIN_BATCH_TIME 0 1870 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1871 1872 /* 1873 * Minimum number of groups in a flexgroup before we separate out 1874 * directories into the first block group of a flexgroup 1875 */ 1876 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1877 1878 /* 1879 * Structure of a directory entry 1880 */ 1881 #define EXT4_NAME_LEN 255 1882 1883 struct ext4_dir_entry { 1884 __le32 inode; /* Inode number */ 1885 __le16 rec_len; /* Directory entry length */ 1886 __le16 name_len; /* Name length */ 1887 char name[EXT4_NAME_LEN]; /* File name */ 1888 }; 1889 1890 /* 1891 * The new version of the directory entry. Since EXT4 structures are 1892 * stored in intel byte order, and the name_len field could never be 1893 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1894 * file_type field. 1895 */ 1896 struct ext4_dir_entry_2 { 1897 __le32 inode; /* Inode number */ 1898 __le16 rec_len; /* Directory entry length */ 1899 __u8 name_len; /* Name length */ 1900 __u8 file_type; 1901 char name[EXT4_NAME_LEN]; /* File name */ 1902 }; 1903 1904 /* 1905 * This is a bogus directory entry at the end of each leaf block that 1906 * records checksums. 1907 */ 1908 struct ext4_dir_entry_tail { 1909 __le32 det_reserved_zero1; /* Pretend to be unused */ 1910 __le16 det_rec_len; /* 12 */ 1911 __u8 det_reserved_zero2; /* Zero name length */ 1912 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1913 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1914 }; 1915 1916 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1917 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1918 ((blocksize) - \ 1919 sizeof(struct ext4_dir_entry_tail)))) 1920 1921 /* 1922 * Ext4 directory file types. Only the low 3 bits are used. The 1923 * other bits are reserved for now. 1924 */ 1925 #define EXT4_FT_UNKNOWN 0 1926 #define EXT4_FT_REG_FILE 1 1927 #define EXT4_FT_DIR 2 1928 #define EXT4_FT_CHRDEV 3 1929 #define EXT4_FT_BLKDEV 4 1930 #define EXT4_FT_FIFO 5 1931 #define EXT4_FT_SOCK 6 1932 #define EXT4_FT_SYMLINK 7 1933 1934 #define EXT4_FT_MAX 8 1935 1936 #define EXT4_FT_DIR_CSUM 0xDE 1937 1938 /* 1939 * EXT4_DIR_PAD defines the directory entries boundaries 1940 * 1941 * NOTE: It must be a multiple of 4 1942 */ 1943 #define EXT4_DIR_PAD 4 1944 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1945 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1946 ~EXT4_DIR_ROUND) 1947 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1948 1949 /* 1950 * If we ever get support for fs block sizes > page_size, we'll need 1951 * to remove the #if statements in the next two functions... 1952 */ 1953 static inline unsigned int 1954 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1955 { 1956 unsigned len = le16_to_cpu(dlen); 1957 1958 #if (PAGE_SIZE >= 65536) 1959 if (len == EXT4_MAX_REC_LEN || len == 0) 1960 return blocksize; 1961 return (len & 65532) | ((len & 3) << 16); 1962 #else 1963 return len; 1964 #endif 1965 } 1966 1967 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1968 { 1969 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1970 BUG(); 1971 #if (PAGE_SIZE >= 65536) 1972 if (len < 65536) 1973 return cpu_to_le16(len); 1974 if (len == blocksize) { 1975 if (blocksize == 65536) 1976 return cpu_to_le16(EXT4_MAX_REC_LEN); 1977 else 1978 return cpu_to_le16(0); 1979 } 1980 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1981 #else 1982 return cpu_to_le16(len); 1983 #endif 1984 } 1985 1986 /* 1987 * Hash Tree Directory indexing 1988 * (c) Daniel Phillips, 2001 1989 */ 1990 1991 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1992 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1993 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1994 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1995 1996 /* Legal values for the dx_root hash_version field: */ 1997 1998 #define DX_HASH_LEGACY 0 1999 #define DX_HASH_HALF_MD4 1 2000 #define DX_HASH_TEA 2 2001 #define DX_HASH_LEGACY_UNSIGNED 3 2002 #define DX_HASH_HALF_MD4_UNSIGNED 4 2003 #define DX_HASH_TEA_UNSIGNED 5 2004 2005 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2006 const void *address, unsigned int length) 2007 { 2008 struct { 2009 struct shash_desc shash; 2010 char ctx[4]; 2011 } desc; 2012 int err; 2013 2014 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2015 2016 desc.shash.tfm = sbi->s_chksum_driver; 2017 desc.shash.flags = 0; 2018 *(u32 *)desc.ctx = crc; 2019 2020 err = crypto_shash_update(&desc.shash, address, length); 2021 BUG_ON(err); 2022 2023 return *(u32 *)desc.ctx; 2024 } 2025 2026 #ifdef __KERNEL__ 2027 2028 /* hash info structure used by the directory hash */ 2029 struct dx_hash_info 2030 { 2031 u32 hash; 2032 u32 minor_hash; 2033 int hash_version; 2034 u32 *seed; 2035 }; 2036 2037 2038 /* 32 and 64 bit signed EOF for dx directories */ 2039 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2040 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2041 2042 2043 /* 2044 * Control parameters used by ext4_htree_next_block 2045 */ 2046 #define HASH_NB_ALWAYS 1 2047 2048 struct ext4_filename { 2049 const struct qstr *usr_fname; 2050 struct fscrypt_str disk_name; 2051 struct dx_hash_info hinfo; 2052 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2053 struct fscrypt_str crypto_buf; 2054 #endif 2055 }; 2056 2057 #define fname_name(p) ((p)->disk_name.name) 2058 #define fname_len(p) ((p)->disk_name.len) 2059 2060 /* 2061 * Describe an inode's exact location on disk and in memory 2062 */ 2063 struct ext4_iloc 2064 { 2065 struct buffer_head *bh; 2066 unsigned long offset; 2067 ext4_group_t block_group; 2068 }; 2069 2070 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2071 { 2072 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2073 } 2074 2075 /* 2076 * This structure is stuffed into the struct file's private_data field 2077 * for directories. It is where we put information so that we can do 2078 * readdir operations in hash tree order. 2079 */ 2080 struct dir_private_info { 2081 struct rb_root root; 2082 struct rb_node *curr_node; 2083 struct fname *extra_fname; 2084 loff_t last_pos; 2085 __u32 curr_hash; 2086 __u32 curr_minor_hash; 2087 __u32 next_hash; 2088 }; 2089 2090 /* calculate the first block number of the group */ 2091 static inline ext4_fsblk_t 2092 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2093 { 2094 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2095 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2096 } 2097 2098 /* 2099 * Special error return code only used by dx_probe() and its callers. 2100 */ 2101 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2102 2103 /* 2104 * Timeout and state flag for lazy initialization inode thread. 2105 */ 2106 #define EXT4_DEF_LI_WAIT_MULT 10 2107 #define EXT4_DEF_LI_MAX_START_DELAY 5 2108 #define EXT4_LAZYINIT_QUIT 0x0001 2109 #define EXT4_LAZYINIT_RUNNING 0x0002 2110 2111 /* 2112 * Lazy inode table initialization info 2113 */ 2114 struct ext4_lazy_init { 2115 unsigned long li_state; 2116 struct list_head li_request_list; 2117 struct mutex li_list_mtx; 2118 }; 2119 2120 struct ext4_li_request { 2121 struct super_block *lr_super; 2122 struct ext4_sb_info *lr_sbi; 2123 ext4_group_t lr_next_group; 2124 struct list_head lr_request; 2125 unsigned long lr_next_sched; 2126 unsigned long lr_timeout; 2127 }; 2128 2129 struct ext4_features { 2130 struct kobject f_kobj; 2131 struct completion f_kobj_unregister; 2132 }; 2133 2134 /* 2135 * This structure will be used for multiple mount protection. It will be 2136 * written into the block number saved in the s_mmp_block field in the 2137 * superblock. Programs that check MMP should assume that if 2138 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2139 * to use the filesystem, regardless of how old the timestamp is. 2140 */ 2141 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2142 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2143 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2144 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2145 2146 struct mmp_struct { 2147 __le32 mmp_magic; /* Magic number for MMP */ 2148 __le32 mmp_seq; /* Sequence no. updated periodically */ 2149 2150 /* 2151 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2152 * purposes and do not affect the correctness of the algorithm 2153 */ 2154 __le64 mmp_time; /* Time last updated */ 2155 char mmp_nodename[64]; /* Node which last updated MMP block */ 2156 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2157 2158 /* 2159 * mmp_check_interval is used to verify if the MMP block has been 2160 * updated on the block device. The value is updated based on the 2161 * maximum time to write the MMP block during an update cycle. 2162 */ 2163 __le16 mmp_check_interval; 2164 2165 __le16 mmp_pad1; 2166 __le32 mmp_pad2[226]; 2167 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2168 }; 2169 2170 /* arguments passed to the mmp thread */ 2171 struct mmpd_data { 2172 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2173 struct super_block *sb; /* super block of the fs */ 2174 }; 2175 2176 /* 2177 * Check interval multiplier 2178 * The MMP block is written every update interval and initially checked every 2179 * update interval x the multiplier (the value is then adapted based on the 2180 * write latency). The reason is that writes can be delayed under load and we 2181 * don't want readers to incorrectly assume that the filesystem is no longer 2182 * in use. 2183 */ 2184 #define EXT4_MMP_CHECK_MULT 2UL 2185 2186 /* 2187 * Minimum interval for MMP checking in seconds. 2188 */ 2189 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2190 2191 /* 2192 * Maximum interval for MMP checking in seconds. 2193 */ 2194 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2195 2196 /* 2197 * Function prototypes 2198 */ 2199 2200 /* 2201 * Ok, these declarations are also in <linux/kernel.h> but none of the 2202 * ext4 source programs needs to include it so they are duplicated here. 2203 */ 2204 # define NORET_TYPE /**/ 2205 # define ATTRIB_NORET __attribute__((noreturn)) 2206 # define NORET_AND noreturn, 2207 2208 /* bitmap.c */ 2209 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2210 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2211 struct ext4_group_desc *gdp, 2212 struct buffer_head *bh, int sz); 2213 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2214 struct ext4_group_desc *gdp, 2215 struct buffer_head *bh, int sz); 2216 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2217 struct ext4_group_desc *gdp, 2218 struct buffer_head *bh); 2219 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2220 struct ext4_group_desc *gdp, 2221 struct buffer_head *bh); 2222 2223 /* balloc.c */ 2224 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2225 ext4_fsblk_t blocknr, 2226 ext4_group_t *blockgrpp, 2227 ext4_grpblk_t *offsetp); 2228 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2229 ext4_fsblk_t block); 2230 2231 extern unsigned int ext4_block_group(struct super_block *sb, 2232 ext4_fsblk_t blocknr); 2233 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2234 ext4_fsblk_t blocknr); 2235 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2236 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2237 ext4_group_t group); 2238 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2239 ext4_fsblk_t goal, 2240 unsigned int flags, 2241 unsigned long *count, 2242 int *errp); 2243 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2244 s64 nclusters, unsigned int flags); 2245 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2246 extern void ext4_check_blocks_bitmap(struct super_block *); 2247 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2248 ext4_group_t block_group, 2249 struct buffer_head ** bh); 2250 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2251 2252 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2253 ext4_group_t block_group); 2254 extern int ext4_wait_block_bitmap(struct super_block *sb, 2255 ext4_group_t block_group, 2256 struct buffer_head *bh); 2257 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2258 ext4_group_t block_group); 2259 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2260 ext4_group_t block_group, 2261 struct ext4_group_desc *gdp); 2262 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2263 2264 static inline bool ext4_encrypted_inode(struct inode *inode) 2265 { 2266 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2267 } 2268 2269 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2270 static inline int ext4_fname_setup_filename(struct inode *dir, 2271 const struct qstr *iname, 2272 int lookup, struct ext4_filename *fname) 2273 { 2274 struct fscrypt_name name; 2275 int err; 2276 2277 memset(fname, 0, sizeof(struct ext4_filename)); 2278 2279 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2280 2281 fname->usr_fname = name.usr_fname; 2282 fname->disk_name = name.disk_name; 2283 fname->hinfo.hash = name.hash; 2284 fname->hinfo.minor_hash = name.minor_hash; 2285 fname->crypto_buf = name.crypto_buf; 2286 return err; 2287 } 2288 2289 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2290 { 2291 struct fscrypt_name name; 2292 2293 name.crypto_buf = fname->crypto_buf; 2294 fscrypt_free_filename(&name); 2295 2296 fname->crypto_buf.name = NULL; 2297 fname->usr_fname = NULL; 2298 fname->disk_name.name = NULL; 2299 } 2300 #else 2301 static inline int ext4_fname_setup_filename(struct inode *dir, 2302 const struct qstr *iname, 2303 int lookup, struct ext4_filename *fname) 2304 { 2305 fname->usr_fname = iname; 2306 fname->disk_name.name = (unsigned char *) iname->name; 2307 fname->disk_name.len = iname->len; 2308 return 0; 2309 } 2310 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2311 2312 #define fscrypt_set_d_op(i) 2313 #define fscrypt_get_ctx fscrypt_notsupp_get_ctx 2314 #define fscrypt_release_ctx fscrypt_notsupp_release_ctx 2315 #define fscrypt_encrypt_page fscrypt_notsupp_encrypt_page 2316 #define fscrypt_decrypt_page fscrypt_notsupp_decrypt_page 2317 #define fscrypt_decrypt_bio_pages fscrypt_notsupp_decrypt_bio_pages 2318 #define fscrypt_pullback_bio_page fscrypt_notsupp_pullback_bio_page 2319 #define fscrypt_restore_control_page fscrypt_notsupp_restore_control_page 2320 #define fscrypt_zeroout_range fscrypt_notsupp_zeroout_range 2321 #define fscrypt_ioctl_set_policy fscrypt_notsupp_ioctl_set_policy 2322 #define fscrypt_ioctl_get_policy fscrypt_notsupp_ioctl_get_policy 2323 #define fscrypt_has_permitted_context fscrypt_notsupp_has_permitted_context 2324 #define fscrypt_inherit_context fscrypt_notsupp_inherit_context 2325 #define fscrypt_get_encryption_info fscrypt_notsupp_get_encryption_info 2326 #define fscrypt_put_encryption_info fscrypt_notsupp_put_encryption_info 2327 #define fscrypt_setup_filename fscrypt_notsupp_setup_filename 2328 #define fscrypt_free_filename fscrypt_notsupp_free_filename 2329 #define fscrypt_fname_encrypted_size fscrypt_notsupp_fname_encrypted_size 2330 #define fscrypt_fname_alloc_buffer fscrypt_notsupp_fname_alloc_buffer 2331 #define fscrypt_fname_free_buffer fscrypt_notsupp_fname_free_buffer 2332 #define fscrypt_fname_disk_to_usr fscrypt_notsupp_fname_disk_to_usr 2333 #define fscrypt_fname_usr_to_disk fscrypt_notsupp_fname_usr_to_disk 2334 #endif 2335 2336 /* dir.c */ 2337 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2338 struct file *, 2339 struct ext4_dir_entry_2 *, 2340 struct buffer_head *, char *, int, 2341 unsigned int); 2342 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2343 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2344 (de), (bh), (buf), (size), (offset))) 2345 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2346 __u32 minor_hash, 2347 struct ext4_dir_entry_2 *dirent, 2348 struct fscrypt_str *ent_name); 2349 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2350 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2351 struct buffer_head *bh, 2352 void *buf, int buf_size, 2353 struct ext4_filename *fname, 2354 struct ext4_dir_entry_2 **dest_de); 2355 int ext4_insert_dentry(struct inode *dir, 2356 struct inode *inode, 2357 struct ext4_dir_entry_2 *de, 2358 int buf_size, 2359 struct ext4_filename *fname); 2360 static inline void ext4_update_dx_flag(struct inode *inode) 2361 { 2362 if (!ext4_has_feature_dir_index(inode->i_sb)) 2363 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2364 } 2365 static unsigned char ext4_filetype_table[] = { 2366 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2367 }; 2368 2369 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2370 { 2371 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2372 return DT_UNKNOWN; 2373 2374 return ext4_filetype_table[filetype]; 2375 } 2376 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2377 void *buf, int buf_size); 2378 2379 /* fsync.c */ 2380 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2381 2382 /* hash.c */ 2383 extern int ext4fs_dirhash(const char *name, int len, struct 2384 dx_hash_info *hinfo); 2385 2386 /* ialloc.c */ 2387 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2388 const struct qstr *qstr, __u32 goal, 2389 uid_t *owner, int handle_type, 2390 unsigned int line_no, int nblocks); 2391 2392 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2393 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2394 0, 0, 0) 2395 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2396 type, nblocks) \ 2397 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2398 (type), __LINE__, (nblocks)) 2399 2400 2401 extern void ext4_free_inode(handle_t *, struct inode *); 2402 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2403 extern unsigned long ext4_count_free_inodes(struct super_block *); 2404 extern unsigned long ext4_count_dirs(struct super_block *); 2405 extern void ext4_check_inodes_bitmap(struct super_block *); 2406 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2407 extern int ext4_init_inode_table(struct super_block *sb, 2408 ext4_group_t group, int barrier); 2409 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2410 2411 /* mballoc.c */ 2412 extern const struct file_operations ext4_seq_mb_groups_fops; 2413 extern long ext4_mb_stats; 2414 extern long ext4_mb_max_to_scan; 2415 extern int ext4_mb_init(struct super_block *); 2416 extern int ext4_mb_release(struct super_block *); 2417 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2418 struct ext4_allocation_request *, int *); 2419 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2420 extern void ext4_discard_preallocations(struct inode *); 2421 extern int __init ext4_init_mballoc(void); 2422 extern void ext4_exit_mballoc(void); 2423 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2424 struct buffer_head *bh, ext4_fsblk_t block, 2425 unsigned long count, int flags); 2426 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2427 ext4_group_t ngroups); 2428 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2429 ext4_group_t i, struct ext4_group_desc *desc); 2430 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2431 ext4_fsblk_t block, unsigned long count); 2432 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2433 2434 /* inode.c */ 2435 int ext4_inode_is_fast_symlink(struct inode *inode); 2436 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2437 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2438 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2439 struct buffer_head *bh_result, int create); 2440 int ext4_get_block(struct inode *inode, sector_t iblock, 2441 struct buffer_head *bh_result, int create); 2442 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2443 struct buffer_head *bh_result, int create); 2444 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2445 struct buffer_head *bh, int create); 2446 int ext4_walk_page_buffers(handle_t *handle, 2447 struct buffer_head *head, 2448 unsigned from, 2449 unsigned to, 2450 int *partial, 2451 int (*fn)(handle_t *handle, 2452 struct buffer_head *bh)); 2453 int do_journal_get_write_access(handle_t *handle, 2454 struct buffer_head *bh); 2455 #define FALL_BACK_TO_NONDELALLOC 1 2456 #define CONVERT_INLINE_DATA 2 2457 2458 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2459 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2460 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2461 extern int ext4_setattr(struct dentry *, struct iattr *); 2462 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2463 struct kstat *stat); 2464 extern void ext4_evict_inode(struct inode *); 2465 extern void ext4_clear_inode(struct inode *); 2466 extern int ext4_sync_inode(handle_t *, struct inode *); 2467 extern void ext4_dirty_inode(struct inode *, int); 2468 extern int ext4_change_inode_journal_flag(struct inode *, int); 2469 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2470 extern int ext4_inode_attach_jinode(struct inode *inode); 2471 extern int ext4_can_truncate(struct inode *inode); 2472 extern int ext4_truncate(struct inode *); 2473 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2474 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2475 extern void ext4_set_inode_flags(struct inode *); 2476 extern void ext4_get_inode_flags(struct ext4_inode_info *); 2477 extern int ext4_alloc_da_blocks(struct inode *inode); 2478 extern void ext4_set_aops(struct inode *inode); 2479 extern int ext4_writepage_trans_blocks(struct inode *); 2480 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2481 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2482 loff_t lstart, loff_t lend); 2483 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2484 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf); 2485 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2486 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2487 extern void ext4_da_update_reserve_space(struct inode *inode, 2488 int used, int quota_claim); 2489 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2490 ext4_fsblk_t pblk, ext4_lblk_t len); 2491 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk, 2492 unsigned int map_len, 2493 struct extent_status *result); 2494 2495 /* indirect.c */ 2496 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2497 struct ext4_map_blocks *map, int flags); 2498 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2499 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2500 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2501 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2502 ext4_lblk_t start, ext4_lblk_t end); 2503 2504 /* ioctl.c */ 2505 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2506 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2507 2508 /* migrate.c */ 2509 extern int ext4_ext_migrate(struct inode *); 2510 extern int ext4_ind_migrate(struct inode *inode); 2511 2512 /* namei.c */ 2513 extern int ext4_dirent_csum_verify(struct inode *inode, 2514 struct ext4_dir_entry *dirent); 2515 extern int ext4_orphan_add(handle_t *, struct inode *); 2516 extern int ext4_orphan_del(handle_t *, struct inode *); 2517 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2518 __u32 start_minor_hash, __u32 *next_hash); 2519 extern int ext4_search_dir(struct buffer_head *bh, 2520 char *search_buf, 2521 int buf_size, 2522 struct inode *dir, 2523 struct ext4_filename *fname, 2524 const struct qstr *d_name, 2525 unsigned int offset, 2526 struct ext4_dir_entry_2 **res_dir); 2527 extern int ext4_generic_delete_entry(handle_t *handle, 2528 struct inode *dir, 2529 struct ext4_dir_entry_2 *de_del, 2530 struct buffer_head *bh, 2531 void *entry_buf, 2532 int buf_size, 2533 int csum_size); 2534 extern bool ext4_empty_dir(struct inode *inode); 2535 2536 /* resize.c */ 2537 extern int ext4_group_add(struct super_block *sb, 2538 struct ext4_new_group_data *input); 2539 extern int ext4_group_extend(struct super_block *sb, 2540 struct ext4_super_block *es, 2541 ext4_fsblk_t n_blocks_count); 2542 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2543 2544 /* super.c */ 2545 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2546 extern int ext4_calculate_overhead(struct super_block *sb); 2547 extern void ext4_superblock_csum_set(struct super_block *sb); 2548 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2549 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2550 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2551 ext4_group_t ngroup); 2552 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2553 char nbuf[16]); 2554 2555 extern __printf(4, 5) 2556 void __ext4_error(struct super_block *, const char *, unsigned int, 2557 const char *, ...); 2558 extern __printf(5, 6) 2559 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2560 const char *, ...); 2561 extern __printf(5, 6) 2562 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2563 const char *, ...); 2564 extern void __ext4_std_error(struct super_block *, const char *, 2565 unsigned int, int); 2566 extern __printf(4, 5) 2567 void __ext4_abort(struct super_block *, const char *, unsigned int, 2568 const char *, ...); 2569 extern __printf(4, 5) 2570 void __ext4_warning(struct super_block *, const char *, unsigned int, 2571 const char *, ...); 2572 extern __printf(4, 5) 2573 void __ext4_warning_inode(const struct inode *inode, const char *function, 2574 unsigned int line, const char *fmt, ...); 2575 extern __printf(3, 4) 2576 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2577 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2578 const char *, unsigned int, const char *); 2579 extern __printf(7, 8) 2580 void __ext4_grp_locked_error(const char *, unsigned int, 2581 struct super_block *, ext4_group_t, 2582 unsigned long, ext4_fsblk_t, 2583 const char *, ...); 2584 2585 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2586 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2587 2588 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2589 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2590 2591 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2592 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2593 2594 #ifdef CONFIG_PRINTK 2595 2596 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2597 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2598 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2599 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2600 #define ext4_error(sb, fmt, ...) \ 2601 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2602 #define ext4_abort(sb, fmt, ...) \ 2603 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2604 #define ext4_warning(sb, fmt, ...) \ 2605 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2606 #define ext4_warning_inode(inode, fmt, ...) \ 2607 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2608 #define ext4_msg(sb, level, fmt, ...) \ 2609 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2610 #define dump_mmp_msg(sb, mmp, msg) \ 2611 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2612 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2613 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2614 fmt, ##__VA_ARGS__) 2615 2616 #else 2617 2618 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2619 do { \ 2620 no_printk(fmt, ##__VA_ARGS__); \ 2621 __ext4_error_inode(inode, "", 0, block, " "); \ 2622 } while (0) 2623 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2624 do { \ 2625 no_printk(fmt, ##__VA_ARGS__); \ 2626 __ext4_error_file(file, "", 0, block, " "); \ 2627 } while (0) 2628 #define ext4_error(sb, fmt, ...) \ 2629 do { \ 2630 no_printk(fmt, ##__VA_ARGS__); \ 2631 __ext4_error(sb, "", 0, " "); \ 2632 } while (0) 2633 #define ext4_abort(sb, fmt, ...) \ 2634 do { \ 2635 no_printk(fmt, ##__VA_ARGS__); \ 2636 __ext4_abort(sb, "", 0, " "); \ 2637 } while (0) 2638 #define ext4_warning(sb, fmt, ...) \ 2639 do { \ 2640 no_printk(fmt, ##__VA_ARGS__); \ 2641 __ext4_warning(sb, "", 0, " "); \ 2642 } while (0) 2643 #define ext4_warning_inode(inode, fmt, ...) \ 2644 do { \ 2645 no_printk(fmt, ##__VA_ARGS__); \ 2646 __ext4_warning_inode(inode, "", 0, " "); \ 2647 } while (0) 2648 #define ext4_msg(sb, level, fmt, ...) \ 2649 do { \ 2650 no_printk(fmt, ##__VA_ARGS__); \ 2651 __ext4_msg(sb, "", " "); \ 2652 } while (0) 2653 #define dump_mmp_msg(sb, mmp, msg) \ 2654 __dump_mmp_msg(sb, mmp, "", 0, "") 2655 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2656 do { \ 2657 no_printk(fmt, ##__VA_ARGS__); \ 2658 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2659 } while (0) 2660 2661 #endif 2662 2663 extern void ext4_update_dynamic_rev(struct super_block *sb); 2664 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2665 __u32 compat); 2666 extern int ext4_update_rocompat_feature(handle_t *handle, 2667 struct super_block *sb, __u32 rocompat); 2668 extern int ext4_update_incompat_feature(handle_t *handle, 2669 struct super_block *sb, __u32 incompat); 2670 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2671 struct ext4_group_desc *bg); 2672 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2673 struct ext4_group_desc *bg); 2674 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2675 struct ext4_group_desc *bg); 2676 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2677 struct ext4_group_desc *bg); 2678 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2679 struct ext4_group_desc *bg); 2680 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2681 struct ext4_group_desc *bg); 2682 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2683 struct ext4_group_desc *bg); 2684 extern void ext4_block_bitmap_set(struct super_block *sb, 2685 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2686 extern void ext4_inode_bitmap_set(struct super_block *sb, 2687 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2688 extern void ext4_inode_table_set(struct super_block *sb, 2689 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2690 extern void ext4_free_group_clusters_set(struct super_block *sb, 2691 struct ext4_group_desc *bg, 2692 __u32 count); 2693 extern void ext4_free_inodes_set(struct super_block *sb, 2694 struct ext4_group_desc *bg, __u32 count); 2695 extern void ext4_used_dirs_set(struct super_block *sb, 2696 struct ext4_group_desc *bg, __u32 count); 2697 extern void ext4_itable_unused_set(struct super_block *sb, 2698 struct ext4_group_desc *bg, __u32 count); 2699 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2700 struct ext4_group_desc *gdp); 2701 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2702 struct ext4_group_desc *gdp); 2703 extern int ext4_register_li_request(struct super_block *sb, 2704 ext4_group_t first_not_zeroed); 2705 2706 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2707 { 2708 return ext4_has_feature_gdt_csum(sb) || 2709 EXT4_SB(sb)->s_chksum_driver != NULL; 2710 } 2711 2712 static inline int ext4_has_metadata_csum(struct super_block *sb) 2713 { 2714 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2715 !EXT4_SB(sb)->s_chksum_driver); 2716 2717 return (EXT4_SB(sb)->s_chksum_driver != NULL); 2718 } 2719 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2720 { 2721 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2722 le32_to_cpu(es->s_blocks_count_lo); 2723 } 2724 2725 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2726 { 2727 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2728 le32_to_cpu(es->s_r_blocks_count_lo); 2729 } 2730 2731 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2732 { 2733 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2734 le32_to_cpu(es->s_free_blocks_count_lo); 2735 } 2736 2737 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2738 ext4_fsblk_t blk) 2739 { 2740 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2741 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2742 } 2743 2744 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2745 ext4_fsblk_t blk) 2746 { 2747 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2748 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2749 } 2750 2751 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2752 ext4_fsblk_t blk) 2753 { 2754 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2755 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2756 } 2757 2758 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2759 { 2760 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2761 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2762 le32_to_cpu(raw_inode->i_size_lo); 2763 else 2764 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2765 } 2766 2767 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2768 { 2769 raw_inode->i_size_lo = cpu_to_le32(i_size); 2770 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2771 } 2772 2773 static inline 2774 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2775 ext4_group_t group) 2776 { 2777 struct ext4_group_info ***grp_info; 2778 long indexv, indexh; 2779 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2780 grp_info = EXT4_SB(sb)->s_group_info; 2781 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2782 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2783 return grp_info[indexv][indexh]; 2784 } 2785 2786 /* 2787 * Reading s_groups_count requires using smp_rmb() afterwards. See 2788 * the locking protocol documented in the comments of ext4_group_add() 2789 * in resize.c 2790 */ 2791 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2792 { 2793 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2794 2795 smp_rmb(); 2796 return ngroups; 2797 } 2798 2799 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2800 ext4_group_t block_group) 2801 { 2802 return block_group >> sbi->s_log_groups_per_flex; 2803 } 2804 2805 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2806 { 2807 return 1 << sbi->s_log_groups_per_flex; 2808 } 2809 2810 #define ext4_std_error(sb, errno) \ 2811 do { \ 2812 if ((errno)) \ 2813 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2814 } while (0) 2815 2816 #ifdef CONFIG_SMP 2817 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2818 * counters. So we need to make sure we have free clusters more 2819 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2820 */ 2821 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2822 #else 2823 #define EXT4_FREECLUSTERS_WATERMARK 0 2824 #endif 2825 2826 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2827 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2828 { 2829 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2830 !inode_is_locked(inode)); 2831 down_write(&EXT4_I(inode)->i_data_sem); 2832 if (newsize > EXT4_I(inode)->i_disksize) 2833 EXT4_I(inode)->i_disksize = newsize; 2834 up_write(&EXT4_I(inode)->i_data_sem); 2835 } 2836 2837 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2838 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2839 { 2840 int changed = 0; 2841 2842 if (newsize > inode->i_size) { 2843 i_size_write(inode, newsize); 2844 changed = 1; 2845 } 2846 if (newsize > EXT4_I(inode)->i_disksize) { 2847 ext4_update_i_disksize(inode, newsize); 2848 changed |= 2; 2849 } 2850 return changed; 2851 } 2852 2853 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2854 loff_t len); 2855 2856 struct ext4_group_info { 2857 unsigned long bb_state; 2858 struct rb_root bb_free_root; 2859 ext4_grpblk_t bb_first_free; /* first free block */ 2860 ext4_grpblk_t bb_free; /* total free blocks */ 2861 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2862 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2863 struct list_head bb_prealloc_list; 2864 #ifdef DOUBLE_CHECK 2865 void *bb_bitmap; 2866 #endif 2867 struct rw_semaphore alloc_sem; 2868 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2869 * regions, index is order. 2870 * bb_counters[3] = 5 means 2871 * 5 free 8-block regions. */ 2872 }; 2873 2874 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2875 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2876 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2877 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2878 2879 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2880 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2881 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2882 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2883 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2884 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2885 2886 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2887 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2888 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2889 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2890 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2891 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2892 2893 #define EXT4_MAX_CONTENTION 8 2894 #define EXT4_CONTENTION_THRESHOLD 2 2895 2896 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2897 ext4_group_t group) 2898 { 2899 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2900 } 2901 2902 /* 2903 * Returns true if the filesystem is busy enough that attempts to 2904 * access the block group locks has run into contention. 2905 */ 2906 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2907 { 2908 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2909 } 2910 2911 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2912 { 2913 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2914 if (spin_trylock(lock)) 2915 /* 2916 * We're able to grab the lock right away, so drop the 2917 * lock contention counter. 2918 */ 2919 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2920 else { 2921 /* 2922 * The lock is busy, so bump the contention counter, 2923 * and then wait on the spin lock. 2924 */ 2925 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2926 EXT4_MAX_CONTENTION); 2927 spin_lock(lock); 2928 } 2929 } 2930 2931 static inline void ext4_unlock_group(struct super_block *sb, 2932 ext4_group_t group) 2933 { 2934 spin_unlock(ext4_group_lock_ptr(sb, group)); 2935 } 2936 2937 /* 2938 * Block validity checking 2939 */ 2940 #define ext4_check_indirect_blockref(inode, bh) \ 2941 ext4_check_blockref(__func__, __LINE__, inode, \ 2942 (__le32 *)(bh)->b_data, \ 2943 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2944 2945 #define ext4_ind_check_inode(inode) \ 2946 ext4_check_blockref(__func__, __LINE__, inode, \ 2947 EXT4_I(inode)->i_data, \ 2948 EXT4_NDIR_BLOCKS) 2949 2950 /* 2951 * Inodes and files operations 2952 */ 2953 2954 /* dir.c */ 2955 extern const struct file_operations ext4_dir_operations; 2956 2957 /* file.c */ 2958 extern const struct inode_operations ext4_file_inode_operations; 2959 extern const struct file_operations ext4_file_operations; 2960 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2961 2962 /* inline.c */ 2963 extern int ext4_get_max_inline_size(struct inode *inode); 2964 extern int ext4_find_inline_data_nolock(struct inode *inode); 2965 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2966 unsigned int len); 2967 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2968 2969 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2970 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2971 struct inode *inode, 2972 loff_t pos, unsigned len, 2973 unsigned flags, 2974 struct page **pagep); 2975 extern int ext4_write_inline_data_end(struct inode *inode, 2976 loff_t pos, unsigned len, 2977 unsigned copied, 2978 struct page *page); 2979 extern struct buffer_head * 2980 ext4_journalled_write_inline_data(struct inode *inode, 2981 unsigned len, 2982 struct page *page); 2983 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2984 struct inode *inode, 2985 loff_t pos, unsigned len, 2986 unsigned flags, 2987 struct page **pagep, 2988 void **fsdata); 2989 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2990 unsigned len, unsigned copied, 2991 struct page *page); 2992 extern int ext4_try_add_inline_entry(handle_t *handle, 2993 struct ext4_filename *fname, 2994 struct inode *dir, struct inode *inode); 2995 extern int ext4_try_create_inline_dir(handle_t *handle, 2996 struct inode *parent, 2997 struct inode *inode); 2998 extern int ext4_read_inline_dir(struct file *filp, 2999 struct dir_context *ctx, 3000 int *has_inline_data); 3001 extern int htree_inlinedir_to_tree(struct file *dir_file, 3002 struct inode *dir, ext4_lblk_t block, 3003 struct dx_hash_info *hinfo, 3004 __u32 start_hash, __u32 start_minor_hash, 3005 int *has_inline_data); 3006 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3007 struct ext4_filename *fname, 3008 const struct qstr *d_name, 3009 struct ext4_dir_entry_2 **res_dir, 3010 int *has_inline_data); 3011 extern int ext4_delete_inline_entry(handle_t *handle, 3012 struct inode *dir, 3013 struct ext4_dir_entry_2 *de_del, 3014 struct buffer_head *bh, 3015 int *has_inline_data); 3016 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3017 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3018 struct ext4_dir_entry_2 **parent_de, 3019 int *retval); 3020 extern int ext4_inline_data_fiemap(struct inode *inode, 3021 struct fiemap_extent_info *fieinfo, 3022 int *has_inline, __u64 start, __u64 len); 3023 extern int ext4_try_to_evict_inline_data(handle_t *handle, 3024 struct inode *inode, 3025 int needed); 3026 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3027 3028 extern int ext4_convert_inline_data(struct inode *inode); 3029 3030 static inline int ext4_has_inline_data(struct inode *inode) 3031 { 3032 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3033 EXT4_I(inode)->i_inline_off; 3034 } 3035 3036 /* namei.c */ 3037 extern const struct inode_operations ext4_dir_inode_operations; 3038 extern const struct inode_operations ext4_special_inode_operations; 3039 extern struct dentry *ext4_get_parent(struct dentry *child); 3040 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3041 struct ext4_dir_entry_2 *de, 3042 int blocksize, int csum_size, 3043 unsigned int parent_ino, int dotdot_real_len); 3044 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3045 unsigned int blocksize); 3046 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3047 struct inode *inode, 3048 struct buffer_head *bh); 3049 #define S_SHIFT 12 3050 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 3051 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3052 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3053 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3054 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3055 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3056 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3057 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3058 }; 3059 3060 static inline void ext4_set_de_type(struct super_block *sb, 3061 struct ext4_dir_entry_2 *de, 3062 umode_t mode) { 3063 if (ext4_has_feature_filetype(sb)) 3064 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3065 } 3066 3067 /* readpages.c */ 3068 extern int ext4_mpage_readpages(struct address_space *mapping, 3069 struct list_head *pages, struct page *page, 3070 unsigned nr_pages); 3071 3072 /* symlink.c */ 3073 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3074 extern const struct inode_operations ext4_symlink_inode_operations; 3075 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3076 3077 /* sysfs.c */ 3078 extern int ext4_register_sysfs(struct super_block *sb); 3079 extern void ext4_unregister_sysfs(struct super_block *sb); 3080 extern int __init ext4_init_sysfs(void); 3081 extern void ext4_exit_sysfs(void); 3082 3083 /* block_validity */ 3084 extern void ext4_release_system_zone(struct super_block *sb); 3085 extern int ext4_setup_system_zone(struct super_block *sb); 3086 extern int __init ext4_init_system_zone(void); 3087 extern void ext4_exit_system_zone(void); 3088 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3089 ext4_fsblk_t start_blk, 3090 unsigned int count); 3091 extern int ext4_check_blockref(const char *, unsigned int, 3092 struct inode *, __le32 *, unsigned int); 3093 3094 /* extents.c */ 3095 struct ext4_ext_path; 3096 struct ext4_extent; 3097 3098 /* 3099 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3100 * __le32. 3101 */ 3102 #define EXT_MAX_BLOCKS 0xffffffff 3103 3104 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3105 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3106 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3107 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3108 struct ext4_map_blocks *map, int flags); 3109 extern int ext4_ext_truncate(handle_t *, struct inode *); 3110 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3111 ext4_lblk_t end); 3112 extern void ext4_ext_init(struct super_block *); 3113 extern void ext4_ext_release(struct super_block *); 3114 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3115 loff_t len); 3116 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3117 loff_t offset, ssize_t len); 3118 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3119 struct ext4_map_blocks *map, int flags); 3120 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3121 ext4_lblk_t lblocks); 3122 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3123 int num, 3124 struct ext4_ext_path *path); 3125 extern int ext4_can_extents_be_merged(struct inode *inode, 3126 struct ext4_extent *ex1, 3127 struct ext4_extent *ex2); 3128 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3129 struct ext4_ext_path **, 3130 struct ext4_extent *, int); 3131 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3132 struct ext4_ext_path **, 3133 int flags); 3134 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3135 extern int ext4_ext_check_inode(struct inode *inode); 3136 extern int ext4_find_delalloc_range(struct inode *inode, 3137 ext4_lblk_t lblk_start, 3138 ext4_lblk_t lblk_end); 3139 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3140 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3141 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3142 __u64 start, __u64 len); 3143 extern int ext4_ext_precache(struct inode *inode); 3144 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3145 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3146 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3147 struct inode *inode2, ext4_lblk_t lblk1, 3148 ext4_lblk_t lblk2, ext4_lblk_t count, 3149 int mark_unwritten,int *err); 3150 3151 /* move_extent.c */ 3152 extern void ext4_double_down_write_data_sem(struct inode *first, 3153 struct inode *second); 3154 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3155 struct inode *donor_inode); 3156 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3157 __u64 start_orig, __u64 start_donor, 3158 __u64 len, __u64 *moved_len); 3159 3160 /* page-io.c */ 3161 extern int __init ext4_init_pageio(void); 3162 extern void ext4_exit_pageio(void); 3163 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3164 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3165 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3166 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3167 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3168 struct writeback_control *wbc); 3169 extern void ext4_end_io_rsv_work(struct work_struct *work); 3170 extern void ext4_io_submit(struct ext4_io_submit *io); 3171 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3172 struct page *page, 3173 int len, 3174 struct writeback_control *wbc, 3175 bool keep_towrite); 3176 3177 /* mmp.c */ 3178 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3179 3180 /* 3181 * Add new method to test whether block and inode bitmaps are properly 3182 * initialized. With uninit_bg reading the block from disk is not enough 3183 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3184 */ 3185 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3186 3187 static inline int bitmap_uptodate(struct buffer_head *bh) 3188 { 3189 return (buffer_uptodate(bh) && 3190 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3191 } 3192 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3193 { 3194 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3195 } 3196 3197 /* 3198 * Disable DIO read nolock optimization, so new dioreaders will be forced 3199 * to grab i_mutex 3200 */ 3201 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 3202 { 3203 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3204 smp_mb(); 3205 } 3206 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 3207 { 3208 smp_mb(); 3209 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3210 } 3211 3212 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3213 3214 /* For ioend & aio unwritten conversion wait queues */ 3215 #define EXT4_WQ_HASH_SZ 37 3216 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3217 EXT4_WQ_HASH_SZ]) 3218 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3219 3220 #define EXT4_RESIZING 0 3221 extern int ext4_resize_begin(struct super_block *sb); 3222 extern void ext4_resize_end(struct super_block *sb); 3223 3224 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3225 struct ext4_io_end *io_end) 3226 { 3227 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3228 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3229 atomic_inc(&EXT4_I(inode)->i_unwritten); 3230 } 3231 } 3232 3233 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3234 { 3235 struct inode *inode = io_end->inode; 3236 3237 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3238 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3239 /* Wake up anyone waiting on unwritten extent conversion */ 3240 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3241 wake_up_all(ext4_ioend_wq(inode)); 3242 } 3243 } 3244 3245 extern struct iomap_ops ext4_iomap_ops; 3246 3247 #endif /* __KERNEL__ */ 3248 3249 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3250 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3251 3252 #endif /* _EXT4_H */ 3253