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/sched/signal.h> 32 #include <linux/blockgroup_lock.h> 33 #include <linux/percpu_counter.h> 34 #include <linux/ratelimit.h> 35 #include <crypto/hash.h> 36 #ifdef CONFIG_EXT4_FS_ENCRYPTION 37 #include <linux/fscrypt_supp.h> 38 #else 39 #include <linux/fscrypt_notsupp.h> 40 #endif 41 #include <linux/falloc.h> 42 #include <linux/percpu-rwsem.h> 43 #ifdef __KERNEL__ 44 #include <linux/compat.h> 45 #endif 46 47 /* 48 * The fourth extended filesystem constants/structures 49 */ 50 51 /* 52 * with AGGRESSIVE_CHECK allocator runs consistency checks over 53 * structures. these checks slow things down a lot 54 */ 55 #define AGGRESSIVE_CHECK__ 56 57 /* 58 * with DOUBLE_CHECK defined mballoc creates persistent in-core 59 * bitmaps, maintains and uses them to check for double allocations 60 */ 61 #define DOUBLE_CHECK__ 62 63 /* 64 * Define EXT4FS_DEBUG to produce debug messages 65 */ 66 #undef EXT4FS_DEBUG 67 68 /* 69 * Debug code 70 */ 71 #ifdef EXT4FS_DEBUG 72 #define ext4_debug(f, a...) \ 73 do { \ 74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 75 __FILE__, __LINE__, __func__); \ 76 printk(KERN_DEBUG f, ## a); \ 77 } while (0) 78 #else 79 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 80 #endif 81 82 /* 83 * Turn on EXT_DEBUG to get lots of info about extents operations. 84 */ 85 #define EXT_DEBUG__ 86 #ifdef EXT_DEBUG 87 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 88 #else 89 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 90 #endif 91 92 /* data type for block offset of block group */ 93 typedef int ext4_grpblk_t; 94 95 /* data type for filesystem-wide blocks number */ 96 typedef unsigned long long ext4_fsblk_t; 97 98 /* data type for file logical block number */ 99 typedef __u32 ext4_lblk_t; 100 101 /* data type for block group number */ 102 typedef unsigned int ext4_group_t; 103 104 enum SHIFT_DIRECTION { 105 SHIFT_LEFT = 0, 106 SHIFT_RIGHT, 107 }; 108 109 /* 110 * Flags used in mballoc's allocation_context flags field. 111 * 112 * Also used to show what's going on for debugging purposes when the 113 * flag field is exported via the traceport interface 114 */ 115 116 /* prefer goal again. length */ 117 #define EXT4_MB_HINT_MERGE 0x0001 118 /* blocks already reserved */ 119 #define EXT4_MB_HINT_RESERVED 0x0002 120 /* metadata is being allocated */ 121 #define EXT4_MB_HINT_METADATA 0x0004 122 /* first blocks in the file */ 123 #define EXT4_MB_HINT_FIRST 0x0008 124 /* search for the best chunk */ 125 #define EXT4_MB_HINT_BEST 0x0010 126 /* data is being allocated */ 127 #define EXT4_MB_HINT_DATA 0x0020 128 /* don't preallocate (for tails) */ 129 #define EXT4_MB_HINT_NOPREALLOC 0x0040 130 /* allocate for locality group */ 131 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 132 /* allocate goal blocks or none */ 133 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 134 /* goal is meaningful */ 135 #define EXT4_MB_HINT_TRY_GOAL 0x0200 136 /* blocks already pre-reserved by delayed allocation */ 137 #define EXT4_MB_DELALLOC_RESERVED 0x0400 138 /* We are doing stream allocation */ 139 #define EXT4_MB_STREAM_ALLOC 0x0800 140 /* Use reserved root blocks if needed */ 141 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000 142 /* Use blocks from reserved pool */ 143 #define EXT4_MB_USE_RESERVED 0x2000 144 145 struct ext4_allocation_request { 146 /* target inode for block we're allocating */ 147 struct inode *inode; 148 /* how many blocks we want to allocate */ 149 unsigned int len; 150 /* logical block in target inode */ 151 ext4_lblk_t logical; 152 /* the closest logical allocated block to the left */ 153 ext4_lblk_t lleft; 154 /* the closest logical allocated block to the right */ 155 ext4_lblk_t lright; 156 /* phys. target (a hint) */ 157 ext4_fsblk_t goal; 158 /* phys. block for the closest logical allocated block to the left */ 159 ext4_fsblk_t pleft; 160 /* phys. block for the closest logical allocated block to the right */ 161 ext4_fsblk_t pright; 162 /* flags. see above EXT4_MB_HINT_* */ 163 unsigned int flags; 164 }; 165 166 /* 167 * Logical to physical block mapping, used by ext4_map_blocks() 168 * 169 * This structure is used to pass requests into ext4_map_blocks() as 170 * well as to store the information returned by ext4_map_blocks(). It 171 * takes less room on the stack than a struct buffer_head. 172 */ 173 #define EXT4_MAP_NEW (1 << BH_New) 174 #define EXT4_MAP_MAPPED (1 << BH_Mapped) 175 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 176 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 177 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY) 179 180 struct ext4_map_blocks { 181 ext4_fsblk_t m_pblk; 182 ext4_lblk_t m_lblk; 183 unsigned int m_len; 184 unsigned int m_flags; 185 }; 186 187 /* 188 * Flags for ext4_io_end->flags 189 */ 190 #define EXT4_IO_END_UNWRITTEN 0x0001 191 192 /* 193 * For converting unwritten extents on a work queue. 'handle' is used for 194 * buffered writeback. 195 */ 196 typedef struct ext4_io_end { 197 struct list_head list; /* per-file finished IO list */ 198 handle_t *handle; /* handle reserved for extent 199 * conversion */ 200 struct inode *inode; /* file being written to */ 201 struct bio *bio; /* Linked list of completed 202 * bios covering the extent */ 203 unsigned int flag; /* unwritten or not */ 204 atomic_t count; /* reference counter */ 205 loff_t offset; /* offset in the file */ 206 ssize_t size; /* size of the extent */ 207 } ext4_io_end_t; 208 209 struct ext4_io_submit { 210 struct writeback_control *io_wbc; 211 struct bio *io_bio; 212 ext4_io_end_t *io_end; 213 sector_t io_next_block; 214 }; 215 216 /* 217 * Special inodes numbers 218 */ 219 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 220 #define EXT4_ROOT_INO 2 /* Root inode */ 221 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 222 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 223 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 224 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 225 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 226 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 227 228 /* First non-reserved inode for old ext4 filesystems */ 229 #define EXT4_GOOD_OLD_FIRST_INO 11 230 231 /* 232 * Maximal count of links to a file 233 */ 234 #define EXT4_LINK_MAX 65000 235 236 /* 237 * Macro-instructions used to manage several block sizes 238 */ 239 #define EXT4_MIN_BLOCK_SIZE 1024 240 #define EXT4_MAX_BLOCK_SIZE 65536 241 #define EXT4_MIN_BLOCK_LOG_SIZE 10 242 #define EXT4_MAX_BLOCK_LOG_SIZE 16 243 #define EXT4_MAX_CLUSTER_LOG_SIZE 30 244 #ifdef __KERNEL__ 245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 246 #else 247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 248 #endif 249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 251 EXT4_SB(s)->s_cluster_bits) 252 #ifdef __KERNEL__ 253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 255 #else 256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 257 #endif 258 #ifdef __KERNEL__ 259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 262 #else 263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 264 EXT4_GOOD_OLD_INODE_SIZE : \ 265 (s)->s_inode_size) 266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 267 EXT4_GOOD_OLD_FIRST_INO : \ 268 (s)->s_first_ino) 269 #endif 270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 271 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 273 blkbits)) 274 275 /* Translate a block number to a cluster number */ 276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 277 /* Translate a cluster number to a block number */ 278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 279 /* Translate # of blks to # of clusters */ 280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 281 (sbi)->s_cluster_bits) 282 /* Mask out the low bits to get the starting block of the cluster */ 283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 287 /* Get the cluster offset */ 288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 292 293 /* 294 * Structure of a blocks group descriptor 295 */ 296 struct ext4_group_desc 297 { 298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 300 __le32 bg_inode_table_lo; /* Inodes table block */ 301 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 302 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 303 __le16 bg_used_dirs_count_lo; /* Directories count */ 304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 308 __le16 bg_itable_unused_lo; /* Unused inodes count */ 309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 312 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 320 __u32 bg_reserved; 321 }; 322 323 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 325 sizeof(__le16)) 326 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 328 sizeof(__le16)) 329 330 /* 331 * Structure of a flex block group info 332 */ 333 334 struct flex_groups { 335 atomic64_t free_clusters; 336 atomic_t free_inodes; 337 atomic_t used_dirs; 338 }; 339 340 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 341 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 342 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 343 344 /* 345 * Macro-instructions used to manage group descriptors 346 */ 347 #define EXT4_MIN_DESC_SIZE 32 348 #define EXT4_MIN_DESC_SIZE_64BIT 64 349 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 350 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 351 #ifdef __KERNEL__ 352 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 353 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 355 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 356 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 357 #else 358 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 359 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 360 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 361 #endif 362 363 /* 364 * Constants relative to the data blocks 365 */ 366 #define EXT4_NDIR_BLOCKS 12 367 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 368 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 369 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 370 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 371 372 /* 373 * Inode flags 374 */ 375 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 376 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 377 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 378 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 379 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 380 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 381 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 382 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 383 /* Reserved for compression usage... */ 384 #define EXT4_DIRTY_FL 0x00000100 385 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 386 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 387 /* nb: was previously EXT2_ECOMPR_FL */ 388 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 389 /* End compression flags --- maybe not all used */ 390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 401 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 402 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 403 404 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */ 405 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 406 407 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */ 408 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \ 409 EXT4_IMMUTABLE_FL | \ 410 EXT4_APPEND_FL | \ 411 EXT4_NODUMP_FL | \ 412 EXT4_NOATIME_FL | \ 413 EXT4_PROJINHERIT_FL) 414 415 /* Flags that should be inherited by new inodes from their parent. */ 416 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 417 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 418 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 419 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 420 EXT4_PROJINHERIT_FL) 421 422 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 423 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 424 425 /* Flags that are appropriate for non-directories/regular files. */ 426 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 427 428 /* Mask out flags that are inappropriate for the given type of inode. */ 429 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 430 { 431 if (S_ISDIR(mode)) 432 return flags; 433 else if (S_ISREG(mode)) 434 return flags & EXT4_REG_FLMASK; 435 else 436 return flags & EXT4_OTHER_FLMASK; 437 } 438 439 /* 440 * Inode flags used for atomic set/get 441 */ 442 enum { 443 EXT4_INODE_SECRM = 0, /* Secure deletion */ 444 EXT4_INODE_UNRM = 1, /* Undelete */ 445 EXT4_INODE_COMPR = 2, /* Compress file */ 446 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 447 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 448 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 449 EXT4_INODE_NODUMP = 6, /* do not dump file */ 450 EXT4_INODE_NOATIME = 7, /* do not update atime */ 451 /* Reserved for compression usage... */ 452 EXT4_INODE_DIRTY = 8, 453 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 454 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 455 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 456 /* End compression flags --- maybe not all used */ 457 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 458 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 459 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 460 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 461 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 462 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 463 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 464 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 465 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 466 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 467 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 468 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 469 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 470 }; 471 472 /* 473 * Since it's pretty easy to mix up bit numbers and hex values, we use a 474 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 475 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 476 * any extra space in the compiled kernel image, otherwise, the build will fail. 477 * It's important that these values are the same, since we are using 478 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 479 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 480 * values found in ext2, ext3 and ext4 filesystems, and of course the values 481 * defined in e2fsprogs. 482 * 483 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 484 */ 485 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 486 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 487 488 static inline void ext4_check_flag_values(void) 489 { 490 CHECK_FLAG_VALUE(SECRM); 491 CHECK_FLAG_VALUE(UNRM); 492 CHECK_FLAG_VALUE(COMPR); 493 CHECK_FLAG_VALUE(SYNC); 494 CHECK_FLAG_VALUE(IMMUTABLE); 495 CHECK_FLAG_VALUE(APPEND); 496 CHECK_FLAG_VALUE(NODUMP); 497 CHECK_FLAG_VALUE(NOATIME); 498 CHECK_FLAG_VALUE(DIRTY); 499 CHECK_FLAG_VALUE(COMPRBLK); 500 CHECK_FLAG_VALUE(NOCOMPR); 501 CHECK_FLAG_VALUE(ENCRYPT); 502 CHECK_FLAG_VALUE(INDEX); 503 CHECK_FLAG_VALUE(IMAGIC); 504 CHECK_FLAG_VALUE(JOURNAL_DATA); 505 CHECK_FLAG_VALUE(NOTAIL); 506 CHECK_FLAG_VALUE(DIRSYNC); 507 CHECK_FLAG_VALUE(TOPDIR); 508 CHECK_FLAG_VALUE(HUGE_FILE); 509 CHECK_FLAG_VALUE(EXTENTS); 510 CHECK_FLAG_VALUE(EA_INODE); 511 CHECK_FLAG_VALUE(EOFBLOCKS); 512 CHECK_FLAG_VALUE(INLINE_DATA); 513 CHECK_FLAG_VALUE(PROJINHERIT); 514 CHECK_FLAG_VALUE(RESERVED); 515 } 516 517 /* Used to pass group descriptor data when online resize is done */ 518 struct ext4_new_group_input { 519 __u32 group; /* Group number for this data */ 520 __u64 block_bitmap; /* Absolute block number of block bitmap */ 521 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 522 __u64 inode_table; /* Absolute block number of inode table start */ 523 __u32 blocks_count; /* Total number of blocks in this group */ 524 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 525 __u16 unused; 526 }; 527 528 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 529 struct compat_ext4_new_group_input { 530 u32 group; 531 compat_u64 block_bitmap; 532 compat_u64 inode_bitmap; 533 compat_u64 inode_table; 534 u32 blocks_count; 535 u16 reserved_blocks; 536 u16 unused; 537 }; 538 #endif 539 540 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 541 struct ext4_new_group_data { 542 __u32 group; 543 __u64 block_bitmap; 544 __u64 inode_bitmap; 545 __u64 inode_table; 546 __u32 blocks_count; 547 __u16 reserved_blocks; 548 __u16 mdata_blocks; 549 __u32 free_clusters_count; 550 }; 551 552 /* Indexes used to index group tables in ext4_new_group_data */ 553 enum { 554 BLOCK_BITMAP = 0, /* block bitmap */ 555 INODE_BITMAP, /* inode bitmap */ 556 INODE_TABLE, /* inode tables */ 557 GROUP_TABLE_COUNT, 558 }; 559 560 /* 561 * Flags used by ext4_map_blocks() 562 */ 563 /* Allocate any needed blocks and/or convert an unwritten 564 extent to be an initialized ext4 */ 565 #define EXT4_GET_BLOCKS_CREATE 0x0001 566 /* Request the creation of an unwritten extent */ 567 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 568 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 569 EXT4_GET_BLOCKS_CREATE) 570 /* Caller is from the delayed allocation writeout path 571 * finally doing the actual allocation of delayed blocks */ 572 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 573 /* caller is from the direct IO path, request to creation of an 574 unwritten extents if not allocated, split the unwritten 575 extent if blocks has been preallocated already*/ 576 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 577 #define EXT4_GET_BLOCKS_CONVERT 0x0010 578 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 579 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 580 /* Convert extent to initialized after IO complete */ 581 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 582 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 583 /* Eventual metadata allocation (due to growing extent tree) 584 * should not fail, so try to use reserved blocks for that.*/ 585 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 586 /* Don't normalize allocation size (used for fallocate) */ 587 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 588 /* Request will not result in inode size update (user for fallocate) */ 589 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 590 /* Convert written extents to unwritten */ 591 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 592 /* Write zeros to newly created written extents */ 593 #define EXT4_GET_BLOCKS_ZERO 0x0200 594 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 595 EXT4_GET_BLOCKS_ZERO) 596 /* Caller will submit data before dropping transaction handle. This 597 * allows jbd2 to avoid submitting data before commit. */ 598 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 599 600 /* 601 * The bit position of these flags must not overlap with any of the 602 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 603 * read_extent_tree_block(), ext4_split_extent_at(), 604 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 605 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 606 * caching the extents when reading from the extent tree while a 607 * truncate or punch hole operation is in progress. 608 */ 609 #define EXT4_EX_NOCACHE 0x40000000 610 #define EXT4_EX_FORCE_CACHE 0x20000000 611 612 /* 613 * Flags used by ext4_free_blocks 614 */ 615 #define EXT4_FREE_BLOCKS_METADATA 0x0001 616 #define EXT4_FREE_BLOCKS_FORGET 0x0002 617 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 618 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 619 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 620 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 621 622 /* 623 * ioctl commands 624 */ 625 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 626 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 627 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 628 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 629 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 630 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 631 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 632 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 633 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 634 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 635 #define EXT4_IOC_MIGRATE _IO('f', 9) 636 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 637 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 638 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 639 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 640 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 641 #define EXT4_IOC_SWAP_BOOT _IO('f', 17) 642 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 643 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 644 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 645 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 646 647 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR 648 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR 649 650 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32) 651 652 /* 653 * Flags for going down operation 654 */ 655 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */ 656 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */ 657 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */ 658 659 660 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 661 /* 662 * ioctl commands in 32 bit emulation 663 */ 664 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 665 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 666 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 667 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 668 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 669 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 670 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 671 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 672 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 673 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 674 #endif 675 676 /* Max physical block we can address w/o extents */ 677 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 678 679 /* 680 * Structure of an inode on the disk 681 */ 682 struct ext4_inode { 683 __le16 i_mode; /* File mode */ 684 __le16 i_uid; /* Low 16 bits of Owner Uid */ 685 __le32 i_size_lo; /* Size in bytes */ 686 __le32 i_atime; /* Access time */ 687 __le32 i_ctime; /* Inode Change time */ 688 __le32 i_mtime; /* Modification time */ 689 __le32 i_dtime; /* Deletion Time */ 690 __le16 i_gid; /* Low 16 bits of Group Id */ 691 __le16 i_links_count; /* Links count */ 692 __le32 i_blocks_lo; /* Blocks count */ 693 __le32 i_flags; /* File flags */ 694 union { 695 struct { 696 __le32 l_i_version; 697 } linux1; 698 struct { 699 __u32 h_i_translator; 700 } hurd1; 701 struct { 702 __u32 m_i_reserved1; 703 } masix1; 704 } osd1; /* OS dependent 1 */ 705 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 706 __le32 i_generation; /* File version (for NFS) */ 707 __le32 i_file_acl_lo; /* File ACL */ 708 __le32 i_size_high; 709 __le32 i_obso_faddr; /* Obsoleted fragment address */ 710 union { 711 struct { 712 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 713 __le16 l_i_file_acl_high; 714 __le16 l_i_uid_high; /* these 2 fields */ 715 __le16 l_i_gid_high; /* were reserved2[0] */ 716 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 717 __le16 l_i_reserved; 718 } linux2; 719 struct { 720 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 721 __u16 h_i_mode_high; 722 __u16 h_i_uid_high; 723 __u16 h_i_gid_high; 724 __u32 h_i_author; 725 } hurd2; 726 struct { 727 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 728 __le16 m_i_file_acl_high; 729 __u32 m_i_reserved2[2]; 730 } masix2; 731 } osd2; /* OS dependent 2 */ 732 __le16 i_extra_isize; 733 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 734 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 735 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 736 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 737 __le32 i_crtime; /* File Creation time */ 738 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 739 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 740 __le32 i_projid; /* Project ID */ 741 }; 742 743 struct move_extent { 744 __u32 reserved; /* should be zero */ 745 __u32 donor_fd; /* donor file descriptor */ 746 __u64 orig_start; /* logical start offset in block for orig */ 747 __u64 donor_start; /* logical start offset in block for donor */ 748 __u64 len; /* block length to be moved */ 749 __u64 moved_len; /* moved block length */ 750 }; 751 752 #define EXT4_EPOCH_BITS 2 753 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 754 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 755 756 /* 757 * Extended fields will fit into an inode if the filesystem was formatted 758 * with large inodes (-I 256 or larger) and there are not currently any EAs 759 * consuming all of the available space. For new inodes we always reserve 760 * enough space for the kernel's known extended fields, but for inodes 761 * created with an old kernel this might not have been the case. None of 762 * the extended inode fields is critical for correct filesystem operation. 763 * This macro checks if a certain field fits in the inode. Note that 764 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 765 */ 766 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 767 ((offsetof(typeof(*ext4_inode), field) + \ 768 sizeof((ext4_inode)->field)) \ 769 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 770 (einode)->i_extra_isize)) \ 771 772 /* 773 * We use an encoding that preserves the times for extra epoch "00": 774 * 775 * extra msb of adjust for signed 776 * epoch 32-bit 32-bit tv_sec to 777 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 778 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 779 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 780 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 781 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 782 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 783 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 784 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 785 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 786 * 787 * Note that previous versions of the kernel on 64-bit systems would 788 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 789 * 1970. e2fsck will correct this, assuming that it is run on the 790 * affected filesystem before 2242. 791 */ 792 793 static inline __le32 ext4_encode_extra_time(struct timespec *time) 794 { 795 u32 extra = sizeof(time->tv_sec) > 4 ? 796 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0; 797 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS)); 798 } 799 800 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 801 { 802 if (unlikely(sizeof(time->tv_sec) > 4 && 803 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) { 804 805 #if 1 806 /* Handle legacy encoding of pre-1970 dates with epoch 807 * bits 1,1. (This backwards compatibility may be removed 808 * at the discretion of the ext4 developers.) 809 */ 810 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK; 811 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0) 812 extra_bits = 0; 813 time->tv_sec += extra_bits << 32; 814 #else 815 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 816 #endif 817 } 818 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 819 } 820 821 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 822 do { \ 823 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 824 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 825 (raw_inode)->xtime ## _extra = \ 826 ext4_encode_extra_time(&(inode)->xtime); \ 827 } while (0) 828 829 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 830 do { \ 831 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 832 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 833 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 834 (raw_inode)->xtime ## _extra = \ 835 ext4_encode_extra_time(&(einode)->xtime); \ 836 } while (0) 837 838 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 839 do { \ 840 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 841 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 842 ext4_decode_extra_time(&(inode)->xtime, \ 843 raw_inode->xtime ## _extra); \ 844 else \ 845 (inode)->xtime.tv_nsec = 0; \ 846 } while (0) 847 848 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 849 do { \ 850 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 851 (einode)->xtime.tv_sec = \ 852 (signed)le32_to_cpu((raw_inode)->xtime); \ 853 else \ 854 (einode)->xtime.tv_sec = 0; \ 855 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 856 ext4_decode_extra_time(&(einode)->xtime, \ 857 raw_inode->xtime ## _extra); \ 858 else \ 859 (einode)->xtime.tv_nsec = 0; \ 860 } while (0) 861 862 #define i_disk_version osd1.linux1.l_i_version 863 864 #if defined(__KERNEL__) || defined(__linux__) 865 #define i_reserved1 osd1.linux1.l_i_reserved1 866 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 867 #define i_blocks_high osd2.linux2.l_i_blocks_high 868 #define i_uid_low i_uid 869 #define i_gid_low i_gid 870 #define i_uid_high osd2.linux2.l_i_uid_high 871 #define i_gid_high osd2.linux2.l_i_gid_high 872 #define i_checksum_lo osd2.linux2.l_i_checksum_lo 873 874 #elif defined(__GNU__) 875 876 #define i_translator osd1.hurd1.h_i_translator 877 #define i_uid_high osd2.hurd2.h_i_uid_high 878 #define i_gid_high osd2.hurd2.h_i_gid_high 879 #define i_author osd2.hurd2.h_i_author 880 881 #elif defined(__masix__) 882 883 #define i_reserved1 osd1.masix1.m_i_reserved1 884 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 885 #define i_reserved2 osd2.masix2.m_i_reserved2 886 887 #endif /* defined(__KERNEL__) || defined(__linux__) */ 888 889 #include "extents_status.h" 890 891 /* 892 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 893 * 894 * These are needed to avoid lockdep false positives when we need to 895 * allocate blocks to the quota inode during ext4_map_blocks(), while 896 * holding i_data_sem for a normal (non-quota) inode. Since we don't 897 * do quota tracking for the quota inode, this avoids deadlock (as 898 * well as infinite recursion, since it isn't turtles all the way 899 * down...) 900 * 901 * I_DATA_SEM_NORMAL - Used for most inodes 902 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 903 * where the second inode has larger inode number 904 * than the first 905 * I_DATA_SEM_QUOTA - Used for quota inodes only 906 */ 907 enum { 908 I_DATA_SEM_NORMAL = 0, 909 I_DATA_SEM_OTHER, 910 I_DATA_SEM_QUOTA, 911 }; 912 913 914 /* 915 * fourth extended file system inode data in memory 916 */ 917 struct ext4_inode_info { 918 __le32 i_data[15]; /* unconverted */ 919 __u32 i_dtime; 920 ext4_fsblk_t i_file_acl; 921 922 /* 923 * i_block_group is the number of the block group which contains 924 * this file's inode. Constant across the lifetime of the inode, 925 * it is used for making block allocation decisions - we try to 926 * place a file's data blocks near its inode block, and new inodes 927 * near to their parent directory's inode. 928 */ 929 ext4_group_t i_block_group; 930 ext4_lblk_t i_dir_start_lookup; 931 #if (BITS_PER_LONG < 64) 932 unsigned long i_state_flags; /* Dynamic state flags */ 933 #endif 934 unsigned long i_flags; 935 936 /* 937 * Extended attributes can be read independently of the main file 938 * data. Taking i_mutex even when reading would cause contention 939 * between readers of EAs and writers of regular file data, so 940 * instead we synchronize on xattr_sem when reading or changing 941 * EAs. 942 */ 943 struct rw_semaphore xattr_sem; 944 945 struct list_head i_orphan; /* unlinked but open inodes */ 946 947 /* 948 * i_disksize keeps track of what the inode size is ON DISK, not 949 * in memory. During truncate, i_size is set to the new size by 950 * the VFS prior to calling ext4_truncate(), but the filesystem won't 951 * set i_disksize to 0 until the truncate is actually under way. 952 * 953 * The intent is that i_disksize always represents the blocks which 954 * are used by this file. This allows recovery to restart truncate 955 * on orphans if we crash during truncate. We actually write i_disksize 956 * into the on-disk inode when writing inodes out, instead of i_size. 957 * 958 * The only time when i_disksize and i_size may be different is when 959 * a truncate is in progress. The only things which change i_disksize 960 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 961 */ 962 loff_t i_disksize; 963 964 /* 965 * i_data_sem is for serialising ext4_truncate() against 966 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 967 * data tree are chopped off during truncate. We can't do that in 968 * ext4 because whenever we perform intermediate commits during 969 * truncate, the inode and all the metadata blocks *must* be in a 970 * consistent state which allows truncation of the orphans to restart 971 * during recovery. Hence we must fix the get_block-vs-truncate race 972 * by other means, so we have i_data_sem. 973 */ 974 struct rw_semaphore i_data_sem; 975 /* 976 * i_mmap_sem is for serializing page faults with truncate / punch hole 977 * operations. We have to make sure that new page cannot be faulted in 978 * a section of the inode that is being punched. We cannot easily use 979 * i_data_sem for this since we need protection for the whole punch 980 * operation and i_data_sem ranks below transaction start so we have 981 * to occasionally drop it. 982 */ 983 struct rw_semaphore i_mmap_sem; 984 struct inode vfs_inode; 985 struct jbd2_inode *jinode; 986 987 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 988 989 /* 990 * File creation time. Its function is same as that of 991 * struct timespec i_{a,c,m}time in the generic inode. 992 */ 993 struct timespec i_crtime; 994 995 /* mballoc */ 996 struct list_head i_prealloc_list; 997 spinlock_t i_prealloc_lock; 998 999 /* extents status tree */ 1000 struct ext4_es_tree i_es_tree; 1001 rwlock_t i_es_lock; 1002 struct list_head i_es_list; 1003 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1004 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1005 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1006 extents to shrink. Protected by 1007 i_es_lock */ 1008 1009 /* ialloc */ 1010 ext4_group_t i_last_alloc_group; 1011 1012 /* allocation reservation info for delalloc */ 1013 /* In case of bigalloc, this refer to clusters rather than blocks */ 1014 unsigned int i_reserved_data_blocks; 1015 ext4_lblk_t i_da_metadata_calc_last_lblock; 1016 int i_da_metadata_calc_len; 1017 1018 /* on-disk additional length */ 1019 __u16 i_extra_isize; 1020 1021 /* Indicate the inline data space. */ 1022 u16 i_inline_off; 1023 u16 i_inline_size; 1024 1025 #ifdef CONFIG_QUOTA 1026 /* quota space reservation, managed internally by quota code */ 1027 qsize_t i_reserved_quota; 1028 #endif 1029 1030 /* Lock protecting lists below */ 1031 spinlock_t i_completed_io_lock; 1032 /* 1033 * Completed IOs that need unwritten extents handling and have 1034 * transaction reserved 1035 */ 1036 struct list_head i_rsv_conversion_list; 1037 struct work_struct i_rsv_conversion_work; 1038 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 1039 1040 spinlock_t i_block_reservation_lock; 1041 1042 /* 1043 * Transactions that contain inode's metadata needed to complete 1044 * fsync and fdatasync, respectively. 1045 */ 1046 tid_t i_sync_tid; 1047 tid_t i_datasync_tid; 1048 1049 #ifdef CONFIG_QUOTA 1050 struct dquot *i_dquot[MAXQUOTAS]; 1051 #endif 1052 1053 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1054 __u32 i_csum_seed; 1055 1056 kprojid_t i_projid; 1057 }; 1058 1059 /* 1060 * File system states 1061 */ 1062 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1063 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1064 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1065 1066 /* 1067 * Misc. filesystem flags 1068 */ 1069 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1070 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1071 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1072 1073 /* 1074 * Mount flags set via mount options or defaults 1075 */ 1076 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1077 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1078 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1079 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1080 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1081 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1082 #define EXT4_MOUNT_ERRORS_MASK 0x00070 1083 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1084 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1085 #ifdef CONFIG_FS_DAX 1086 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */ 1087 #else 1088 #define EXT4_MOUNT_DAX 0 1089 #endif 1090 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1091 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1092 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1093 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1094 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1095 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1096 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1097 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1098 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1099 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1100 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1101 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1102 * enable enforcement for hidden 1103 * quota files */ 1104 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1105 * enforcement for hidden quota 1106 * files */ 1107 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1108 * enforcement */ 1109 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1110 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1111 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1112 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1113 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1114 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1115 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1116 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1117 1118 /* 1119 * Mount flags set either automatically (could not be set by mount option) 1120 * based on per file system feature or property or in special cases such as 1121 * distinguishing between explicit mount option definition and default. 1122 */ 1123 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1124 specified delalloc */ 1125 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1126 size of blocksize * 8 1127 blocks */ 1128 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1129 file systems */ 1130 1131 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1132 specified journal checksum */ 1133 1134 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1135 ~EXT4_MOUNT_##opt 1136 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1137 EXT4_MOUNT_##opt 1138 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1139 EXT4_MOUNT_##opt) 1140 1141 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1142 ~EXT4_MOUNT2_##opt 1143 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1144 EXT4_MOUNT2_##opt 1145 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1146 EXT4_MOUNT2_##opt) 1147 1148 #define ext4_test_and_set_bit __test_and_set_bit_le 1149 #define ext4_set_bit __set_bit_le 1150 #define ext4_set_bit_atomic ext2_set_bit_atomic 1151 #define ext4_test_and_clear_bit __test_and_clear_bit_le 1152 #define ext4_clear_bit __clear_bit_le 1153 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 1154 #define ext4_test_bit test_bit_le 1155 #define ext4_find_next_zero_bit find_next_zero_bit_le 1156 #define ext4_find_next_bit find_next_bit_le 1157 1158 extern void ext4_set_bits(void *bm, int cur, int len); 1159 1160 /* 1161 * Maximal mount counts between two filesystem checks 1162 */ 1163 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1164 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1165 1166 /* 1167 * Behaviour when detecting errors 1168 */ 1169 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1170 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1171 #define EXT4_ERRORS_PANIC 3 /* Panic */ 1172 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1173 1174 /* Metadata checksum algorithm codes */ 1175 #define EXT4_CRC32C_CHKSUM 1 1176 1177 /* 1178 * Structure of the super block 1179 */ 1180 struct ext4_super_block { 1181 /*00*/ __le32 s_inodes_count; /* Inodes count */ 1182 __le32 s_blocks_count_lo; /* Blocks count */ 1183 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1184 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1185 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1186 __le32 s_first_data_block; /* First Data Block */ 1187 __le32 s_log_block_size; /* Block size */ 1188 __le32 s_log_cluster_size; /* Allocation cluster size */ 1189 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1190 __le32 s_clusters_per_group; /* # Clusters per group */ 1191 __le32 s_inodes_per_group; /* # Inodes per group */ 1192 __le32 s_mtime; /* Mount time */ 1193 /*30*/ __le32 s_wtime; /* Write time */ 1194 __le16 s_mnt_count; /* Mount count */ 1195 __le16 s_max_mnt_count; /* Maximal mount count */ 1196 __le16 s_magic; /* Magic signature */ 1197 __le16 s_state; /* File system state */ 1198 __le16 s_errors; /* Behaviour when detecting errors */ 1199 __le16 s_minor_rev_level; /* minor revision level */ 1200 /*40*/ __le32 s_lastcheck; /* time of last check */ 1201 __le32 s_checkinterval; /* max. time between checks */ 1202 __le32 s_creator_os; /* OS */ 1203 __le32 s_rev_level; /* Revision level */ 1204 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1205 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1206 /* 1207 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1208 * 1209 * Note: the difference between the compatible feature set and 1210 * the incompatible feature set is that if there is a bit set 1211 * in the incompatible feature set that the kernel doesn't 1212 * know about, it should refuse to mount the filesystem. 1213 * 1214 * e2fsck's requirements are more strict; if it doesn't know 1215 * about a feature in either the compatible or incompatible 1216 * feature set, it must abort and not try to meddle with 1217 * things it doesn't understand... 1218 */ 1219 __le32 s_first_ino; /* First non-reserved inode */ 1220 __le16 s_inode_size; /* size of inode structure */ 1221 __le16 s_block_group_nr; /* block group # of this superblock */ 1222 __le32 s_feature_compat; /* compatible feature set */ 1223 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1224 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1225 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1226 /*78*/ char s_volume_name[16]; /* volume name */ 1227 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1228 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1229 /* 1230 * Performance hints. Directory preallocation should only 1231 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1232 */ 1233 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1234 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1235 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1236 /* 1237 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1238 */ 1239 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1240 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1241 __le32 s_journal_dev; /* device number of journal file */ 1242 __le32 s_last_orphan; /* start of list of inodes to delete */ 1243 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1244 __u8 s_def_hash_version; /* Default hash version to use */ 1245 __u8 s_jnl_backup_type; 1246 __le16 s_desc_size; /* size of group descriptor */ 1247 /*100*/ __le32 s_default_mount_opts; 1248 __le32 s_first_meta_bg; /* First metablock block group */ 1249 __le32 s_mkfs_time; /* When the filesystem was created */ 1250 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1251 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1252 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1253 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1254 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1255 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1256 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1257 __le32 s_flags; /* Miscellaneous flags */ 1258 __le16 s_raid_stride; /* RAID stride */ 1259 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1260 __le64 s_mmp_block; /* Block for multi-mount protection */ 1261 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1262 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1263 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1264 __u8 s_encryption_level; /* versioning level for encryption */ 1265 __u8 s_reserved_pad; /* Padding to next 32bits */ 1266 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1267 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1268 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1269 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1270 snapshot's future use */ 1271 __le32 s_snapshot_list; /* inode number of the head of the 1272 on-disk snapshot list */ 1273 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1274 __le32 s_error_count; /* number of fs errors */ 1275 __le32 s_first_error_time; /* first time an error happened */ 1276 __le32 s_first_error_ino; /* inode involved in first error */ 1277 __le64 s_first_error_block; /* block involved of first error */ 1278 __u8 s_first_error_func[32]; /* function where the error happened */ 1279 __le32 s_first_error_line; /* line number where error happened */ 1280 __le32 s_last_error_time; /* most recent time of an error */ 1281 __le32 s_last_error_ino; /* inode involved in last error */ 1282 __le32 s_last_error_line; /* line number where error happened */ 1283 __le64 s_last_error_block; /* block involved of last error */ 1284 __u8 s_last_error_func[32]; /* function where the error happened */ 1285 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1286 __u8 s_mount_opts[64]; 1287 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1288 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1289 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1290 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1291 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1292 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1293 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1294 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1295 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1296 __le32 s_reserved[98]; /* Padding to the end of the block */ 1297 __le32 s_checksum; /* crc32c(superblock) */ 1298 }; 1299 1300 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1301 1302 #ifdef __KERNEL__ 1303 1304 /* 1305 * run-time mount flags 1306 */ 1307 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 1308 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1309 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004 1310 1311 #ifdef CONFIG_EXT4_FS_ENCRYPTION 1312 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \ 1313 EXT4_MF_TEST_DUMMY_ENCRYPTION)) 1314 #else 1315 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1316 #endif 1317 1318 /* Number of quota types we support */ 1319 #define EXT4_MAXQUOTAS 3 1320 1321 /* 1322 * fourth extended-fs super-block data in memory 1323 */ 1324 struct ext4_sb_info { 1325 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1326 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1327 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1328 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1329 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1330 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1331 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1332 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1333 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1334 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1335 unsigned long s_overhead; /* # of fs overhead clusters */ 1336 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1337 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1338 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1339 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1340 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1341 struct buffer_head **s_group_desc; 1342 unsigned int s_mount_opt; 1343 unsigned int s_mount_opt2; 1344 unsigned int s_mount_flags; 1345 unsigned int s_def_mount_opt; 1346 ext4_fsblk_t s_sb_block; 1347 atomic64_t s_resv_clusters; 1348 kuid_t s_resuid; 1349 kgid_t s_resgid; 1350 unsigned short s_mount_state; 1351 unsigned short s_pad; 1352 int s_addr_per_block_bits; 1353 int s_desc_per_block_bits; 1354 int s_inode_size; 1355 int s_first_ino; 1356 unsigned int s_inode_readahead_blks; 1357 unsigned int s_inode_goal; 1358 spinlock_t s_next_gen_lock; 1359 u32 s_next_generation; 1360 u32 s_hash_seed[4]; 1361 int s_def_hash_version; 1362 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1363 struct percpu_counter s_freeclusters_counter; 1364 struct percpu_counter s_freeinodes_counter; 1365 struct percpu_counter s_dirs_counter; 1366 struct percpu_counter s_dirtyclusters_counter; 1367 struct blockgroup_lock *s_blockgroup_lock; 1368 struct proc_dir_entry *s_proc; 1369 struct kobject s_kobj; 1370 struct completion s_kobj_unregister; 1371 struct super_block *s_sb; 1372 1373 /* Journaling */ 1374 struct journal_s *s_journal; 1375 struct list_head s_orphan; 1376 struct mutex s_orphan_lock; 1377 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1378 unsigned long s_commit_interval; 1379 u32 s_max_batch_time; 1380 u32 s_min_batch_time; 1381 struct block_device *journal_bdev; 1382 #ifdef CONFIG_QUOTA 1383 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */ 1384 int s_jquota_fmt; /* Format of quota to use */ 1385 #endif 1386 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1387 struct rb_root system_blks; 1388 1389 #ifdef EXTENTS_STATS 1390 /* ext4 extents stats */ 1391 unsigned long s_ext_min; 1392 unsigned long s_ext_max; 1393 unsigned long s_depth_max; 1394 spinlock_t s_ext_stats_lock; 1395 unsigned long s_ext_blocks; 1396 unsigned long s_ext_extents; 1397 #endif 1398 1399 /* for buddy allocator */ 1400 struct ext4_group_info ***s_group_info; 1401 struct inode *s_buddy_cache; 1402 spinlock_t s_md_lock; 1403 unsigned short *s_mb_offsets; 1404 unsigned int *s_mb_maxs; 1405 unsigned int s_group_info_size; 1406 unsigned int s_mb_free_pending; 1407 struct list_head s_freed_data_list; /* List of blocks to be freed 1408 after commit completed */ 1409 1410 /* tunables */ 1411 unsigned long s_stripe; 1412 unsigned int s_mb_stream_request; 1413 unsigned int s_mb_max_to_scan; 1414 unsigned int s_mb_min_to_scan; 1415 unsigned int s_mb_stats; 1416 unsigned int s_mb_order2_reqs; 1417 unsigned int s_mb_group_prealloc; 1418 unsigned int s_max_dir_size_kb; 1419 /* where last allocation was done - for stream allocation */ 1420 unsigned long s_mb_last_group; 1421 unsigned long s_mb_last_start; 1422 1423 /* stats for buddy allocator */ 1424 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1425 atomic_t s_bal_success; /* we found long enough chunks */ 1426 atomic_t s_bal_allocated; /* in blocks */ 1427 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1428 atomic_t s_bal_goals; /* goal hits */ 1429 atomic_t s_bal_breaks; /* too long searches */ 1430 atomic_t s_bal_2orders; /* 2^order hits */ 1431 spinlock_t s_bal_lock; 1432 unsigned long s_mb_buddies_generated; 1433 unsigned long long s_mb_generation_time; 1434 atomic_t s_mb_lost_chunks; 1435 atomic_t s_mb_preallocated; 1436 atomic_t s_mb_discarded; 1437 atomic_t s_lock_busy; 1438 1439 /* locality groups */ 1440 struct ext4_locality_group __percpu *s_locality_groups; 1441 1442 /* for write statistics */ 1443 unsigned long s_sectors_written_start; 1444 u64 s_kbytes_written; 1445 1446 /* the size of zero-out chunk */ 1447 unsigned int s_extent_max_zeroout_kb; 1448 1449 unsigned int s_log_groups_per_flex; 1450 struct flex_groups *s_flex_groups; 1451 ext4_group_t s_flex_groups_allocated; 1452 1453 /* workqueue for reserved extent conversions (buffered io) */ 1454 struct workqueue_struct *rsv_conversion_wq; 1455 1456 /* timer for periodic error stats printing */ 1457 struct timer_list s_err_report; 1458 1459 /* Lazy inode table initialization info */ 1460 struct ext4_li_request *s_li_request; 1461 /* Wait multiplier for lazy initialization thread */ 1462 unsigned int s_li_wait_mult; 1463 1464 /* Kernel thread for multiple mount protection */ 1465 struct task_struct *s_mmp_tsk; 1466 1467 /* record the last minlen when FITRIM is called. */ 1468 atomic_t s_last_trim_minblks; 1469 1470 /* Reference to checksum algorithm driver via cryptoapi */ 1471 struct crypto_shash *s_chksum_driver; 1472 1473 /* Precomputed FS UUID checksum for seeding other checksums */ 1474 __u32 s_csum_seed; 1475 1476 /* Reclaim extents from extent status tree */ 1477 struct shrinker s_es_shrinker; 1478 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1479 long s_es_nr_inode; 1480 struct ext4_es_stats s_es_stats; 1481 struct mb_cache *s_ea_block_cache; 1482 struct mb_cache *s_ea_inode_cache; 1483 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1484 1485 /* Ratelimit ext4 messages. */ 1486 struct ratelimit_state s_err_ratelimit_state; 1487 struct ratelimit_state s_warning_ratelimit_state; 1488 struct ratelimit_state s_msg_ratelimit_state; 1489 1490 /* Barrier between changing inodes' journal flags and writepages ops. */ 1491 struct percpu_rw_semaphore s_journal_flag_rwsem; 1492 struct dax_device *s_daxdev; 1493 }; 1494 1495 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1496 { 1497 return sb->s_fs_info; 1498 } 1499 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1500 { 1501 return container_of(inode, struct ext4_inode_info, vfs_inode); 1502 } 1503 1504 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1505 { 1506 return ino == EXT4_ROOT_INO || 1507 ino == EXT4_USR_QUOTA_INO || 1508 ino == EXT4_GRP_QUOTA_INO || 1509 ino == EXT4_BOOT_LOADER_INO || 1510 ino == EXT4_JOURNAL_INO || 1511 ino == EXT4_RESIZE_INO || 1512 (ino >= EXT4_FIRST_INO(sb) && 1513 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1514 } 1515 1516 /* 1517 * Inode dynamic state flags 1518 */ 1519 enum { 1520 EXT4_STATE_JDATA, /* journaled data exists */ 1521 EXT4_STATE_NEW, /* inode is newly created */ 1522 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1523 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1524 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1525 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1526 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1527 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1528 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1529 nolocking */ 1530 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1531 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1532 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1533 }; 1534 1535 #define EXT4_INODE_BIT_FNS(name, field, offset) \ 1536 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1537 { \ 1538 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1539 } \ 1540 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1541 { \ 1542 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1543 } \ 1544 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1545 { \ 1546 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1547 } 1548 1549 /* Add these declarations here only so that these functions can be 1550 * found by name. Otherwise, they are very hard to locate. */ 1551 static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1552 static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1553 static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1554 EXT4_INODE_BIT_FNS(flag, flags, 0) 1555 1556 /* Add these declarations here only so that these functions can be 1557 * found by name. Otherwise, they are very hard to locate. */ 1558 static inline int ext4_test_inode_state(struct inode *inode, int bit); 1559 static inline void ext4_set_inode_state(struct inode *inode, int bit); 1560 static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1561 #if (BITS_PER_LONG < 64) 1562 EXT4_INODE_BIT_FNS(state, state_flags, 0) 1563 1564 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1565 { 1566 (ei)->i_state_flags = 0; 1567 } 1568 #else 1569 EXT4_INODE_BIT_FNS(state, flags, 32) 1570 1571 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1572 { 1573 /* We depend on the fact that callers will set i_flags */ 1574 } 1575 #endif 1576 #else 1577 /* Assume that user mode programs are passing in an ext4fs superblock, not 1578 * a kernel struct super_block. This will allow us to call the feature-test 1579 * macros from user land. */ 1580 #define EXT4_SB(sb) (sb) 1581 #endif 1582 1583 /* 1584 * Returns true if the inode is inode is encrypted 1585 */ 1586 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1587 1588 /* 1589 * Codes for operating systems 1590 */ 1591 #define EXT4_OS_LINUX 0 1592 #define EXT4_OS_HURD 1 1593 #define EXT4_OS_MASIX 2 1594 #define EXT4_OS_FREEBSD 3 1595 #define EXT4_OS_LITES 4 1596 1597 /* 1598 * Revision levels 1599 */ 1600 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1601 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1602 1603 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1604 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1605 1606 #define EXT4_GOOD_OLD_INODE_SIZE 128 1607 1608 /* 1609 * Feature set definitions 1610 */ 1611 1612 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1613 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1614 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1615 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1616 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1617 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1618 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 1619 1620 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1621 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1622 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1623 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1624 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1625 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1626 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1627 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1628 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1629 /* 1630 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1631 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1632 * all other data structures' checksums. However, the METADATA_CSUM and 1633 * GDT_CSUM bits are mutually exclusive. 1634 */ 1635 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1636 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 1637 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 1638 1639 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1640 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1641 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1642 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1643 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1644 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1645 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1646 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1647 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1648 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1649 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1650 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 1651 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1652 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1653 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 1654 1655 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 1656 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1657 { \ 1658 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1659 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 1660 } \ 1661 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1662 { \ 1663 EXT4_SB(sb)->s_es->s_feature_compat |= \ 1664 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1665 } \ 1666 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1667 { \ 1668 EXT4_SB(sb)->s_es->s_feature_compat &= \ 1669 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 1670 } 1671 1672 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 1673 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1674 { \ 1675 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1676 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 1677 } \ 1678 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1679 { \ 1680 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 1681 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1682 } \ 1683 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1684 { \ 1685 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 1686 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 1687 } 1688 1689 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 1690 static inline bool ext4_has_feature_##name(struct super_block *sb) \ 1691 { \ 1692 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1693 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 1694 } \ 1695 static inline void ext4_set_feature_##name(struct super_block *sb) \ 1696 { \ 1697 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 1698 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1699 } \ 1700 static inline void ext4_clear_feature_##name(struct super_block *sb) \ 1701 { \ 1702 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 1703 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 1704 } 1705 1706 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 1707 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 1708 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 1709 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 1710 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 1711 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 1712 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 1713 1714 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 1715 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 1716 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 1717 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 1718 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 1719 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 1720 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 1721 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 1722 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 1723 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 1724 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 1725 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 1726 1727 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 1728 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 1729 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 1730 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 1731 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 1732 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 1733 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 1734 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 1735 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 1736 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 1737 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 1738 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 1739 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 1740 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 1741 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 1742 1743 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1744 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1745 EXT4_FEATURE_INCOMPAT_META_BG) 1746 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1747 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1748 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1749 1750 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1751 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1752 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1753 EXT4_FEATURE_INCOMPAT_META_BG) 1754 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1755 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1756 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1757 1758 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1759 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1760 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1761 EXT4_FEATURE_INCOMPAT_META_BG| \ 1762 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1763 EXT4_FEATURE_INCOMPAT_64BIT| \ 1764 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1765 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 1766 EXT4_FEATURE_INCOMPAT_MMP | \ 1767 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 1768 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 1769 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 1770 EXT4_FEATURE_INCOMPAT_LARGEDIR) 1771 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1772 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1773 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1774 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1775 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1776 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1777 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1778 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1779 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1780 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 1781 EXT4_FEATURE_RO_COMPAT_PROJECT) 1782 1783 #define EXTN_FEATURE_FUNCS(ver) \ 1784 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 1785 { \ 1786 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 1787 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 1788 } \ 1789 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 1790 { \ 1791 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 1792 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 1793 } \ 1794 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 1795 { \ 1796 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 1797 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 1798 } 1799 1800 EXTN_FEATURE_FUNCS(2) 1801 EXTN_FEATURE_FUNCS(3) 1802 EXTN_FEATURE_FUNCS(4) 1803 1804 static inline bool ext4_has_compat_features(struct super_block *sb) 1805 { 1806 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 1807 } 1808 static inline bool ext4_has_ro_compat_features(struct super_block *sb) 1809 { 1810 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 1811 } 1812 static inline bool ext4_has_incompat_features(struct super_block *sb) 1813 { 1814 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 1815 } 1816 1817 /* 1818 * Superblock flags 1819 */ 1820 #define EXT4_FLAGS_RESIZING 0 1821 #define EXT4_FLAGS_SHUTDOWN 1 1822 1823 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi) 1824 { 1825 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags); 1826 } 1827 1828 1829 /* 1830 * Default values for user and/or group using reserved blocks 1831 */ 1832 #define EXT4_DEF_RESUID 0 1833 #define EXT4_DEF_RESGID 0 1834 1835 /* 1836 * Default project ID 1837 */ 1838 #define EXT4_DEF_PROJID 0 1839 1840 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1841 1842 /* 1843 * Default mount options 1844 */ 1845 #define EXT4_DEFM_DEBUG 0x0001 1846 #define EXT4_DEFM_BSDGROUPS 0x0002 1847 #define EXT4_DEFM_XATTR_USER 0x0004 1848 #define EXT4_DEFM_ACL 0x0008 1849 #define EXT4_DEFM_UID16 0x0010 1850 #define EXT4_DEFM_JMODE 0x0060 1851 #define EXT4_DEFM_JMODE_DATA 0x0020 1852 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1853 #define EXT4_DEFM_JMODE_WBACK 0x0060 1854 #define EXT4_DEFM_NOBARRIER 0x0100 1855 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1856 #define EXT4_DEFM_DISCARD 0x0400 1857 #define EXT4_DEFM_NODELALLOC 0x0800 1858 1859 /* 1860 * Default journal batch times 1861 */ 1862 #define EXT4_DEF_MIN_BATCH_TIME 0 1863 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1864 1865 /* 1866 * Minimum number of groups in a flexgroup before we separate out 1867 * directories into the first block group of a flexgroup 1868 */ 1869 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1870 1871 /* 1872 * Structure of a directory entry 1873 */ 1874 #define EXT4_NAME_LEN 255 1875 1876 struct ext4_dir_entry { 1877 __le32 inode; /* Inode number */ 1878 __le16 rec_len; /* Directory entry length */ 1879 __le16 name_len; /* Name length */ 1880 char name[EXT4_NAME_LEN]; /* File name */ 1881 }; 1882 1883 /* 1884 * The new version of the directory entry. Since EXT4 structures are 1885 * stored in intel byte order, and the name_len field could never be 1886 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1887 * file_type field. 1888 */ 1889 struct ext4_dir_entry_2 { 1890 __le32 inode; /* Inode number */ 1891 __le16 rec_len; /* Directory entry length */ 1892 __u8 name_len; /* Name length */ 1893 __u8 file_type; 1894 char name[EXT4_NAME_LEN]; /* File name */ 1895 }; 1896 1897 /* 1898 * This is a bogus directory entry at the end of each leaf block that 1899 * records checksums. 1900 */ 1901 struct ext4_dir_entry_tail { 1902 __le32 det_reserved_zero1; /* Pretend to be unused */ 1903 __le16 det_rec_len; /* 12 */ 1904 __u8 det_reserved_zero2; /* Zero name length */ 1905 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1906 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1907 }; 1908 1909 #define EXT4_DIRENT_TAIL(block, blocksize) \ 1910 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1911 ((blocksize) - \ 1912 sizeof(struct ext4_dir_entry_tail)))) 1913 1914 /* 1915 * Ext4 directory file types. Only the low 3 bits are used. The 1916 * other bits are reserved for now. 1917 */ 1918 #define EXT4_FT_UNKNOWN 0 1919 #define EXT4_FT_REG_FILE 1 1920 #define EXT4_FT_DIR 2 1921 #define EXT4_FT_CHRDEV 3 1922 #define EXT4_FT_BLKDEV 4 1923 #define EXT4_FT_FIFO 5 1924 #define EXT4_FT_SOCK 6 1925 #define EXT4_FT_SYMLINK 7 1926 1927 #define EXT4_FT_MAX 8 1928 1929 #define EXT4_FT_DIR_CSUM 0xDE 1930 1931 /* 1932 * EXT4_DIR_PAD defines the directory entries boundaries 1933 * 1934 * NOTE: It must be a multiple of 4 1935 */ 1936 #define EXT4_DIR_PAD 4 1937 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1938 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1939 ~EXT4_DIR_ROUND) 1940 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1941 1942 /* 1943 * If we ever get support for fs block sizes > page_size, we'll need 1944 * to remove the #if statements in the next two functions... 1945 */ 1946 static inline unsigned int 1947 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1948 { 1949 unsigned len = le16_to_cpu(dlen); 1950 1951 #if (PAGE_SIZE >= 65536) 1952 if (len == EXT4_MAX_REC_LEN || len == 0) 1953 return blocksize; 1954 return (len & 65532) | ((len & 3) << 16); 1955 #else 1956 return len; 1957 #endif 1958 } 1959 1960 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1961 { 1962 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1963 BUG(); 1964 #if (PAGE_SIZE >= 65536) 1965 if (len < 65536) 1966 return cpu_to_le16(len); 1967 if (len == blocksize) { 1968 if (blocksize == 65536) 1969 return cpu_to_le16(EXT4_MAX_REC_LEN); 1970 else 1971 return cpu_to_le16(0); 1972 } 1973 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1974 #else 1975 return cpu_to_le16(len); 1976 #endif 1977 } 1978 1979 /* 1980 * Hash Tree Directory indexing 1981 * (c) Daniel Phillips, 2001 1982 */ 1983 1984 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 1985 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1986 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 1987 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 1988 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1989 1990 /* Legal values for the dx_root hash_version field: */ 1991 1992 #define DX_HASH_LEGACY 0 1993 #define DX_HASH_HALF_MD4 1 1994 #define DX_HASH_TEA 2 1995 #define DX_HASH_LEGACY_UNSIGNED 3 1996 #define DX_HASH_HALF_MD4_UNSIGNED 4 1997 #define DX_HASH_TEA_UNSIGNED 5 1998 1999 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 2000 const void *address, unsigned int length) 2001 { 2002 struct { 2003 struct shash_desc shash; 2004 char ctx[4]; 2005 } desc; 2006 int err; 2007 2008 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 2009 2010 desc.shash.tfm = sbi->s_chksum_driver; 2011 desc.shash.flags = 0; 2012 *(u32 *)desc.ctx = crc; 2013 2014 err = crypto_shash_update(&desc.shash, address, length); 2015 BUG_ON(err); 2016 2017 return *(u32 *)desc.ctx; 2018 } 2019 2020 #ifdef __KERNEL__ 2021 2022 /* hash info structure used by the directory hash */ 2023 struct dx_hash_info 2024 { 2025 u32 hash; 2026 u32 minor_hash; 2027 int hash_version; 2028 u32 *seed; 2029 }; 2030 2031 2032 /* 32 and 64 bit signed EOF for dx directories */ 2033 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2034 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2035 2036 2037 /* 2038 * Control parameters used by ext4_htree_next_block 2039 */ 2040 #define HASH_NB_ALWAYS 1 2041 2042 struct ext4_filename { 2043 const struct qstr *usr_fname; 2044 struct fscrypt_str disk_name; 2045 struct dx_hash_info hinfo; 2046 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2047 struct fscrypt_str crypto_buf; 2048 #endif 2049 }; 2050 2051 #define fname_name(p) ((p)->disk_name.name) 2052 #define fname_len(p) ((p)->disk_name.len) 2053 2054 /* 2055 * Describe an inode's exact location on disk and in memory 2056 */ 2057 struct ext4_iloc 2058 { 2059 struct buffer_head *bh; 2060 unsigned long offset; 2061 ext4_group_t block_group; 2062 }; 2063 2064 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2065 { 2066 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2067 } 2068 2069 static inline bool ext4_is_quota_file(struct inode *inode) 2070 { 2071 return IS_NOQUOTA(inode) && 2072 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 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 /* htree levels for ext4 */ 2104 #define EXT4_HTREE_LEVEL_COMPAT 2 2105 #define EXT4_HTREE_LEVEL 3 2106 2107 static inline int ext4_dir_htree_level(struct super_block *sb) 2108 { 2109 return ext4_has_feature_largedir(sb) ? 2110 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2111 } 2112 2113 /* 2114 * Timeout and state flag for lazy initialization inode thread. 2115 */ 2116 #define EXT4_DEF_LI_WAIT_MULT 10 2117 #define EXT4_DEF_LI_MAX_START_DELAY 5 2118 #define EXT4_LAZYINIT_QUIT 0x0001 2119 #define EXT4_LAZYINIT_RUNNING 0x0002 2120 2121 /* 2122 * Lazy inode table initialization info 2123 */ 2124 struct ext4_lazy_init { 2125 unsigned long li_state; 2126 struct list_head li_request_list; 2127 struct mutex li_list_mtx; 2128 }; 2129 2130 struct ext4_li_request { 2131 struct super_block *lr_super; 2132 struct ext4_sb_info *lr_sbi; 2133 ext4_group_t lr_next_group; 2134 struct list_head lr_request; 2135 unsigned long lr_next_sched; 2136 unsigned long lr_timeout; 2137 }; 2138 2139 struct ext4_features { 2140 struct kobject f_kobj; 2141 struct completion f_kobj_unregister; 2142 }; 2143 2144 /* 2145 * This structure will be used for multiple mount protection. It will be 2146 * written into the block number saved in the s_mmp_block field in the 2147 * superblock. Programs that check MMP should assume that if 2148 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2149 * to use the filesystem, regardless of how old the timestamp is. 2150 */ 2151 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2152 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2153 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2154 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2155 2156 struct mmp_struct { 2157 __le32 mmp_magic; /* Magic number for MMP */ 2158 __le32 mmp_seq; /* Sequence no. updated periodically */ 2159 2160 /* 2161 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2162 * purposes and do not affect the correctness of the algorithm 2163 */ 2164 __le64 mmp_time; /* Time last updated */ 2165 char mmp_nodename[64]; /* Node which last updated MMP block */ 2166 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2167 2168 /* 2169 * mmp_check_interval is used to verify if the MMP block has been 2170 * updated on the block device. The value is updated based on the 2171 * maximum time to write the MMP block during an update cycle. 2172 */ 2173 __le16 mmp_check_interval; 2174 2175 __le16 mmp_pad1; 2176 __le32 mmp_pad2[226]; 2177 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2178 }; 2179 2180 /* arguments passed to the mmp thread */ 2181 struct mmpd_data { 2182 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2183 struct super_block *sb; /* super block of the fs */ 2184 }; 2185 2186 /* 2187 * Check interval multiplier 2188 * The MMP block is written every update interval and initially checked every 2189 * update interval x the multiplier (the value is then adapted based on the 2190 * write latency). The reason is that writes can be delayed under load and we 2191 * don't want readers to incorrectly assume that the filesystem is no longer 2192 * in use. 2193 */ 2194 #define EXT4_MMP_CHECK_MULT 2UL 2195 2196 /* 2197 * Minimum interval for MMP checking in seconds. 2198 */ 2199 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2200 2201 /* 2202 * Maximum interval for MMP checking in seconds. 2203 */ 2204 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2205 2206 /* 2207 * Function prototypes 2208 */ 2209 2210 /* 2211 * Ok, these declarations are also in <linux/kernel.h> but none of the 2212 * ext4 source programs needs to include it so they are duplicated here. 2213 */ 2214 # define NORET_TYPE /**/ 2215 # define ATTRIB_NORET __attribute__((noreturn)) 2216 # define NORET_AND noreturn, 2217 2218 /* bitmap.c */ 2219 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2220 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2221 struct ext4_group_desc *gdp, 2222 struct buffer_head *bh, int sz); 2223 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2224 struct ext4_group_desc *gdp, 2225 struct buffer_head *bh, int sz); 2226 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 2227 struct ext4_group_desc *gdp, 2228 struct buffer_head *bh); 2229 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 2230 struct ext4_group_desc *gdp, 2231 struct buffer_head *bh); 2232 2233 /* balloc.c */ 2234 extern void ext4_get_group_no_and_offset(struct super_block *sb, 2235 ext4_fsblk_t blocknr, 2236 ext4_group_t *blockgrpp, 2237 ext4_grpblk_t *offsetp); 2238 extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2239 ext4_fsblk_t block); 2240 2241 extern unsigned int ext4_block_group(struct super_block *sb, 2242 ext4_fsblk_t blocknr); 2243 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 2244 ext4_fsblk_t blocknr); 2245 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2246 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2247 ext4_group_t group); 2248 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2249 ext4_fsblk_t goal, 2250 unsigned int flags, 2251 unsigned long *count, 2252 int *errp); 2253 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2254 s64 nclusters, unsigned int flags); 2255 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2256 extern void ext4_check_blocks_bitmap(struct super_block *); 2257 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2258 ext4_group_t block_group, 2259 struct buffer_head ** bh); 2260 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2261 2262 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2263 ext4_group_t block_group); 2264 extern int ext4_wait_block_bitmap(struct super_block *sb, 2265 ext4_group_t block_group, 2266 struct buffer_head *bh); 2267 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2268 ext4_group_t block_group); 2269 extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2270 ext4_group_t block_group, 2271 struct ext4_group_desc *gdp); 2272 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2273 2274 static inline bool ext4_encrypted_inode(struct inode *inode) 2275 { 2276 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT); 2277 } 2278 2279 #ifdef CONFIG_EXT4_FS_ENCRYPTION 2280 static inline int ext4_fname_setup_filename(struct inode *dir, 2281 const struct qstr *iname, 2282 int lookup, struct ext4_filename *fname) 2283 { 2284 struct fscrypt_name name; 2285 int err; 2286 2287 memset(fname, 0, sizeof(struct ext4_filename)); 2288 2289 err = fscrypt_setup_filename(dir, iname, lookup, &name); 2290 2291 fname->usr_fname = name.usr_fname; 2292 fname->disk_name = name.disk_name; 2293 fname->hinfo.hash = name.hash; 2294 fname->hinfo.minor_hash = name.minor_hash; 2295 fname->crypto_buf = name.crypto_buf; 2296 return err; 2297 } 2298 2299 static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2300 { 2301 struct fscrypt_name name; 2302 2303 name.crypto_buf = fname->crypto_buf; 2304 fscrypt_free_filename(&name); 2305 2306 fname->crypto_buf.name = NULL; 2307 fname->usr_fname = NULL; 2308 fname->disk_name.name = NULL; 2309 } 2310 #else 2311 static inline int ext4_fname_setup_filename(struct inode *dir, 2312 const struct qstr *iname, 2313 int lookup, struct ext4_filename *fname) 2314 { 2315 fname->usr_fname = iname; 2316 fname->disk_name.name = (unsigned char *) iname->name; 2317 fname->disk_name.len = iname->len; 2318 return 0; 2319 } 2320 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { } 2321 2322 #endif 2323 2324 /* dir.c */ 2325 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2326 struct file *, 2327 struct ext4_dir_entry_2 *, 2328 struct buffer_head *, char *, int, 2329 unsigned int); 2330 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2331 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2332 (de), (bh), (buf), (size), (offset))) 2333 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2334 __u32 minor_hash, 2335 struct ext4_dir_entry_2 *dirent, 2336 struct fscrypt_str *ent_name); 2337 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2338 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2339 struct buffer_head *bh, 2340 void *buf, int buf_size, 2341 struct ext4_filename *fname, 2342 struct ext4_dir_entry_2 **dest_de); 2343 void ext4_insert_dentry(struct inode *inode, 2344 struct ext4_dir_entry_2 *de, 2345 int buf_size, 2346 struct ext4_filename *fname); 2347 static inline void ext4_update_dx_flag(struct inode *inode) 2348 { 2349 if (!ext4_has_feature_dir_index(inode->i_sb)) 2350 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2351 } 2352 static const unsigned char ext4_filetype_table[] = { 2353 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2354 }; 2355 2356 static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2357 { 2358 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2359 return DT_UNKNOWN; 2360 2361 return ext4_filetype_table[filetype]; 2362 } 2363 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2364 void *buf, int buf_size); 2365 2366 /* fsync.c */ 2367 extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2368 2369 /* hash.c */ 2370 extern int ext4fs_dirhash(const char *name, int len, struct 2371 dx_hash_info *hinfo); 2372 2373 /* ialloc.c */ 2374 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2375 const struct qstr *qstr, __u32 goal, 2376 uid_t *owner, __u32 i_flags, 2377 int handle_type, unsigned int line_no, 2378 int nblocks); 2379 2380 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2381 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2382 i_flags, 0, 0, 0) 2383 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2384 type, nblocks) \ 2385 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2386 0, (type), __LINE__, (nblocks)) 2387 2388 2389 extern void ext4_free_inode(handle_t *, struct inode *); 2390 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2391 extern unsigned long ext4_count_free_inodes(struct super_block *); 2392 extern unsigned long ext4_count_dirs(struct super_block *); 2393 extern void ext4_check_inodes_bitmap(struct super_block *); 2394 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2395 extern int ext4_init_inode_table(struct super_block *sb, 2396 ext4_group_t group, int barrier); 2397 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2398 2399 /* mballoc.c */ 2400 extern const struct file_operations ext4_seq_mb_groups_fops; 2401 extern long ext4_mb_stats; 2402 extern long ext4_mb_max_to_scan; 2403 extern int ext4_mb_init(struct super_block *); 2404 extern int ext4_mb_release(struct super_block *); 2405 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2406 struct ext4_allocation_request *, int *); 2407 extern int ext4_mb_reserve_blocks(struct super_block *, int); 2408 extern void ext4_discard_preallocations(struct inode *); 2409 extern int __init ext4_init_mballoc(void); 2410 extern void ext4_exit_mballoc(void); 2411 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2412 struct buffer_head *bh, ext4_fsblk_t block, 2413 unsigned long count, int flags); 2414 extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2415 ext4_group_t ngroups); 2416 extern int ext4_mb_add_groupinfo(struct super_block *sb, 2417 ext4_group_t i, struct ext4_group_desc *desc); 2418 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2419 ext4_fsblk_t block, unsigned long count); 2420 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2421 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 2422 2423 /* inode.c */ 2424 int ext4_inode_is_fast_symlink(struct inode *inode); 2425 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 2426 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 2427 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 2428 bool wait, struct buffer_head **bhs); 2429 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 2430 struct buffer_head *bh_result, int create); 2431 int ext4_get_block(struct inode *inode, sector_t iblock, 2432 struct buffer_head *bh_result, int create); 2433 int ext4_dio_get_block(struct inode *inode, sector_t iblock, 2434 struct buffer_head *bh_result, int create); 2435 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2436 struct buffer_head *bh, int create); 2437 int ext4_walk_page_buffers(handle_t *handle, 2438 struct buffer_head *head, 2439 unsigned from, 2440 unsigned to, 2441 int *partial, 2442 int (*fn)(handle_t *handle, 2443 struct buffer_head *bh)); 2444 int do_journal_get_write_access(handle_t *handle, 2445 struct buffer_head *bh); 2446 #define FALL_BACK_TO_NONDELALLOC 1 2447 #define CONVERT_INLINE_DATA 2 2448 2449 extern struct inode *ext4_iget(struct super_block *, unsigned long); 2450 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long); 2451 extern int ext4_write_inode(struct inode *, struct writeback_control *); 2452 extern int ext4_setattr(struct dentry *, struct iattr *); 2453 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int); 2454 extern void ext4_evict_inode(struct inode *); 2455 extern void ext4_clear_inode(struct inode *); 2456 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int); 2457 extern int ext4_sync_inode(handle_t *, struct inode *); 2458 extern void ext4_dirty_inode(struct inode *, int); 2459 extern int ext4_change_inode_journal_flag(struct inode *, int); 2460 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2461 extern int ext4_inode_attach_jinode(struct inode *inode); 2462 extern int ext4_can_truncate(struct inode *inode); 2463 extern int ext4_truncate(struct inode *); 2464 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2465 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2466 extern void ext4_set_inode_flags(struct inode *); 2467 extern int ext4_alloc_da_blocks(struct inode *inode); 2468 extern void ext4_set_aops(struct inode *inode); 2469 extern int ext4_writepage_trans_blocks(struct inode *); 2470 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2471 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2472 loff_t lstart, loff_t lend); 2473 extern int ext4_page_mkwrite(struct vm_fault *vmf); 2474 extern int ext4_filemap_fault(struct vm_fault *vmf); 2475 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2476 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 2477 extern void ext4_da_update_reserve_space(struct inode *inode, 2478 int used, int quota_claim); 2479 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 2480 ext4_fsblk_t pblk, ext4_lblk_t len); 2481 2482 /* indirect.c */ 2483 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2484 struct ext4_map_blocks *map, int flags); 2485 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2486 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2487 extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2488 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 2489 ext4_lblk_t start, ext4_lblk_t end); 2490 2491 /* ioctl.c */ 2492 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2493 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2494 2495 /* migrate.c */ 2496 extern int ext4_ext_migrate(struct inode *); 2497 extern int ext4_ind_migrate(struct inode *inode); 2498 2499 /* namei.c */ 2500 extern int ext4_dirent_csum_verify(struct inode *inode, 2501 struct ext4_dir_entry *dirent); 2502 extern int ext4_orphan_add(handle_t *, struct inode *); 2503 extern int ext4_orphan_del(handle_t *, struct inode *); 2504 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2505 __u32 start_minor_hash, __u32 *next_hash); 2506 extern int ext4_search_dir(struct buffer_head *bh, 2507 char *search_buf, 2508 int buf_size, 2509 struct inode *dir, 2510 struct ext4_filename *fname, 2511 unsigned int offset, 2512 struct ext4_dir_entry_2 **res_dir); 2513 extern int ext4_generic_delete_entry(handle_t *handle, 2514 struct inode *dir, 2515 struct ext4_dir_entry_2 *de_del, 2516 struct buffer_head *bh, 2517 void *entry_buf, 2518 int buf_size, 2519 int csum_size); 2520 extern bool ext4_empty_dir(struct inode *inode); 2521 2522 /* resize.c */ 2523 extern int ext4_group_add(struct super_block *sb, 2524 struct ext4_new_group_data *input); 2525 extern int ext4_group_extend(struct super_block *sb, 2526 struct ext4_super_block *es, 2527 ext4_fsblk_t n_blocks_count); 2528 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2529 2530 /* super.c */ 2531 extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 2532 extern int ext4_calculate_overhead(struct super_block *sb); 2533 extern void ext4_superblock_csum_set(struct super_block *sb); 2534 extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2535 extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2536 extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2537 ext4_group_t ngroup); 2538 extern const char *ext4_decode_error(struct super_block *sb, int errno, 2539 char nbuf[16]); 2540 2541 extern __printf(4, 5) 2542 void __ext4_error(struct super_block *, const char *, unsigned int, 2543 const char *, ...); 2544 extern __printf(5, 6) 2545 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2546 const char *, ...); 2547 extern __printf(5, 6) 2548 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2549 const char *, ...); 2550 extern void __ext4_std_error(struct super_block *, const char *, 2551 unsigned int, int); 2552 extern __printf(4, 5) 2553 void __ext4_abort(struct super_block *, const char *, unsigned int, 2554 const char *, ...); 2555 extern __printf(4, 5) 2556 void __ext4_warning(struct super_block *, const char *, unsigned int, 2557 const char *, ...); 2558 extern __printf(4, 5) 2559 void __ext4_warning_inode(const struct inode *inode, const char *function, 2560 unsigned int line, const char *fmt, ...); 2561 extern __printf(3, 4) 2562 void __ext4_msg(struct super_block *, const char *, const char *, ...); 2563 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2564 const char *, unsigned int, const char *); 2565 extern __printf(7, 8) 2566 void __ext4_grp_locked_error(const char *, unsigned int, 2567 struct super_block *, ext4_group_t, 2568 unsigned long, ext4_fsblk_t, 2569 const char *, ...); 2570 2571 #define EXT4_ERROR_INODE(inode, fmt, a...) \ 2572 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 2573 2574 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 2575 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 2576 2577 #define EXT4_ERROR_FILE(file, block, fmt, a...) \ 2578 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 2579 2580 #ifdef CONFIG_PRINTK 2581 2582 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2583 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2584 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2585 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2586 #define ext4_error(sb, fmt, ...) \ 2587 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2588 #define ext4_abort(sb, fmt, ...) \ 2589 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2590 #define ext4_warning(sb, fmt, ...) \ 2591 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2592 #define ext4_warning_inode(inode, fmt, ...) \ 2593 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 2594 #define ext4_msg(sb, level, fmt, ...) \ 2595 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2596 #define dump_mmp_msg(sb, mmp, msg) \ 2597 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2598 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2599 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2600 fmt, ##__VA_ARGS__) 2601 2602 #else 2603 2604 #define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2605 do { \ 2606 no_printk(fmt, ##__VA_ARGS__); \ 2607 __ext4_error_inode(inode, "", 0, block, " "); \ 2608 } while (0) 2609 #define ext4_error_file(file, func, line, block, fmt, ...) \ 2610 do { \ 2611 no_printk(fmt, ##__VA_ARGS__); \ 2612 __ext4_error_file(file, "", 0, block, " "); \ 2613 } while (0) 2614 #define ext4_error(sb, fmt, ...) \ 2615 do { \ 2616 no_printk(fmt, ##__VA_ARGS__); \ 2617 __ext4_error(sb, "", 0, " "); \ 2618 } while (0) 2619 #define ext4_abort(sb, fmt, ...) \ 2620 do { \ 2621 no_printk(fmt, ##__VA_ARGS__); \ 2622 __ext4_abort(sb, "", 0, " "); \ 2623 } while (0) 2624 #define ext4_warning(sb, fmt, ...) \ 2625 do { \ 2626 no_printk(fmt, ##__VA_ARGS__); \ 2627 __ext4_warning(sb, "", 0, " "); \ 2628 } while (0) 2629 #define ext4_warning_inode(inode, fmt, ...) \ 2630 do { \ 2631 no_printk(fmt, ##__VA_ARGS__); \ 2632 __ext4_warning_inode(inode, "", 0, " "); \ 2633 } while (0) 2634 #define ext4_msg(sb, level, fmt, ...) \ 2635 do { \ 2636 no_printk(fmt, ##__VA_ARGS__); \ 2637 __ext4_msg(sb, "", " "); \ 2638 } while (0) 2639 #define dump_mmp_msg(sb, mmp, msg) \ 2640 __dump_mmp_msg(sb, mmp, "", 0, "") 2641 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2642 do { \ 2643 no_printk(fmt, ##__VA_ARGS__); \ 2644 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2645 } while (0) 2646 2647 #endif 2648 2649 extern void ext4_update_dynamic_rev(struct super_block *sb); 2650 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2651 __u32 compat); 2652 extern int ext4_update_rocompat_feature(handle_t *handle, 2653 struct super_block *sb, __u32 rocompat); 2654 extern int ext4_update_incompat_feature(handle_t *handle, 2655 struct super_block *sb, __u32 incompat); 2656 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2657 struct ext4_group_desc *bg); 2658 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2659 struct ext4_group_desc *bg); 2660 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2661 struct ext4_group_desc *bg); 2662 extern __u32 ext4_free_group_clusters(struct super_block *sb, 2663 struct ext4_group_desc *bg); 2664 extern __u32 ext4_free_inodes_count(struct super_block *sb, 2665 struct ext4_group_desc *bg); 2666 extern __u32 ext4_used_dirs_count(struct super_block *sb, 2667 struct ext4_group_desc *bg); 2668 extern __u32 ext4_itable_unused_count(struct super_block *sb, 2669 struct ext4_group_desc *bg); 2670 extern void ext4_block_bitmap_set(struct super_block *sb, 2671 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2672 extern void ext4_inode_bitmap_set(struct super_block *sb, 2673 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2674 extern void ext4_inode_table_set(struct super_block *sb, 2675 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2676 extern void ext4_free_group_clusters_set(struct super_block *sb, 2677 struct ext4_group_desc *bg, 2678 __u32 count); 2679 extern void ext4_free_inodes_set(struct super_block *sb, 2680 struct ext4_group_desc *bg, __u32 count); 2681 extern void ext4_used_dirs_set(struct super_block *sb, 2682 struct ext4_group_desc *bg, __u32 count); 2683 extern void ext4_itable_unused_set(struct super_block *sb, 2684 struct ext4_group_desc *bg, __u32 count); 2685 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2686 struct ext4_group_desc *gdp); 2687 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2688 struct ext4_group_desc *gdp); 2689 extern int ext4_register_li_request(struct super_block *sb, 2690 ext4_group_t first_not_zeroed); 2691 2692 static inline int ext4_has_metadata_csum(struct super_block *sb) 2693 { 2694 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) && 2695 !EXT4_SB(sb)->s_chksum_driver); 2696 2697 return ext4_has_feature_metadata_csum(sb) && 2698 (EXT4_SB(sb)->s_chksum_driver != NULL); 2699 } 2700 2701 static inline int ext4_has_group_desc_csum(struct super_block *sb) 2702 { 2703 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb); 2704 } 2705 2706 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2707 { 2708 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2709 le32_to_cpu(es->s_blocks_count_lo); 2710 } 2711 2712 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2713 { 2714 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2715 le32_to_cpu(es->s_r_blocks_count_lo); 2716 } 2717 2718 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2719 { 2720 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2721 le32_to_cpu(es->s_free_blocks_count_lo); 2722 } 2723 2724 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2725 ext4_fsblk_t blk) 2726 { 2727 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2728 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2729 } 2730 2731 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2732 ext4_fsblk_t blk) 2733 { 2734 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2735 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2736 } 2737 2738 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2739 ext4_fsblk_t blk) 2740 { 2741 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2742 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2743 } 2744 2745 static inline loff_t ext4_isize(struct super_block *sb, 2746 struct ext4_inode *raw_inode) 2747 { 2748 if (ext4_has_feature_largedir(sb) || 2749 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2750 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2751 le32_to_cpu(raw_inode->i_size_lo); 2752 2753 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2754 } 2755 2756 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2757 { 2758 raw_inode->i_size_lo = cpu_to_le32(i_size); 2759 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2760 } 2761 2762 static inline 2763 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2764 ext4_group_t group) 2765 { 2766 struct ext4_group_info ***grp_info; 2767 long indexv, indexh; 2768 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2769 grp_info = EXT4_SB(sb)->s_group_info; 2770 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2771 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2772 return grp_info[indexv][indexh]; 2773 } 2774 2775 /* 2776 * Reading s_groups_count requires using smp_rmb() afterwards. See 2777 * the locking protocol documented in the comments of ext4_group_add() 2778 * in resize.c 2779 */ 2780 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2781 { 2782 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2783 2784 smp_rmb(); 2785 return ngroups; 2786 } 2787 2788 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2789 ext4_group_t block_group) 2790 { 2791 return block_group >> sbi->s_log_groups_per_flex; 2792 } 2793 2794 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2795 { 2796 return 1 << sbi->s_log_groups_per_flex; 2797 } 2798 2799 #define ext4_std_error(sb, errno) \ 2800 do { \ 2801 if ((errno)) \ 2802 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2803 } while (0) 2804 2805 #ifdef CONFIG_SMP 2806 /* Each CPU can accumulate percpu_counter_batch clusters in their local 2807 * counters. So we need to make sure we have free clusters more 2808 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2809 */ 2810 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2811 #else 2812 #define EXT4_FREECLUSTERS_WATERMARK 0 2813 #endif 2814 2815 /* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2816 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2817 { 2818 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2819 !inode_is_locked(inode)); 2820 down_write(&EXT4_I(inode)->i_data_sem); 2821 if (newsize > EXT4_I(inode)->i_disksize) 2822 EXT4_I(inode)->i_disksize = newsize; 2823 up_write(&EXT4_I(inode)->i_data_sem); 2824 } 2825 2826 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */ 2827 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 2828 { 2829 int changed = 0; 2830 2831 if (newsize > inode->i_size) { 2832 i_size_write(inode, newsize); 2833 changed = 1; 2834 } 2835 if (newsize > EXT4_I(inode)->i_disksize) { 2836 ext4_update_i_disksize(inode, newsize); 2837 changed |= 2; 2838 } 2839 return changed; 2840 } 2841 2842 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 2843 loff_t len); 2844 2845 struct ext4_group_info { 2846 unsigned long bb_state; 2847 struct rb_root bb_free_root; 2848 ext4_grpblk_t bb_first_free; /* first free block */ 2849 ext4_grpblk_t bb_free; /* total free blocks */ 2850 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2851 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2852 struct list_head bb_prealloc_list; 2853 #ifdef DOUBLE_CHECK 2854 void *bb_bitmap; 2855 #endif 2856 struct rw_semaphore alloc_sem; 2857 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2858 * regions, index is order. 2859 * bb_counters[3] = 5 means 2860 * 5 free 8-block regions. */ 2861 }; 2862 2863 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2864 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2865 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2866 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2867 2868 #define EXT4_MB_GRP_NEED_INIT(grp) \ 2869 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2870 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2871 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2872 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2873 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2874 2875 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2876 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2877 #define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2878 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2879 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2880 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2881 2882 #define EXT4_MAX_CONTENTION 8 2883 #define EXT4_CONTENTION_THRESHOLD 2 2884 2885 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2886 ext4_group_t group) 2887 { 2888 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2889 } 2890 2891 /* 2892 * Returns true if the filesystem is busy enough that attempts to 2893 * access the block group locks has run into contention. 2894 */ 2895 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2896 { 2897 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2898 } 2899 2900 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2901 { 2902 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2903 if (spin_trylock(lock)) 2904 /* 2905 * We're able to grab the lock right away, so drop the 2906 * lock contention counter. 2907 */ 2908 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2909 else { 2910 /* 2911 * The lock is busy, so bump the contention counter, 2912 * and then wait on the spin lock. 2913 */ 2914 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2915 EXT4_MAX_CONTENTION); 2916 spin_lock(lock); 2917 } 2918 } 2919 2920 static inline void ext4_unlock_group(struct super_block *sb, 2921 ext4_group_t group) 2922 { 2923 spin_unlock(ext4_group_lock_ptr(sb, group)); 2924 } 2925 2926 /* 2927 * Block validity checking 2928 */ 2929 #define ext4_check_indirect_blockref(inode, bh) \ 2930 ext4_check_blockref(__func__, __LINE__, inode, \ 2931 (__le32 *)(bh)->b_data, \ 2932 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2933 2934 #define ext4_ind_check_inode(inode) \ 2935 ext4_check_blockref(__func__, __LINE__, inode, \ 2936 EXT4_I(inode)->i_data, \ 2937 EXT4_NDIR_BLOCKS) 2938 2939 /* 2940 * Inodes and files operations 2941 */ 2942 2943 /* dir.c */ 2944 extern const struct file_operations ext4_dir_operations; 2945 2946 /* file.c */ 2947 extern const struct inode_operations ext4_file_inode_operations; 2948 extern const struct file_operations ext4_file_operations; 2949 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2950 2951 /* inline.c */ 2952 extern int ext4_get_max_inline_size(struct inode *inode); 2953 extern int ext4_find_inline_data_nolock(struct inode *inode); 2954 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2955 unsigned int len); 2956 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2957 2958 extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2959 extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2960 struct inode *inode, 2961 loff_t pos, unsigned len, 2962 unsigned flags, 2963 struct page **pagep); 2964 extern int ext4_write_inline_data_end(struct inode *inode, 2965 loff_t pos, unsigned len, 2966 unsigned copied, 2967 struct page *page); 2968 extern struct buffer_head * 2969 ext4_journalled_write_inline_data(struct inode *inode, 2970 unsigned len, 2971 struct page *page); 2972 extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2973 struct inode *inode, 2974 loff_t pos, unsigned len, 2975 unsigned flags, 2976 struct page **pagep, 2977 void **fsdata); 2978 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2979 unsigned len, unsigned copied, 2980 struct page *page); 2981 extern int ext4_try_add_inline_entry(handle_t *handle, 2982 struct ext4_filename *fname, 2983 struct inode *dir, struct inode *inode); 2984 extern int ext4_try_create_inline_dir(handle_t *handle, 2985 struct inode *parent, 2986 struct inode *inode); 2987 extern int ext4_read_inline_dir(struct file *filp, 2988 struct dir_context *ctx, 2989 int *has_inline_data); 2990 extern int htree_inlinedir_to_tree(struct file *dir_file, 2991 struct inode *dir, ext4_lblk_t block, 2992 struct dx_hash_info *hinfo, 2993 __u32 start_hash, __u32 start_minor_hash, 2994 int *has_inline_data); 2995 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2996 struct ext4_filename *fname, 2997 struct ext4_dir_entry_2 **res_dir, 2998 int *has_inline_data); 2999 extern int ext4_delete_inline_entry(handle_t *handle, 3000 struct inode *dir, 3001 struct ext4_dir_entry_2 *de_del, 3002 struct buffer_head *bh, 3003 int *has_inline_data); 3004 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3005 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3006 struct ext4_dir_entry_2 **parent_de, 3007 int *retval); 3008 extern int ext4_inline_data_fiemap(struct inode *inode, 3009 struct fiemap_extent_info *fieinfo, 3010 int *has_inline, __u64 start, __u64 len); 3011 3012 struct iomap; 3013 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3014 3015 extern int ext4_try_to_evict_inline_data(handle_t *handle, 3016 struct inode *inode, 3017 int needed); 3018 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3019 3020 extern int ext4_convert_inline_data(struct inode *inode); 3021 3022 static inline int ext4_has_inline_data(struct inode *inode) 3023 { 3024 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3025 EXT4_I(inode)->i_inline_off; 3026 } 3027 3028 /* namei.c */ 3029 extern const struct inode_operations ext4_dir_inode_operations; 3030 extern const struct inode_operations ext4_special_inode_operations; 3031 extern struct dentry *ext4_get_parent(struct dentry *child); 3032 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 3033 struct ext4_dir_entry_2 *de, 3034 int blocksize, int csum_size, 3035 unsigned int parent_ino, int dotdot_real_len); 3036 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 3037 unsigned int blocksize); 3038 extern int ext4_handle_dirty_dirent_node(handle_t *handle, 3039 struct inode *inode, 3040 struct buffer_head *bh); 3041 #define S_SHIFT 12 3042 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3043 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3044 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3045 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3046 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3047 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3048 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3049 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3050 }; 3051 3052 static inline void ext4_set_de_type(struct super_block *sb, 3053 struct ext4_dir_entry_2 *de, 3054 umode_t mode) { 3055 if (ext4_has_feature_filetype(sb)) 3056 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3057 } 3058 3059 /* readpages.c */ 3060 extern int ext4_mpage_readpages(struct address_space *mapping, 3061 struct list_head *pages, struct page *page, 3062 unsigned nr_pages); 3063 3064 /* symlink.c */ 3065 extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3066 extern const struct inode_operations ext4_symlink_inode_operations; 3067 extern const struct inode_operations ext4_fast_symlink_inode_operations; 3068 3069 /* sysfs.c */ 3070 extern int ext4_register_sysfs(struct super_block *sb); 3071 extern void ext4_unregister_sysfs(struct super_block *sb); 3072 extern int __init ext4_init_sysfs(void); 3073 extern void ext4_exit_sysfs(void); 3074 3075 /* block_validity */ 3076 extern void ext4_release_system_zone(struct super_block *sb); 3077 extern int ext4_setup_system_zone(struct super_block *sb); 3078 extern int __init ext4_init_system_zone(void); 3079 extern void ext4_exit_system_zone(void); 3080 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 3081 ext4_fsblk_t start_blk, 3082 unsigned int count); 3083 extern int ext4_check_blockref(const char *, unsigned int, 3084 struct inode *, __le32 *, unsigned int); 3085 3086 /* extents.c */ 3087 struct ext4_ext_path; 3088 struct ext4_extent; 3089 3090 /* 3091 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3092 * __le32. 3093 */ 3094 #define EXT_MAX_BLOCKS 0xffffffff 3095 3096 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 3097 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 3098 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3099 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3100 struct ext4_map_blocks *map, int flags); 3101 extern int ext4_ext_truncate(handle_t *, struct inode *); 3102 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3103 ext4_lblk_t end); 3104 extern void ext4_ext_init(struct super_block *); 3105 extern void ext4_ext_release(struct super_block *); 3106 extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3107 loff_t len); 3108 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3109 loff_t offset, ssize_t len); 3110 extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3111 struct ext4_map_blocks *map, int flags); 3112 extern int ext4_ext_calc_metadata_amount(struct inode *inode, 3113 ext4_lblk_t lblocks); 3114 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3115 int num, 3116 struct ext4_ext_path *path); 3117 extern int ext4_can_extents_be_merged(struct inode *inode, 3118 struct ext4_extent *ex1, 3119 struct ext4_extent *ex2); 3120 extern int ext4_ext_insert_extent(handle_t *, struct inode *, 3121 struct ext4_ext_path **, 3122 struct ext4_extent *, int); 3123 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3124 struct ext4_ext_path **, 3125 int flags); 3126 extern void ext4_ext_drop_refs(struct ext4_ext_path *); 3127 extern int ext4_ext_check_inode(struct inode *inode); 3128 extern int ext4_find_delalloc_range(struct inode *inode, 3129 ext4_lblk_t lblk_start, 3130 ext4_lblk_t lblk_end); 3131 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 3132 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3133 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3134 __u64 start, __u64 len); 3135 extern int ext4_ext_precache(struct inode *inode); 3136 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len); 3137 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len); 3138 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3139 struct inode *inode2, ext4_lblk_t lblk1, 3140 ext4_lblk_t lblk2, ext4_lblk_t count, 3141 int mark_unwritten,int *err); 3142 3143 /* move_extent.c */ 3144 extern void ext4_double_down_write_data_sem(struct inode *first, 3145 struct inode *second); 3146 extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3147 struct inode *donor_inode); 3148 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3149 __u64 start_orig, __u64 start_donor, 3150 __u64 len, __u64 *moved_len); 3151 3152 /* page-io.c */ 3153 extern int __init ext4_init_pageio(void); 3154 extern void ext4_exit_pageio(void); 3155 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3156 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3157 extern int ext4_put_io_end(ext4_io_end_t *io_end); 3158 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3159 extern void ext4_io_submit_init(struct ext4_io_submit *io, 3160 struct writeback_control *wbc); 3161 extern void ext4_end_io_rsv_work(struct work_struct *work); 3162 extern void ext4_io_submit(struct ext4_io_submit *io); 3163 extern int ext4_bio_write_page(struct ext4_io_submit *io, 3164 struct page *page, 3165 int len, 3166 struct writeback_control *wbc, 3167 bool keep_towrite); 3168 3169 /* mmp.c */ 3170 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3171 3172 /* 3173 * Add new method to test whether block and inode bitmaps are properly 3174 * initialized. With uninit_bg reading the block from disk is not enough 3175 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3176 */ 3177 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3178 3179 static inline int bitmap_uptodate(struct buffer_head *bh) 3180 { 3181 return (buffer_uptodate(bh) && 3182 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3183 } 3184 static inline void set_bitmap_uptodate(struct buffer_head *bh) 3185 { 3186 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3187 } 3188 3189 /* 3190 * Disable DIO read nolock optimization, so new dioreaders will be forced 3191 * to grab i_mutex 3192 */ 3193 static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 3194 { 3195 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3196 smp_mb(); 3197 } 3198 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 3199 { 3200 smp_mb(); 3201 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 3202 } 3203 3204 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 3205 3206 /* For ioend & aio unwritten conversion wait queues */ 3207 #define EXT4_WQ_HASH_SZ 37 3208 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 3209 EXT4_WQ_HASH_SZ]) 3210 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 3211 3212 extern int ext4_resize_begin(struct super_block *sb); 3213 extern void ext4_resize_end(struct super_block *sb); 3214 3215 static inline void ext4_set_io_unwritten_flag(struct inode *inode, 3216 struct ext4_io_end *io_end) 3217 { 3218 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 3219 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3220 atomic_inc(&EXT4_I(inode)->i_unwritten); 3221 } 3222 } 3223 3224 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3225 { 3226 struct inode *inode = io_end->inode; 3227 3228 if (io_end->flag & EXT4_IO_END_UNWRITTEN) { 3229 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3230 /* Wake up anyone waiting on unwritten extent conversion */ 3231 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten)) 3232 wake_up_all(ext4_ioend_wq(inode)); 3233 } 3234 } 3235 3236 extern const struct iomap_ops ext4_iomap_ops; 3237 3238 #endif /* __KERNEL__ */ 3239 3240 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 3241 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3242 3243 #endif /* _EXT4_H */ 3244