1 /* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16 #ifndef _EXT4_H 17 #define _EXT4_H 18 19 #include <linux/types.h> 20 #include <linux/blkdev.h> 21 #include <linux/magic.h> 22 #include <linux/jbd2.h> 23 #include <linux/quota.h> 24 #include <linux/rwsem.h> 25 #include <linux/rbtree.h> 26 #include <linux/seqlock.h> 27 #include <linux/mutex.h> 28 #include <linux/timer.h> 29 #include <linux/wait.h> 30 #include <linux/blockgroup_lock.h> 31 #include <linux/percpu_counter.h> 32 33 /* 34 * The fourth extended filesystem constants/structures 35 */ 36 37 /* 38 * Define EXT4FS_DEBUG to produce debug messages 39 */ 40 #undef EXT4FS_DEBUG 41 42 /* 43 * Debug code 44 */ 45 #ifdef EXT4FS_DEBUG 46 #define ext4_debug(f, a...) \ 47 do { \ 48 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 49 __FILE__, __LINE__, __func__); \ 50 printk(KERN_DEBUG f, ## a); \ 51 } while (0) 52 #else 53 #define ext4_debug(f, a...) do {} while (0) 54 #endif 55 56 /* data type for block offset of block group */ 57 typedef int ext4_grpblk_t; 58 59 /* data type for filesystem-wide blocks number */ 60 typedef unsigned long long ext4_fsblk_t; 61 62 /* data type for file logical block number */ 63 typedef __u32 ext4_lblk_t; 64 65 /* data type for block group number */ 66 typedef unsigned int ext4_group_t; 67 68 /* 69 * Flags used in mballoc's allocation_context flags field. 70 * 71 * Also used to show what's going on for debugging purposes when the 72 * flag field is exported via the traceport interface 73 */ 74 75 /* prefer goal again. length */ 76 #define EXT4_MB_HINT_MERGE 0x0001 77 /* blocks already reserved */ 78 #define EXT4_MB_HINT_RESERVED 0x0002 79 /* metadata is being allocated */ 80 #define EXT4_MB_HINT_METADATA 0x0004 81 /* first blocks in the file */ 82 #define EXT4_MB_HINT_FIRST 0x0008 83 /* search for the best chunk */ 84 #define EXT4_MB_HINT_BEST 0x0010 85 /* data is being allocated */ 86 #define EXT4_MB_HINT_DATA 0x0020 87 /* don't preallocate (for tails) */ 88 #define EXT4_MB_HINT_NOPREALLOC 0x0040 89 /* allocate for locality group */ 90 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080 91 /* allocate goal blocks or none */ 92 #define EXT4_MB_HINT_GOAL_ONLY 0x0100 93 /* goal is meaningful */ 94 #define EXT4_MB_HINT_TRY_GOAL 0x0200 95 /* blocks already pre-reserved by delayed allocation */ 96 #define EXT4_MB_DELALLOC_RESERVED 0x0400 97 /* We are doing stream allocation */ 98 #define EXT4_MB_STREAM_ALLOC 0x0800 99 100 101 struct ext4_allocation_request { 102 /* target inode for block we're allocating */ 103 struct inode *inode; 104 /* how many blocks we want to allocate */ 105 unsigned int len; 106 /* logical block in target inode */ 107 ext4_lblk_t logical; 108 /* the closest logical allocated block to the left */ 109 ext4_lblk_t lleft; 110 /* the closest logical allocated block to the right */ 111 ext4_lblk_t lright; 112 /* phys. target (a hint) */ 113 ext4_fsblk_t goal; 114 /* phys. block for the closest logical allocated block to the left */ 115 ext4_fsblk_t pleft; 116 /* phys. block for the closest logical allocated block to the right */ 117 ext4_fsblk_t pright; 118 /* flags. see above EXT4_MB_HINT_* */ 119 unsigned int flags; 120 }; 121 122 /* 123 * For delayed allocation tracking 124 */ 125 struct mpage_da_data { 126 struct inode *inode; 127 sector_t b_blocknr; /* start block number of extent */ 128 size_t b_size; /* size of extent */ 129 unsigned long b_state; /* state of the extent */ 130 unsigned long first_page, next_page; /* extent of pages */ 131 struct writeback_control *wbc; 132 int io_done; 133 int pages_written; 134 int retval; 135 }; 136 #define EXT4_IO_UNWRITTEN 0x1 137 typedef struct ext4_io_end { 138 struct list_head list; /* per-file finished AIO list */ 139 struct inode *inode; /* file being written to */ 140 unsigned int flag; /* unwritten or not */ 141 struct page *page; /* page struct for buffer write */ 142 loff_t offset; /* offset in the file */ 143 ssize_t size; /* size of the extent */ 144 struct work_struct work; /* data work queue */ 145 } ext4_io_end_t; 146 147 /* 148 * Special inodes numbers 149 */ 150 #define EXT4_BAD_INO 1 /* Bad blocks inode */ 151 #define EXT4_ROOT_INO 2 /* Root inode */ 152 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 153 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 154 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 155 #define EXT4_JOURNAL_INO 8 /* Journal inode */ 156 157 /* First non-reserved inode for old ext4 filesystems */ 158 #define EXT4_GOOD_OLD_FIRST_INO 11 159 160 /* 161 * Maximal count of links to a file 162 */ 163 #define EXT4_LINK_MAX 65000 164 165 /* 166 * Macro-instructions used to manage several block sizes 167 */ 168 #define EXT4_MIN_BLOCK_SIZE 1024 169 #define EXT4_MAX_BLOCK_SIZE 65536 170 #define EXT4_MIN_BLOCK_LOG_SIZE 10 171 #ifdef __KERNEL__ 172 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 173 #else 174 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 175 #endif 176 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 177 #ifdef __KERNEL__ 178 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 179 #else 180 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 181 #endif 182 #ifdef __KERNEL__ 183 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 184 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 185 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 186 #else 187 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 188 EXT4_GOOD_OLD_INODE_SIZE : \ 189 (s)->s_inode_size) 190 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 191 EXT4_GOOD_OLD_FIRST_INO : \ 192 (s)->s_first_ino) 193 #endif 194 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 195 196 /* 197 * Structure of a blocks group descriptor 198 */ 199 struct ext4_group_desc 200 { 201 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 202 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 203 __le32 bg_inode_table_lo; /* Inodes table block */ 204 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 205 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 206 __le16 bg_used_dirs_count_lo; /* Directories count */ 207 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 208 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */ 209 __le16 bg_itable_unused_lo; /* Unused inodes count */ 210 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 211 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 212 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 213 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 214 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 215 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 216 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 217 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 218 __u32 bg_reserved2[3]; 219 }; 220 221 /* 222 * Structure of a flex block group info 223 */ 224 225 struct flex_groups { 226 atomic_t free_inodes; 227 atomic_t free_blocks; 228 atomic_t used_dirs; 229 }; 230 231 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 232 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 233 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 234 235 /* 236 * Macro-instructions used to manage group descriptors 237 */ 238 #define EXT4_MIN_DESC_SIZE 32 239 #define EXT4_MIN_DESC_SIZE_64BIT 64 240 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 241 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 242 #ifdef __KERNEL__ 243 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 244 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 245 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 246 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 247 #else 248 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 249 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 250 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 251 #endif 252 253 /* 254 * Constants relative to the data blocks 255 */ 256 #define EXT4_NDIR_BLOCKS 12 257 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 258 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 259 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 260 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 261 262 /* 263 * Inode flags 264 */ 265 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 266 #define EXT4_UNRM_FL 0x00000002 /* Undelete */ 267 #define EXT4_COMPR_FL 0x00000004 /* Compress file */ 268 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 269 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 270 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 271 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 272 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 273 /* Reserved for compression usage... */ 274 #define EXT4_DIRTY_FL 0x00000100 275 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 276 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 277 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 278 /* End compression flags --- maybe not all used */ 279 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 280 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 281 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 282 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 283 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 284 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 285 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 286 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 287 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 288 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 289 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 290 291 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 292 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */ 293 294 /* Flags that should be inherited by new inodes from their parent. */ 295 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 296 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\ 297 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 298 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 299 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 300 301 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 302 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 303 304 /* Flags that are appropriate for non-directories/regular files. */ 305 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 306 307 /* Mask out flags that are inappropriate for the given type of inode. */ 308 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 309 { 310 if (S_ISDIR(mode)) 311 return flags; 312 else if (S_ISREG(mode)) 313 return flags & EXT4_REG_FLMASK; 314 else 315 return flags & EXT4_OTHER_FLMASK; 316 } 317 318 /* Used to pass group descriptor data when online resize is done */ 319 struct ext4_new_group_input { 320 __u32 group; /* Group number for this data */ 321 __u64 block_bitmap; /* Absolute block number of block bitmap */ 322 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 323 __u64 inode_table; /* Absolute block number of inode table start */ 324 __u32 blocks_count; /* Total number of blocks in this group */ 325 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 326 __u16 unused; 327 }; 328 329 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 330 struct ext4_new_group_data { 331 __u32 group; 332 __u64 block_bitmap; 333 __u64 inode_bitmap; 334 __u64 inode_table; 335 __u32 blocks_count; 336 __u16 reserved_blocks; 337 __u16 unused; 338 __u32 free_blocks_count; 339 }; 340 341 /* 342 * Flags used by ext4_get_blocks() 343 */ 344 /* Allocate any needed blocks and/or convert an unitialized 345 extent to be an initialized ext4 */ 346 #define EXT4_GET_BLOCKS_CREATE 0x0001 347 /* Request the creation of an unitialized extent */ 348 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 349 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 350 EXT4_GET_BLOCKS_CREATE) 351 /* Caller is from the delayed allocation writeout path, 352 so set the magic i_delalloc_reserve_flag after taking the 353 inode allocation semaphore for */ 354 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 355 /* caller is from the direct IO path, request to creation of an 356 unitialized extents if not allocated, split the uninitialized 357 extent if blocks has been preallocated already*/ 358 #define EXT4_GET_BLOCKS_PRE_IO 0x0008 359 #define EXT4_GET_BLOCKS_CONVERT 0x0010 360 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 361 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 362 /* Convert extent to initialized after IO complete */ 363 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 364 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 365 366 /* 367 * Flags used by ext4_free_blocks 368 */ 369 #define EXT4_FREE_BLOCKS_METADATA 0x0001 370 #define EXT4_FREE_BLOCKS_FORGET 0x0002 371 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 372 373 /* 374 * ioctl commands 375 */ 376 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 377 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 378 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 379 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 380 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 381 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 382 #ifdef CONFIG_JBD2_DEBUG 383 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long) 384 #endif 385 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 386 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 387 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 388 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 389 #define EXT4_IOC_MIGRATE _IO('f', 9) 390 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 391 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 392 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 393 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 394 395 /* 396 * ioctl commands in 32 bit emulation 397 */ 398 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 399 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 400 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 401 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 402 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 403 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 404 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 405 #ifdef CONFIG_JBD2_DEBUG 406 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int) 407 #endif 408 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 409 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 410 411 412 /* 413 * Mount options 414 */ 415 struct ext4_mount_options { 416 unsigned long s_mount_opt; 417 uid_t s_resuid; 418 gid_t s_resgid; 419 unsigned long s_commit_interval; 420 u32 s_min_batch_time, s_max_batch_time; 421 #ifdef CONFIG_QUOTA 422 int s_jquota_fmt; 423 char *s_qf_names[MAXQUOTAS]; 424 #endif 425 }; 426 427 /* Max physical block we can addres w/o extents */ 428 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 429 430 /* 431 * Structure of an inode on the disk 432 */ 433 struct ext4_inode { 434 __le16 i_mode; /* File mode */ 435 __le16 i_uid; /* Low 16 bits of Owner Uid */ 436 __le32 i_size_lo; /* Size in bytes */ 437 __le32 i_atime; /* Access time */ 438 __le32 i_ctime; /* Inode Change time */ 439 __le32 i_mtime; /* Modification time */ 440 __le32 i_dtime; /* Deletion Time */ 441 __le16 i_gid; /* Low 16 bits of Group Id */ 442 __le16 i_links_count; /* Links count */ 443 __le32 i_blocks_lo; /* Blocks count */ 444 __le32 i_flags; /* File flags */ 445 union { 446 struct { 447 __le32 l_i_version; 448 } linux1; 449 struct { 450 __u32 h_i_translator; 451 } hurd1; 452 struct { 453 __u32 m_i_reserved1; 454 } masix1; 455 } osd1; /* OS dependent 1 */ 456 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 457 __le32 i_generation; /* File version (for NFS) */ 458 __le32 i_file_acl_lo; /* File ACL */ 459 __le32 i_size_high; 460 __le32 i_obso_faddr; /* Obsoleted fragment address */ 461 union { 462 struct { 463 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 464 __le16 l_i_file_acl_high; 465 __le16 l_i_uid_high; /* these 2 fields */ 466 __le16 l_i_gid_high; /* were reserved2[0] */ 467 __u32 l_i_reserved2; 468 } linux2; 469 struct { 470 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 471 __u16 h_i_mode_high; 472 __u16 h_i_uid_high; 473 __u16 h_i_gid_high; 474 __u32 h_i_author; 475 } hurd2; 476 struct { 477 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 478 __le16 m_i_file_acl_high; 479 __u32 m_i_reserved2[2]; 480 } masix2; 481 } osd2; /* OS dependent 2 */ 482 __le16 i_extra_isize; 483 __le16 i_pad1; 484 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 485 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 486 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 487 __le32 i_crtime; /* File Creation time */ 488 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 489 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 490 }; 491 492 struct move_extent { 493 __u32 reserved; /* should be zero */ 494 __u32 donor_fd; /* donor file descriptor */ 495 __u64 orig_start; /* logical start offset in block for orig */ 496 __u64 donor_start; /* logical start offset in block for donor */ 497 __u64 len; /* block length to be moved */ 498 __u64 moved_len; /* moved block length */ 499 }; 500 501 #define EXT4_EPOCH_BITS 2 502 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 503 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 504 505 /* 506 * Extended fields will fit into an inode if the filesystem was formatted 507 * with large inodes (-I 256 or larger) and there are not currently any EAs 508 * consuming all of the available space. For new inodes we always reserve 509 * enough space for the kernel's known extended fields, but for inodes 510 * created with an old kernel this might not have been the case. None of 511 * the extended inode fields is critical for correct filesystem operation. 512 * This macro checks if a certain field fits in the inode. Note that 513 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 514 */ 515 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 516 ((offsetof(typeof(*ext4_inode), field) + \ 517 sizeof((ext4_inode)->field)) \ 518 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 519 (einode)->i_extra_isize)) \ 520 521 static inline __le32 ext4_encode_extra_time(struct timespec *time) 522 { 523 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 524 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 525 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 526 } 527 528 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 529 { 530 if (sizeof(time->tv_sec) > 4) 531 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 532 << 32; 533 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 534 } 535 536 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 537 do { \ 538 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 539 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 540 (raw_inode)->xtime ## _extra = \ 541 ext4_encode_extra_time(&(inode)->xtime); \ 542 } while (0) 543 544 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 545 do { \ 546 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 547 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 548 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 549 (raw_inode)->xtime ## _extra = \ 550 ext4_encode_extra_time(&(einode)->xtime); \ 551 } while (0) 552 553 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 554 do { \ 555 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 556 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 557 ext4_decode_extra_time(&(inode)->xtime, \ 558 raw_inode->xtime ## _extra); \ 559 } while (0) 560 561 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 562 do { \ 563 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 564 (einode)->xtime.tv_sec = \ 565 (signed)le32_to_cpu((raw_inode)->xtime); \ 566 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 567 ext4_decode_extra_time(&(einode)->xtime, \ 568 raw_inode->xtime ## _extra); \ 569 } while (0) 570 571 #define i_disk_version osd1.linux1.l_i_version 572 573 #if defined(__KERNEL__) || defined(__linux__) 574 #define i_reserved1 osd1.linux1.l_i_reserved1 575 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 576 #define i_blocks_high osd2.linux2.l_i_blocks_high 577 #define i_uid_low i_uid 578 #define i_gid_low i_gid 579 #define i_uid_high osd2.linux2.l_i_uid_high 580 #define i_gid_high osd2.linux2.l_i_gid_high 581 #define i_reserved2 osd2.linux2.l_i_reserved2 582 583 #elif defined(__GNU__) 584 585 #define i_translator osd1.hurd1.h_i_translator 586 #define i_uid_high osd2.hurd2.h_i_uid_high 587 #define i_gid_high osd2.hurd2.h_i_gid_high 588 #define i_author osd2.hurd2.h_i_author 589 590 #elif defined(__masix__) 591 592 #define i_reserved1 osd1.masix1.m_i_reserved1 593 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 594 #define i_reserved2 osd2.masix2.m_i_reserved2 595 596 #endif /* defined(__KERNEL__) || defined(__linux__) */ 597 598 /* 599 * storage for cached extent 600 */ 601 struct ext4_ext_cache { 602 ext4_fsblk_t ec_start; 603 ext4_lblk_t ec_block; 604 __u32 ec_len; /* must be 32bit to return holes */ 605 __u32 ec_type; 606 }; 607 608 /* 609 * fourth extended file system inode data in memory 610 */ 611 struct ext4_inode_info { 612 __le32 i_data[15]; /* unconverted */ 613 __u32 i_flags; 614 ext4_fsblk_t i_file_acl; 615 __u32 i_dtime; 616 617 /* 618 * i_block_group is the number of the block group which contains 619 * this file's inode. Constant across the lifetime of the inode, 620 * it is ued for making block allocation decisions - we try to 621 * place a file's data blocks near its inode block, and new inodes 622 * near to their parent directory's inode. 623 */ 624 ext4_group_t i_block_group; 625 unsigned long i_state_flags; /* Dynamic state flags */ 626 627 ext4_lblk_t i_dir_start_lookup; 628 #ifdef CONFIG_EXT4_FS_XATTR 629 /* 630 * Extended attributes can be read independently of the main file 631 * data. Taking i_mutex even when reading would cause contention 632 * between readers of EAs and writers of regular file data, so 633 * instead we synchronize on xattr_sem when reading or changing 634 * EAs. 635 */ 636 struct rw_semaphore xattr_sem; 637 #endif 638 639 struct list_head i_orphan; /* unlinked but open inodes */ 640 641 /* 642 * i_disksize keeps track of what the inode size is ON DISK, not 643 * in memory. During truncate, i_size is set to the new size by 644 * the VFS prior to calling ext4_truncate(), but the filesystem won't 645 * set i_disksize to 0 until the truncate is actually under way. 646 * 647 * The intent is that i_disksize always represents the blocks which 648 * are used by this file. This allows recovery to restart truncate 649 * on orphans if we crash during truncate. We actually write i_disksize 650 * into the on-disk inode when writing inodes out, instead of i_size. 651 * 652 * The only time when i_disksize and i_size may be different is when 653 * a truncate is in progress. The only things which change i_disksize 654 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 655 */ 656 loff_t i_disksize; 657 658 /* 659 * i_data_sem is for serialising ext4_truncate() against 660 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 661 * data tree are chopped off during truncate. We can't do that in 662 * ext4 because whenever we perform intermediate commits during 663 * truncate, the inode and all the metadata blocks *must* be in a 664 * consistent state which allows truncation of the orphans to restart 665 * during recovery. Hence we must fix the get_block-vs-truncate race 666 * by other means, so we have i_data_sem. 667 */ 668 struct rw_semaphore i_data_sem; 669 struct inode vfs_inode; 670 struct jbd2_inode jinode; 671 672 struct ext4_ext_cache i_cached_extent; 673 /* 674 * File creation time. Its function is same as that of 675 * struct timespec i_{a,c,m}time in the generic inode. 676 */ 677 struct timespec i_crtime; 678 679 /* mballoc */ 680 struct list_head i_prealloc_list; 681 spinlock_t i_prealloc_lock; 682 683 /* ialloc */ 684 ext4_group_t i_last_alloc_group; 685 686 /* allocation reservation info for delalloc */ 687 unsigned int i_reserved_data_blocks; 688 unsigned int i_reserved_meta_blocks; 689 unsigned int i_allocated_meta_blocks; 690 unsigned short i_delalloc_reserved_flag; 691 sector_t i_da_metadata_calc_last_lblock; 692 int i_da_metadata_calc_len; 693 694 /* on-disk additional length */ 695 __u16 i_extra_isize; 696 697 spinlock_t i_block_reservation_lock; 698 #ifdef CONFIG_QUOTA 699 /* quota space reservation, managed internally by quota code */ 700 qsize_t i_reserved_quota; 701 #endif 702 703 /* completed IOs that might need unwritten extents handling */ 704 struct list_head i_completed_io_list; 705 spinlock_t i_completed_io_lock; 706 /* current io_end structure for async DIO write*/ 707 ext4_io_end_t *cur_aio_dio; 708 709 /* 710 * Transactions that contain inode's metadata needed to complete 711 * fsync and fdatasync, respectively. 712 */ 713 tid_t i_sync_tid; 714 tid_t i_datasync_tid; 715 }; 716 717 /* 718 * File system states 719 */ 720 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 721 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 722 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 723 724 /* 725 * Misc. filesystem flags 726 */ 727 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 728 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 729 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 730 731 /* 732 * Mount flags 733 */ 734 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */ 735 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 736 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 737 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 738 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 739 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 740 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 741 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 742 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 743 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 744 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 745 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 746 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 747 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 748 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 749 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 750 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 751 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 752 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */ 753 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 754 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 755 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 756 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 757 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 758 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 759 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */ 760 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 761 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 762 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 763 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 764 765 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt 766 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt 767 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 768 EXT4_MOUNT_##opt) 769 770 #define ext4_set_bit ext2_set_bit 771 #define ext4_set_bit_atomic ext2_set_bit_atomic 772 #define ext4_clear_bit ext2_clear_bit 773 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 774 #define ext4_test_bit ext2_test_bit 775 #define ext4_find_first_zero_bit ext2_find_first_zero_bit 776 #define ext4_find_next_zero_bit ext2_find_next_zero_bit 777 #define ext4_find_next_bit ext2_find_next_bit 778 779 /* 780 * Maximal mount counts between two filesystem checks 781 */ 782 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 783 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 784 785 /* 786 * Behaviour when detecting errors 787 */ 788 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 789 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 790 #define EXT4_ERRORS_PANIC 3 /* Panic */ 791 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 792 793 /* 794 * Structure of the super block 795 */ 796 struct ext4_super_block { 797 /*00*/ __le32 s_inodes_count; /* Inodes count */ 798 __le32 s_blocks_count_lo; /* Blocks count */ 799 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 800 __le32 s_free_blocks_count_lo; /* Free blocks count */ 801 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 802 __le32 s_first_data_block; /* First Data Block */ 803 __le32 s_log_block_size; /* Block size */ 804 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */ 805 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 806 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */ 807 __le32 s_inodes_per_group; /* # Inodes per group */ 808 __le32 s_mtime; /* Mount time */ 809 /*30*/ __le32 s_wtime; /* Write time */ 810 __le16 s_mnt_count; /* Mount count */ 811 __le16 s_max_mnt_count; /* Maximal mount count */ 812 __le16 s_magic; /* Magic signature */ 813 __le16 s_state; /* File system state */ 814 __le16 s_errors; /* Behaviour when detecting errors */ 815 __le16 s_minor_rev_level; /* minor revision level */ 816 /*40*/ __le32 s_lastcheck; /* time of last check */ 817 __le32 s_checkinterval; /* max. time between checks */ 818 __le32 s_creator_os; /* OS */ 819 __le32 s_rev_level; /* Revision level */ 820 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 821 __le16 s_def_resgid; /* Default gid for reserved blocks */ 822 /* 823 * These fields are for EXT4_DYNAMIC_REV superblocks only. 824 * 825 * Note: the difference between the compatible feature set and 826 * the incompatible feature set is that if there is a bit set 827 * in the incompatible feature set that the kernel doesn't 828 * know about, it should refuse to mount the filesystem. 829 * 830 * e2fsck's requirements are more strict; if it doesn't know 831 * about a feature in either the compatible or incompatible 832 * feature set, it must abort and not try to meddle with 833 * things it doesn't understand... 834 */ 835 __le32 s_first_ino; /* First non-reserved inode */ 836 __le16 s_inode_size; /* size of inode structure */ 837 __le16 s_block_group_nr; /* block group # of this superblock */ 838 __le32 s_feature_compat; /* compatible feature set */ 839 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 840 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 841 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 842 /*78*/ char s_volume_name[16]; /* volume name */ 843 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 844 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 845 /* 846 * Performance hints. Directory preallocation should only 847 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 848 */ 849 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 850 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 851 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 852 /* 853 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 854 */ 855 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 856 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 857 __le32 s_journal_dev; /* device number of journal file */ 858 __le32 s_last_orphan; /* start of list of inodes to delete */ 859 __le32 s_hash_seed[4]; /* HTREE hash seed */ 860 __u8 s_def_hash_version; /* Default hash version to use */ 861 __u8 s_reserved_char_pad; 862 __le16 s_desc_size; /* size of group descriptor */ 863 /*100*/ __le32 s_default_mount_opts; 864 __le32 s_first_meta_bg; /* First metablock block group */ 865 __le32 s_mkfs_time; /* When the filesystem was created */ 866 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 867 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 868 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 869 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 870 __le32 s_free_blocks_count_hi; /* Free blocks count */ 871 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 872 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 873 __le32 s_flags; /* Miscellaneous flags */ 874 __le16 s_raid_stride; /* RAID stride */ 875 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */ 876 __le64 s_mmp_block; /* Block for multi-mount protection */ 877 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 878 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 879 __u8 s_reserved_char_pad2; 880 __le16 s_reserved_pad; 881 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 882 __u32 s_reserved[160]; /* Padding to the end of the block */ 883 }; 884 885 #ifdef __KERNEL__ 886 887 /* 888 * run-time mount flags 889 */ 890 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 891 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 892 893 /* 894 * fourth extended-fs super-block data in memory 895 */ 896 struct ext4_sb_info { 897 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 898 unsigned long s_inodes_per_block;/* Number of inodes per block */ 899 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 900 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 901 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 902 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 903 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 904 ext4_group_t s_groups_count; /* Number of groups in the fs */ 905 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 906 unsigned long s_overhead_last; /* Last calculated overhead */ 907 unsigned long s_blocks_last; /* Last seen block count */ 908 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 909 struct buffer_head * s_sbh; /* Buffer containing the super block */ 910 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 911 struct buffer_head **s_group_desc; 912 unsigned int s_mount_opt; 913 unsigned int s_mount_flags; 914 ext4_fsblk_t s_sb_block; 915 uid_t s_resuid; 916 gid_t s_resgid; 917 unsigned short s_mount_state; 918 unsigned short s_pad; 919 int s_addr_per_block_bits; 920 int s_desc_per_block_bits; 921 int s_inode_size; 922 int s_first_ino; 923 unsigned int s_inode_readahead_blks; 924 unsigned int s_inode_goal; 925 spinlock_t s_next_gen_lock; 926 u32 s_next_generation; 927 u32 s_hash_seed[4]; 928 int s_def_hash_version; 929 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 930 struct percpu_counter s_freeblocks_counter; 931 struct percpu_counter s_freeinodes_counter; 932 struct percpu_counter s_dirs_counter; 933 struct percpu_counter s_dirtyblocks_counter; 934 struct blockgroup_lock *s_blockgroup_lock; 935 struct proc_dir_entry *s_proc; 936 struct kobject s_kobj; 937 struct completion s_kobj_unregister; 938 939 /* Journaling */ 940 struct inode *s_journal_inode; 941 struct journal_s *s_journal; 942 struct list_head s_orphan; 943 struct mutex s_orphan_lock; 944 struct mutex s_resize_lock; 945 unsigned long s_commit_interval; 946 u32 s_max_batch_time; 947 u32 s_min_batch_time; 948 struct block_device *journal_bdev; 949 #ifdef CONFIG_JBD2_DEBUG 950 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */ 951 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */ 952 #endif 953 #ifdef CONFIG_QUOTA 954 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 955 int s_jquota_fmt; /* Format of quota to use */ 956 #endif 957 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 958 struct rb_root system_blks; 959 960 #ifdef EXTENTS_STATS 961 /* ext4 extents stats */ 962 unsigned long s_ext_min; 963 unsigned long s_ext_max; 964 unsigned long s_depth_max; 965 spinlock_t s_ext_stats_lock; 966 unsigned long s_ext_blocks; 967 unsigned long s_ext_extents; 968 #endif 969 970 /* for buddy allocator */ 971 struct ext4_group_info ***s_group_info; 972 struct inode *s_buddy_cache; 973 long s_blocks_reserved; 974 spinlock_t s_reserve_lock; 975 spinlock_t s_md_lock; 976 tid_t s_last_transaction; 977 unsigned short *s_mb_offsets; 978 unsigned int *s_mb_maxs; 979 980 /* tunables */ 981 unsigned long s_stripe; 982 unsigned int s_mb_stream_request; 983 unsigned int s_mb_max_to_scan; 984 unsigned int s_mb_min_to_scan; 985 unsigned int s_mb_stats; 986 unsigned int s_mb_order2_reqs; 987 unsigned int s_mb_group_prealloc; 988 unsigned int s_max_writeback_mb_bump; 989 /* where last allocation was done - for stream allocation */ 990 unsigned long s_mb_last_group; 991 unsigned long s_mb_last_start; 992 993 /* stats for buddy allocator */ 994 spinlock_t s_mb_pa_lock; 995 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 996 atomic_t s_bal_success; /* we found long enough chunks */ 997 atomic_t s_bal_allocated; /* in blocks */ 998 atomic_t s_bal_ex_scanned; /* total extents scanned */ 999 atomic_t s_bal_goals; /* goal hits */ 1000 atomic_t s_bal_breaks; /* too long searches */ 1001 atomic_t s_bal_2orders; /* 2^order hits */ 1002 spinlock_t s_bal_lock; 1003 unsigned long s_mb_buddies_generated; 1004 unsigned long long s_mb_generation_time; 1005 atomic_t s_mb_lost_chunks; 1006 atomic_t s_mb_preallocated; 1007 atomic_t s_mb_discarded; 1008 atomic_t s_lock_busy; 1009 1010 /* locality groups */ 1011 struct ext4_locality_group *s_locality_groups; 1012 1013 /* for write statistics */ 1014 unsigned long s_sectors_written_start; 1015 u64 s_kbytes_written; 1016 1017 unsigned int s_log_groups_per_flex; 1018 struct flex_groups *s_flex_groups; 1019 1020 /* workqueue for dio unwritten */ 1021 struct workqueue_struct *dio_unwritten_wq; 1022 }; 1023 1024 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1025 { 1026 return sb->s_fs_info; 1027 } 1028 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1029 { 1030 return container_of(inode, struct ext4_inode_info, vfs_inode); 1031 } 1032 1033 static inline struct timespec ext4_current_time(struct inode *inode) 1034 { 1035 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1036 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1037 } 1038 1039 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1040 { 1041 return ino == EXT4_ROOT_INO || 1042 ino == EXT4_JOURNAL_INO || 1043 ino == EXT4_RESIZE_INO || 1044 (ino >= EXT4_FIRST_INO(sb) && 1045 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1046 } 1047 1048 /* 1049 * Inode dynamic state flags 1050 */ 1051 enum { 1052 EXT4_STATE_JDATA, /* journaled data exists */ 1053 EXT4_STATE_NEW, /* inode is newly created */ 1054 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1055 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1056 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1057 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1058 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1059 }; 1060 1061 static inline int ext4_test_inode_state(struct inode *inode, int bit) 1062 { 1063 return test_bit(bit, &EXT4_I(inode)->i_state_flags); 1064 } 1065 1066 static inline void ext4_set_inode_state(struct inode *inode, int bit) 1067 { 1068 set_bit(bit, &EXT4_I(inode)->i_state_flags); 1069 } 1070 1071 static inline void ext4_clear_inode_state(struct inode *inode, int bit) 1072 { 1073 clear_bit(bit, &EXT4_I(inode)->i_state_flags); 1074 } 1075 #else 1076 /* Assume that user mode programs are passing in an ext4fs superblock, not 1077 * a kernel struct super_block. This will allow us to call the feature-test 1078 * macros from user land. */ 1079 #define EXT4_SB(sb) (sb) 1080 #endif 1081 1082 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1083 1084 /* 1085 * Codes for operating systems 1086 */ 1087 #define EXT4_OS_LINUX 0 1088 #define EXT4_OS_HURD 1 1089 #define EXT4_OS_MASIX 2 1090 #define EXT4_OS_FREEBSD 3 1091 #define EXT4_OS_LITES 4 1092 1093 /* 1094 * Revision levels 1095 */ 1096 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1097 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1098 1099 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1100 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1101 1102 #define EXT4_GOOD_OLD_INODE_SIZE 128 1103 1104 /* 1105 * Feature set definitions 1106 */ 1107 1108 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1109 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1110 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1111 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1112 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1113 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1114 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1115 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1116 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1117 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1118 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1119 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1120 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1121 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1122 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1123 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1124 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1125 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1126 1127 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1128 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1129 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1130 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1131 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1132 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1133 1134 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1135 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1136 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1137 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1138 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1139 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1140 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1141 1142 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1143 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1144 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1145 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1146 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1147 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1148 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1149 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1150 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1151 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1152 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1153 1154 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1155 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1156 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1157 EXT4_FEATURE_INCOMPAT_META_BG| \ 1158 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1159 EXT4_FEATURE_INCOMPAT_64BIT| \ 1160 EXT4_FEATURE_INCOMPAT_FLEX_BG) 1161 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1162 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1163 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1164 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1165 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1166 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1167 EXT4_FEATURE_RO_COMPAT_HUGE_FILE) 1168 1169 /* 1170 * Default values for user and/or group using reserved blocks 1171 */ 1172 #define EXT4_DEF_RESUID 0 1173 #define EXT4_DEF_RESGID 0 1174 1175 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1176 1177 /* 1178 * Default mount options 1179 */ 1180 #define EXT4_DEFM_DEBUG 0x0001 1181 #define EXT4_DEFM_BSDGROUPS 0x0002 1182 #define EXT4_DEFM_XATTR_USER 0x0004 1183 #define EXT4_DEFM_ACL 0x0008 1184 #define EXT4_DEFM_UID16 0x0010 1185 #define EXT4_DEFM_JMODE 0x0060 1186 #define EXT4_DEFM_JMODE_DATA 0x0020 1187 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1188 #define EXT4_DEFM_JMODE_WBACK 0x0060 1189 1190 /* 1191 * Default journal batch times 1192 */ 1193 #define EXT4_DEF_MIN_BATCH_TIME 0 1194 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1195 1196 /* 1197 * Minimum number of groups in a flexgroup before we separate out 1198 * directories into the first block group of a flexgroup 1199 */ 1200 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1201 1202 /* 1203 * Structure of a directory entry 1204 */ 1205 #define EXT4_NAME_LEN 255 1206 1207 struct ext4_dir_entry { 1208 __le32 inode; /* Inode number */ 1209 __le16 rec_len; /* Directory entry length */ 1210 __le16 name_len; /* Name length */ 1211 char name[EXT4_NAME_LEN]; /* File name */ 1212 }; 1213 1214 /* 1215 * The new version of the directory entry. Since EXT4 structures are 1216 * stored in intel byte order, and the name_len field could never be 1217 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1218 * file_type field. 1219 */ 1220 struct ext4_dir_entry_2 { 1221 __le32 inode; /* Inode number */ 1222 __le16 rec_len; /* Directory entry length */ 1223 __u8 name_len; /* Name length */ 1224 __u8 file_type; 1225 char name[EXT4_NAME_LEN]; /* File name */ 1226 }; 1227 1228 /* 1229 * Ext4 directory file types. Only the low 3 bits are used. The 1230 * other bits are reserved for now. 1231 */ 1232 #define EXT4_FT_UNKNOWN 0 1233 #define EXT4_FT_REG_FILE 1 1234 #define EXT4_FT_DIR 2 1235 #define EXT4_FT_CHRDEV 3 1236 #define EXT4_FT_BLKDEV 4 1237 #define EXT4_FT_FIFO 5 1238 #define EXT4_FT_SOCK 6 1239 #define EXT4_FT_SYMLINK 7 1240 1241 #define EXT4_FT_MAX 8 1242 1243 /* 1244 * EXT4_DIR_PAD defines the directory entries boundaries 1245 * 1246 * NOTE: It must be a multiple of 4 1247 */ 1248 #define EXT4_DIR_PAD 4 1249 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1250 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1251 ~EXT4_DIR_ROUND) 1252 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1253 1254 /* 1255 * Hash Tree Directory indexing 1256 * (c) Daniel Phillips, 2001 1257 */ 1258 1259 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1260 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1261 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL)) 1262 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1263 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1264 1265 /* Legal values for the dx_root hash_version field: */ 1266 1267 #define DX_HASH_LEGACY 0 1268 #define DX_HASH_HALF_MD4 1 1269 #define DX_HASH_TEA 2 1270 #define DX_HASH_LEGACY_UNSIGNED 3 1271 #define DX_HASH_HALF_MD4_UNSIGNED 4 1272 #define DX_HASH_TEA_UNSIGNED 5 1273 1274 #ifdef __KERNEL__ 1275 1276 /* hash info structure used by the directory hash */ 1277 struct dx_hash_info 1278 { 1279 u32 hash; 1280 u32 minor_hash; 1281 int hash_version; 1282 u32 *seed; 1283 }; 1284 1285 #define EXT4_HTREE_EOF 0x7fffffff 1286 1287 /* 1288 * Control parameters used by ext4_htree_next_block 1289 */ 1290 #define HASH_NB_ALWAYS 1 1291 1292 1293 /* 1294 * Describe an inode's exact location on disk and in memory 1295 */ 1296 struct ext4_iloc 1297 { 1298 struct buffer_head *bh; 1299 unsigned long offset; 1300 ext4_group_t block_group; 1301 }; 1302 1303 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1304 { 1305 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1306 } 1307 1308 /* 1309 * This structure is stuffed into the struct file's private_data field 1310 * for directories. It is where we put information so that we can do 1311 * readdir operations in hash tree order. 1312 */ 1313 struct dir_private_info { 1314 struct rb_root root; 1315 struct rb_node *curr_node; 1316 struct fname *extra_fname; 1317 loff_t last_pos; 1318 __u32 curr_hash; 1319 __u32 curr_minor_hash; 1320 __u32 next_hash; 1321 }; 1322 1323 /* calculate the first block number of the group */ 1324 static inline ext4_fsblk_t 1325 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1326 { 1327 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1328 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1329 } 1330 1331 /* 1332 * Special error return code only used by dx_probe() and its callers. 1333 */ 1334 #define ERR_BAD_DX_DIR -75000 1335 1336 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1337 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1338 1339 extern struct proc_dir_entry *ext4_proc_root; 1340 1341 /* 1342 * Function prototypes 1343 */ 1344 1345 /* 1346 * Ok, these declarations are also in <linux/kernel.h> but none of the 1347 * ext4 source programs needs to include it so they are duplicated here. 1348 */ 1349 # define NORET_TYPE /**/ 1350 # define ATTRIB_NORET __attribute__((noreturn)) 1351 # define NORET_AND noreturn, 1352 1353 /* bitmap.c */ 1354 extern unsigned int ext4_count_free(struct buffer_head *, unsigned); 1355 1356 /* balloc.c */ 1357 extern unsigned int ext4_block_group(struct super_block *sb, 1358 ext4_fsblk_t blocknr); 1359 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1360 ext4_fsblk_t blocknr); 1361 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1362 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1363 ext4_group_t group); 1364 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1365 ext4_fsblk_t goal, unsigned long *count, int *errp); 1366 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1367 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1368 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb, 1369 ext4_fsblk_t block, unsigned long count); 1370 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *); 1371 extern void ext4_check_blocks_bitmap(struct super_block *); 1372 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1373 ext4_group_t block_group, 1374 struct buffer_head ** bh); 1375 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1376 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1377 ext4_group_t block_group); 1378 extern unsigned ext4_init_block_bitmap(struct super_block *sb, 1379 struct buffer_head *bh, 1380 ext4_group_t group, 1381 struct ext4_group_desc *desc); 1382 #define ext4_free_blocks_after_init(sb, group, desc) \ 1383 ext4_init_block_bitmap(sb, NULL, group, desc) 1384 1385 /* dir.c */ 1386 extern int ext4_check_dir_entry(const char *, struct inode *, 1387 struct ext4_dir_entry_2 *, 1388 struct buffer_head *, unsigned int); 1389 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1390 __u32 minor_hash, 1391 struct ext4_dir_entry_2 *dirent); 1392 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1393 1394 /* fsync.c */ 1395 extern int ext4_sync_file(struct file *, struct dentry *, int); 1396 1397 /* hash.c */ 1398 extern int ext4fs_dirhash(const char *name, int len, struct 1399 dx_hash_info *hinfo); 1400 1401 /* ialloc.c */ 1402 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int, 1403 const struct qstr *qstr, __u32 goal); 1404 extern void ext4_free_inode(handle_t *, struct inode *); 1405 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 1406 extern unsigned long ext4_count_free_inodes(struct super_block *); 1407 extern unsigned long ext4_count_dirs(struct super_block *); 1408 extern void ext4_check_inodes_bitmap(struct super_block *); 1409 extern unsigned ext4_init_inode_bitmap(struct super_block *sb, 1410 struct buffer_head *bh, 1411 ext4_group_t group, 1412 struct ext4_group_desc *desc); 1413 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 1414 1415 /* mballoc.c */ 1416 extern long ext4_mb_stats; 1417 extern long ext4_mb_max_to_scan; 1418 extern int ext4_mb_init(struct super_block *, int); 1419 extern int ext4_mb_release(struct super_block *); 1420 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 1421 struct ext4_allocation_request *, int *); 1422 extern int ext4_mb_reserve_blocks(struct super_block *, int); 1423 extern void ext4_discard_preallocations(struct inode *); 1424 extern int __init init_ext4_mballoc(void); 1425 extern void exit_ext4_mballoc(void); 1426 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 1427 struct buffer_head *bh, ext4_fsblk_t block, 1428 unsigned long count, int flags); 1429 extern int ext4_mb_add_groupinfo(struct super_block *sb, 1430 ext4_group_t i, struct ext4_group_desc *desc); 1431 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t); 1432 extern void ext4_mb_put_buddy_cache_lock(struct super_block *, 1433 ext4_group_t, int); 1434 /* inode.c */ 1435 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 1436 ext4_lblk_t, int, int *); 1437 struct buffer_head *ext4_bread(handle_t *, struct inode *, 1438 ext4_lblk_t, int, int *); 1439 int ext4_get_block(struct inode *inode, sector_t iblock, 1440 struct buffer_head *bh_result, int create); 1441 1442 extern struct inode *ext4_iget(struct super_block *, unsigned long); 1443 extern int ext4_write_inode(struct inode *, int); 1444 extern int ext4_setattr(struct dentry *, struct iattr *); 1445 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 1446 struct kstat *stat); 1447 extern void ext4_delete_inode(struct inode *); 1448 extern int ext4_sync_inode(handle_t *, struct inode *); 1449 extern void ext4_dirty_inode(struct inode *); 1450 extern int ext4_change_inode_journal_flag(struct inode *, int); 1451 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 1452 extern int ext4_can_truncate(struct inode *inode); 1453 extern void ext4_truncate(struct inode *); 1454 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 1455 extern void ext4_set_inode_flags(struct inode *); 1456 extern void ext4_get_inode_flags(struct ext4_inode_info *); 1457 extern int ext4_alloc_da_blocks(struct inode *inode); 1458 extern void ext4_set_aops(struct inode *inode); 1459 extern int ext4_writepage_trans_blocks(struct inode *); 1460 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks); 1461 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 1462 extern int ext4_block_truncate_page(handle_t *handle, 1463 struct address_space *mapping, loff_t from); 1464 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 1465 extern qsize_t *ext4_get_reserved_space(struct inode *inode); 1466 extern int flush_completed_IO(struct inode *inode); 1467 extern void ext4_da_update_reserve_space(struct inode *inode, 1468 int used, int quota_claim); 1469 /* ioctl.c */ 1470 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 1471 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 1472 1473 /* migrate.c */ 1474 extern int ext4_ext_migrate(struct inode *); 1475 1476 /* namei.c */ 1477 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize); 1478 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize); 1479 extern int ext4_orphan_add(handle_t *, struct inode *); 1480 extern int ext4_orphan_del(handle_t *, struct inode *); 1481 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 1482 __u32 start_minor_hash, __u32 *next_hash); 1483 1484 /* resize.c */ 1485 extern int ext4_group_add(struct super_block *sb, 1486 struct ext4_new_group_data *input); 1487 extern int ext4_group_extend(struct super_block *sb, 1488 struct ext4_super_block *es, 1489 ext4_fsblk_t n_blocks_count); 1490 1491 /* super.c */ 1492 extern void __ext4_error(struct super_block *, const char *, const char *, ...) 1493 __attribute__ ((format (printf, 3, 4))); 1494 #define ext4_error(sb, message...) __ext4_error(sb, __func__, ## message) 1495 extern void __ext4_std_error(struct super_block *, const char *, int); 1496 extern void ext4_abort(struct super_block *, const char *, const char *, ...) 1497 __attribute__ ((format (printf, 3, 4))); 1498 extern void __ext4_warning(struct super_block *, const char *, 1499 const char *, ...) 1500 __attribute__ ((format (printf, 3, 4))); 1501 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, ## message) 1502 extern void ext4_msg(struct super_block *, const char *, const char *, ...) 1503 __attribute__ ((format (printf, 3, 4))); 1504 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t, 1505 const char *, const char *, ...) 1506 __attribute__ ((format (printf, 4, 5))); 1507 extern void ext4_update_dynamic_rev(struct super_block *sb); 1508 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 1509 __u32 compat); 1510 extern int ext4_update_rocompat_feature(handle_t *handle, 1511 struct super_block *sb, __u32 rocompat); 1512 extern int ext4_update_incompat_feature(handle_t *handle, 1513 struct super_block *sb, __u32 incompat); 1514 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 1515 struct ext4_group_desc *bg); 1516 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 1517 struct ext4_group_desc *bg); 1518 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 1519 struct ext4_group_desc *bg); 1520 extern __u32 ext4_free_blks_count(struct super_block *sb, 1521 struct ext4_group_desc *bg); 1522 extern __u32 ext4_free_inodes_count(struct super_block *sb, 1523 struct ext4_group_desc *bg); 1524 extern __u32 ext4_used_dirs_count(struct super_block *sb, 1525 struct ext4_group_desc *bg); 1526 extern __u32 ext4_itable_unused_count(struct super_block *sb, 1527 struct ext4_group_desc *bg); 1528 extern void ext4_block_bitmap_set(struct super_block *sb, 1529 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1530 extern void ext4_inode_bitmap_set(struct super_block *sb, 1531 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1532 extern void ext4_inode_table_set(struct super_block *sb, 1533 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1534 extern void ext4_free_blks_set(struct super_block *sb, 1535 struct ext4_group_desc *bg, __u32 count); 1536 extern void ext4_free_inodes_set(struct super_block *sb, 1537 struct ext4_group_desc *bg, __u32 count); 1538 extern void ext4_used_dirs_set(struct super_block *sb, 1539 struct ext4_group_desc *bg, __u32 count); 1540 extern void ext4_itable_unused_set(struct super_block *sb, 1541 struct ext4_group_desc *bg, __u32 count); 1542 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group, 1543 struct ext4_group_desc *gdp); 1544 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group, 1545 struct ext4_group_desc *gdp); 1546 1547 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 1548 { 1549 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 1550 le32_to_cpu(es->s_blocks_count_lo); 1551 } 1552 1553 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 1554 { 1555 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 1556 le32_to_cpu(es->s_r_blocks_count_lo); 1557 } 1558 1559 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 1560 { 1561 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 1562 le32_to_cpu(es->s_free_blocks_count_lo); 1563 } 1564 1565 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 1566 ext4_fsblk_t blk) 1567 { 1568 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 1569 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 1570 } 1571 1572 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 1573 ext4_fsblk_t blk) 1574 { 1575 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 1576 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 1577 } 1578 1579 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 1580 ext4_fsblk_t blk) 1581 { 1582 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 1583 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 1584 } 1585 1586 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 1587 { 1588 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 1589 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 1590 le32_to_cpu(raw_inode->i_size_lo); 1591 else 1592 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 1593 } 1594 1595 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 1596 { 1597 raw_inode->i_size_lo = cpu_to_le32(i_size); 1598 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 1599 } 1600 1601 static inline 1602 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 1603 ext4_group_t group) 1604 { 1605 struct ext4_group_info ***grp_info; 1606 long indexv, indexh; 1607 grp_info = EXT4_SB(sb)->s_group_info; 1608 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 1609 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 1610 return grp_info[indexv][indexh]; 1611 } 1612 1613 /* 1614 * Reading s_groups_count requires using smp_rmb() afterwards. See 1615 * the locking protocol documented in the comments of ext4_group_add() 1616 * in resize.c 1617 */ 1618 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 1619 { 1620 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 1621 1622 smp_rmb(); 1623 return ngroups; 1624 } 1625 1626 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 1627 ext4_group_t block_group) 1628 { 1629 return block_group >> sbi->s_log_groups_per_flex; 1630 } 1631 1632 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 1633 { 1634 return 1 << sbi->s_log_groups_per_flex; 1635 } 1636 1637 #define ext4_std_error(sb, errno) \ 1638 do { \ 1639 if ((errno)) \ 1640 __ext4_std_error((sb), __func__, (errno)); \ 1641 } while (0) 1642 1643 #ifdef CONFIG_SMP 1644 /* Each CPU can accumulate percpu_counter_batch blocks in their local 1645 * counters. So we need to make sure we have free blocks more 1646 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 1647 */ 1648 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 1649 #else 1650 #define EXT4_FREEBLOCKS_WATERMARK 0 1651 #endif 1652 1653 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 1654 { 1655 /* 1656 * XXX: replace with spinlock if seen contended -bzzz 1657 */ 1658 down_write(&EXT4_I(inode)->i_data_sem); 1659 if (newsize > EXT4_I(inode)->i_disksize) 1660 EXT4_I(inode)->i_disksize = newsize; 1661 up_write(&EXT4_I(inode)->i_data_sem); 1662 return ; 1663 } 1664 1665 struct ext4_group_info { 1666 unsigned long bb_state; 1667 struct rb_root bb_free_root; 1668 ext4_grpblk_t bb_first_free; /* first free block */ 1669 ext4_grpblk_t bb_free; /* total free blocks */ 1670 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 1671 struct list_head bb_prealloc_list; 1672 #ifdef DOUBLE_CHECK 1673 void *bb_bitmap; 1674 #endif 1675 struct rw_semaphore alloc_sem; 1676 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 1677 * regions, index is order. 1678 * bb_counters[3] = 5 means 1679 * 5 free 8-block regions. */ 1680 }; 1681 1682 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 1683 1684 #define EXT4_MB_GRP_NEED_INIT(grp) \ 1685 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 1686 1687 #define EXT4_MAX_CONTENTION 8 1688 #define EXT4_CONTENTION_THRESHOLD 2 1689 1690 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 1691 ext4_group_t group) 1692 { 1693 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 1694 } 1695 1696 /* 1697 * Returns true if the filesystem is busy enough that attempts to 1698 * access the block group locks has run into contention. 1699 */ 1700 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 1701 { 1702 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 1703 } 1704 1705 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 1706 { 1707 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 1708 if (spin_trylock(lock)) 1709 /* 1710 * We're able to grab the lock right away, so drop the 1711 * lock contention counter. 1712 */ 1713 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 1714 else { 1715 /* 1716 * The lock is busy, so bump the contention counter, 1717 * and then wait on the spin lock. 1718 */ 1719 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 1720 EXT4_MAX_CONTENTION); 1721 spin_lock(lock); 1722 } 1723 } 1724 1725 static inline void ext4_unlock_group(struct super_block *sb, 1726 ext4_group_t group) 1727 { 1728 spin_unlock(ext4_group_lock_ptr(sb, group)); 1729 } 1730 1731 /* 1732 * Inodes and files operations 1733 */ 1734 1735 /* dir.c */ 1736 extern const struct file_operations ext4_dir_operations; 1737 1738 /* file.c */ 1739 extern const struct inode_operations ext4_file_inode_operations; 1740 extern const struct file_operations ext4_file_operations; 1741 1742 /* namei.c */ 1743 extern const struct inode_operations ext4_dir_inode_operations; 1744 extern const struct inode_operations ext4_special_inode_operations; 1745 extern struct dentry *ext4_get_parent(struct dentry *child); 1746 1747 /* symlink.c */ 1748 extern const struct inode_operations ext4_symlink_inode_operations; 1749 extern const struct inode_operations ext4_fast_symlink_inode_operations; 1750 1751 /* block_validity */ 1752 extern void ext4_release_system_zone(struct super_block *sb); 1753 extern int ext4_setup_system_zone(struct super_block *sb); 1754 extern int __init init_ext4_system_zone(void); 1755 extern void exit_ext4_system_zone(void); 1756 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 1757 ext4_fsblk_t start_blk, 1758 unsigned int count); 1759 1760 /* extents.c */ 1761 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 1762 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 1763 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 1764 int chunk); 1765 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, 1766 ext4_lblk_t iblock, unsigned int max_blocks, 1767 struct buffer_head *bh_result, int flags); 1768 extern void ext4_ext_truncate(struct inode *); 1769 extern void ext4_ext_init(struct super_block *); 1770 extern void ext4_ext_release(struct super_block *); 1771 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset, 1772 loff_t len); 1773 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 1774 ssize_t len); 1775 extern int ext4_get_blocks(handle_t *handle, struct inode *inode, 1776 sector_t block, unsigned int max_blocks, 1777 struct buffer_head *bh, int flags); 1778 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 1779 __u64 start, __u64 len); 1780 /* move_extent.c */ 1781 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 1782 __u64 start_orig, __u64 start_donor, 1783 __u64 len, __u64 *moved_len); 1784 1785 1786 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */ 1787 enum ext4_state_bits { 1788 BH_Uninit /* blocks are allocated but uninitialized on disk */ 1789 = BH_JBDPrivateStart, 1790 }; 1791 1792 BUFFER_FNS(Uninit, uninit) 1793 TAS_BUFFER_FNS(Uninit, uninit) 1794 1795 /* 1796 * Add new method to test wether block and inode bitmaps are properly 1797 * initialized. With uninit_bg reading the block from disk is not enough 1798 * to mark the bitmap uptodate. We need to also zero-out the bitmap 1799 */ 1800 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 1801 1802 static inline int bitmap_uptodate(struct buffer_head *bh) 1803 { 1804 return (buffer_uptodate(bh) && 1805 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 1806 } 1807 static inline void set_bitmap_uptodate(struct buffer_head *bh) 1808 { 1809 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 1810 } 1811 1812 #endif /* __KERNEL__ */ 1813 1814 #endif /* _EXT4_H */ 1815