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 DIO_AIO_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 int error; /* I/O error code */ 142 ext4_lblk_t offset; /* offset in the file */ 143 size_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_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 288 289 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */ 290 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */ 291 292 /* Flags that should be inherited by new inodes from their parent. */ 293 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 294 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\ 295 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 296 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 297 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 298 299 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 300 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 301 302 /* Flags that are appropriate for non-directories/regular files. */ 303 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 304 305 /* Mask out flags that are inappropriate for the given type of inode. */ 306 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 307 { 308 if (S_ISDIR(mode)) 309 return flags; 310 else if (S_ISREG(mode)) 311 return flags & EXT4_REG_FLMASK; 312 else 313 return flags & EXT4_OTHER_FLMASK; 314 } 315 316 /* 317 * Inode dynamic state flags 318 */ 319 #define EXT4_STATE_JDATA 0x00000001 /* journaled data exists */ 320 #define EXT4_STATE_NEW 0x00000002 /* inode is newly created */ 321 #define EXT4_STATE_XATTR 0x00000004 /* has in-inode xattrs */ 322 #define EXT4_STATE_NO_EXPAND 0x00000008 /* No space for expansion */ 323 #define EXT4_STATE_DA_ALLOC_CLOSE 0x00000010 /* Alloc DA blks on close */ 324 #define EXT4_STATE_EXT_MIGRATE 0x00000020 /* Inode is migrating */ 325 #define EXT4_STATE_DIO_UNWRITTEN 0x00000040 /* need convert on dio done*/ 326 327 /* Used to pass group descriptor data when online resize is done */ 328 struct ext4_new_group_input { 329 __u32 group; /* Group number for this data */ 330 __u64 block_bitmap; /* Absolute block number of block bitmap */ 331 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 332 __u64 inode_table; /* Absolute block number of inode table start */ 333 __u32 blocks_count; /* Total number of blocks in this group */ 334 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 335 __u16 unused; 336 }; 337 338 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 339 struct ext4_new_group_data { 340 __u32 group; 341 __u64 block_bitmap; 342 __u64 inode_bitmap; 343 __u64 inode_table; 344 __u32 blocks_count; 345 __u16 reserved_blocks; 346 __u16 unused; 347 __u32 free_blocks_count; 348 }; 349 350 /* 351 * Flags used by ext4_get_blocks() 352 */ 353 /* Allocate any needed blocks and/or convert an unitialized 354 extent to be an initialized ext4 */ 355 #define EXT4_GET_BLOCKS_CREATE 0x0001 356 /* Request the creation of an unitialized extent */ 357 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 358 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 359 EXT4_GET_BLOCKS_CREATE) 360 /* Caller is from the delayed allocation writeout path, 361 so set the magic i_delalloc_reserve_flag after taking the 362 inode allocation semaphore for */ 363 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 364 /* Call ext4_da_update_reserve_space() after successfully 365 allocating the blocks */ 366 #define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008 367 /* caller is from the direct IO path, request to creation of an 368 unitialized extents if not allocated, split the uninitialized 369 extent if blocks has been preallocated already*/ 370 #define EXT4_GET_BLOCKS_DIO 0x0010 371 #define EXT4_GET_BLOCKS_CONVERT 0x0020 372 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\ 373 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 374 /* Convert extent to initialized after direct IO complete */ 375 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 376 EXT4_GET_BLOCKS_DIO_CREATE_EXT) 377 378 /* 379 * Flags used by ext4_free_blocks 380 */ 381 #define EXT4_FREE_BLOCKS_METADATA 0x0001 382 #define EXT4_FREE_BLOCKS_FORGET 0x0002 383 384 /* 385 * ioctl commands 386 */ 387 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 388 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 389 #define EXT4_IOC_GETVERSION _IOR('f', 3, long) 390 #define EXT4_IOC_SETVERSION _IOW('f', 4, long) 391 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 392 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 393 #ifdef CONFIG_JBD2_DEBUG 394 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long) 395 #endif 396 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 397 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 398 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 399 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 400 #define EXT4_IOC_MIGRATE _IO('f', 9) 401 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 402 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 403 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 404 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 405 406 /* 407 * ioctl commands in 32 bit emulation 408 */ 409 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 410 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 411 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 412 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 413 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 414 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 415 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 416 #ifdef CONFIG_JBD2_DEBUG 417 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int) 418 #endif 419 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 420 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 421 422 423 /* 424 * Mount options 425 */ 426 struct ext4_mount_options { 427 unsigned long s_mount_opt; 428 uid_t s_resuid; 429 gid_t s_resgid; 430 unsigned long s_commit_interval; 431 u32 s_min_batch_time, s_max_batch_time; 432 #ifdef CONFIG_QUOTA 433 int s_jquota_fmt; 434 char *s_qf_names[MAXQUOTAS]; 435 #endif 436 }; 437 438 /* Max physical block we can addres w/o extents */ 439 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 440 441 /* 442 * Structure of an inode on the disk 443 */ 444 struct ext4_inode { 445 __le16 i_mode; /* File mode */ 446 __le16 i_uid; /* Low 16 bits of Owner Uid */ 447 __le32 i_size_lo; /* Size in bytes */ 448 __le32 i_atime; /* Access time */ 449 __le32 i_ctime; /* Inode Change time */ 450 __le32 i_mtime; /* Modification time */ 451 __le32 i_dtime; /* Deletion Time */ 452 __le16 i_gid; /* Low 16 bits of Group Id */ 453 __le16 i_links_count; /* Links count */ 454 __le32 i_blocks_lo; /* Blocks count */ 455 __le32 i_flags; /* File flags */ 456 union { 457 struct { 458 __le32 l_i_version; 459 } linux1; 460 struct { 461 __u32 h_i_translator; 462 } hurd1; 463 struct { 464 __u32 m_i_reserved1; 465 } masix1; 466 } osd1; /* OS dependent 1 */ 467 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 468 __le32 i_generation; /* File version (for NFS) */ 469 __le32 i_file_acl_lo; /* File ACL */ 470 __le32 i_size_high; 471 __le32 i_obso_faddr; /* Obsoleted fragment address */ 472 union { 473 struct { 474 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 475 __le16 l_i_file_acl_high; 476 __le16 l_i_uid_high; /* these 2 fields */ 477 __le16 l_i_gid_high; /* were reserved2[0] */ 478 __u32 l_i_reserved2; 479 } linux2; 480 struct { 481 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 482 __u16 h_i_mode_high; 483 __u16 h_i_uid_high; 484 __u16 h_i_gid_high; 485 __u32 h_i_author; 486 } hurd2; 487 struct { 488 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 489 __le16 m_i_file_acl_high; 490 __u32 m_i_reserved2[2]; 491 } masix2; 492 } osd2; /* OS dependent 2 */ 493 __le16 i_extra_isize; 494 __le16 i_pad1; 495 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 496 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 497 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 498 __le32 i_crtime; /* File Creation time */ 499 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 500 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 501 }; 502 503 struct move_extent { 504 __u32 reserved; /* should be zero */ 505 __u32 donor_fd; /* donor file descriptor */ 506 __u64 orig_start; /* logical start offset in block for orig */ 507 __u64 donor_start; /* logical start offset in block for donor */ 508 __u64 len; /* block length to be moved */ 509 __u64 moved_len; /* moved block length */ 510 }; 511 512 #define EXT4_EPOCH_BITS 2 513 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 514 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 515 516 /* 517 * Extended fields will fit into an inode if the filesystem was formatted 518 * with large inodes (-I 256 or larger) and there are not currently any EAs 519 * consuming all of the available space. For new inodes we always reserve 520 * enough space for the kernel's known extended fields, but for inodes 521 * created with an old kernel this might not have been the case. None of 522 * the extended inode fields is critical for correct filesystem operation. 523 * This macro checks if a certain field fits in the inode. Note that 524 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 525 */ 526 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 527 ((offsetof(typeof(*ext4_inode), field) + \ 528 sizeof((ext4_inode)->field)) \ 529 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 530 (einode)->i_extra_isize)) \ 531 532 static inline __le32 ext4_encode_extra_time(struct timespec *time) 533 { 534 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 535 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 536 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 537 } 538 539 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 540 { 541 if (sizeof(time->tv_sec) > 4) 542 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 543 << 32; 544 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 545 } 546 547 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 548 do { \ 549 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 550 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 551 (raw_inode)->xtime ## _extra = \ 552 ext4_encode_extra_time(&(inode)->xtime); \ 553 } while (0) 554 555 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 556 do { \ 557 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 558 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 559 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 560 (raw_inode)->xtime ## _extra = \ 561 ext4_encode_extra_time(&(einode)->xtime); \ 562 } while (0) 563 564 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 565 do { \ 566 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 567 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 568 ext4_decode_extra_time(&(inode)->xtime, \ 569 raw_inode->xtime ## _extra); \ 570 } while (0) 571 572 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 573 do { \ 574 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 575 (einode)->xtime.tv_sec = \ 576 (signed)le32_to_cpu((raw_inode)->xtime); \ 577 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 578 ext4_decode_extra_time(&(einode)->xtime, \ 579 raw_inode->xtime ## _extra); \ 580 } while (0) 581 582 #define i_disk_version osd1.linux1.l_i_version 583 584 #if defined(__KERNEL__) || defined(__linux__) 585 #define i_reserved1 osd1.linux1.l_i_reserved1 586 #define i_file_acl_high osd2.linux2.l_i_file_acl_high 587 #define i_blocks_high osd2.linux2.l_i_blocks_high 588 #define i_uid_low i_uid 589 #define i_gid_low i_gid 590 #define i_uid_high osd2.linux2.l_i_uid_high 591 #define i_gid_high osd2.linux2.l_i_gid_high 592 #define i_reserved2 osd2.linux2.l_i_reserved2 593 594 #elif defined(__GNU__) 595 596 #define i_translator osd1.hurd1.h_i_translator 597 #define i_uid_high osd2.hurd2.h_i_uid_high 598 #define i_gid_high osd2.hurd2.h_i_gid_high 599 #define i_author osd2.hurd2.h_i_author 600 601 #elif defined(__masix__) 602 603 #define i_reserved1 osd1.masix1.m_i_reserved1 604 #define i_file_acl_high osd2.masix2.m_i_file_acl_high 605 #define i_reserved2 osd2.masix2.m_i_reserved2 606 607 #endif /* defined(__KERNEL__) || defined(__linux__) */ 608 609 /* 610 * storage for cached extent 611 */ 612 struct ext4_ext_cache { 613 ext4_fsblk_t ec_start; 614 ext4_lblk_t ec_block; 615 __u32 ec_len; /* must be 32bit to return holes */ 616 __u32 ec_type; 617 }; 618 619 /* 620 * fourth extended file system inode data in memory 621 */ 622 struct ext4_inode_info { 623 __le32 i_data[15]; /* unconverted */ 624 __u32 i_flags; 625 ext4_fsblk_t i_file_acl; 626 __u32 i_dtime; 627 628 /* 629 * i_block_group is the number of the block group which contains 630 * this file's inode. Constant across the lifetime of the inode, 631 * it is ued for making block allocation decisions - we try to 632 * place a file's data blocks near its inode block, and new inodes 633 * near to their parent directory's inode. 634 */ 635 ext4_group_t i_block_group; 636 __u32 i_state; /* Dynamic state flags for ext4 */ 637 638 ext4_lblk_t i_dir_start_lookup; 639 #ifdef CONFIG_EXT4_FS_XATTR 640 /* 641 * Extended attributes can be read independently of the main file 642 * data. Taking i_mutex even when reading would cause contention 643 * between readers of EAs and writers of regular file data, so 644 * instead we synchronize on xattr_sem when reading or changing 645 * EAs. 646 */ 647 struct rw_semaphore xattr_sem; 648 #endif 649 650 struct list_head i_orphan; /* unlinked but open inodes */ 651 652 /* 653 * i_disksize keeps track of what the inode size is ON DISK, not 654 * in memory. During truncate, i_size is set to the new size by 655 * the VFS prior to calling ext4_truncate(), but the filesystem won't 656 * set i_disksize to 0 until the truncate is actually under way. 657 * 658 * The intent is that i_disksize always represents the blocks which 659 * are used by this file. This allows recovery to restart truncate 660 * on orphans if we crash during truncate. We actually write i_disksize 661 * into the on-disk inode when writing inodes out, instead of i_size. 662 * 663 * The only time when i_disksize and i_size may be different is when 664 * a truncate is in progress. The only things which change i_disksize 665 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 666 */ 667 loff_t i_disksize; 668 669 /* 670 * i_data_sem is for serialising ext4_truncate() against 671 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 672 * data tree are chopped off during truncate. We can't do that in 673 * ext4 because whenever we perform intermediate commits during 674 * truncate, the inode and all the metadata blocks *must* be in a 675 * consistent state which allows truncation of the orphans to restart 676 * during recovery. Hence we must fix the get_block-vs-truncate race 677 * by other means, so we have i_data_sem. 678 */ 679 struct rw_semaphore i_data_sem; 680 struct inode vfs_inode; 681 struct jbd2_inode jinode; 682 683 struct ext4_ext_cache i_cached_extent; 684 /* 685 * File creation time. Its function is same as that of 686 * struct timespec i_{a,c,m}time in the generic inode. 687 */ 688 struct timespec i_crtime; 689 690 /* mballoc */ 691 struct list_head i_prealloc_list; 692 spinlock_t i_prealloc_lock; 693 694 /* ialloc */ 695 ext4_group_t i_last_alloc_group; 696 697 /* allocation reservation info for delalloc */ 698 unsigned int i_reserved_data_blocks; 699 unsigned int i_reserved_meta_blocks; 700 unsigned int i_allocated_meta_blocks; 701 unsigned short i_delalloc_reserved_flag; 702 703 /* on-disk additional length */ 704 __u16 i_extra_isize; 705 706 spinlock_t i_block_reservation_lock; 707 708 /* completed async DIOs that might need unwritten extents handling */ 709 struct list_head i_aio_dio_complete_list; 710 /* current io_end structure for async DIO write*/ 711 ext4_io_end_t *cur_aio_dio; 712 713 /* 714 * Transactions that contain inode's metadata needed to complete 715 * fsync and fdatasync, respectively. 716 */ 717 tid_t i_sync_tid; 718 tid_t i_datasync_tid; 719 }; 720 721 /* 722 * File system states 723 */ 724 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 725 #define EXT4_ERROR_FS 0x0002 /* Errors detected */ 726 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 727 728 /* 729 * Misc. filesystem flags 730 */ 731 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 732 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 733 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 734 735 /* 736 * Mount flags 737 */ 738 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */ 739 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 740 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 741 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 742 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 743 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 744 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 745 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 746 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 747 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 748 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 749 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 750 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 751 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 752 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 753 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 754 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 755 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 756 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */ 757 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 758 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 759 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 760 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 761 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 762 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */ 763 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 764 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 765 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 766 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 767 768 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt 769 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt 770 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 771 EXT4_MOUNT_##opt) 772 773 #define ext4_set_bit ext2_set_bit 774 #define ext4_set_bit_atomic ext2_set_bit_atomic 775 #define ext4_clear_bit ext2_clear_bit 776 #define ext4_clear_bit_atomic ext2_clear_bit_atomic 777 #define ext4_test_bit ext2_test_bit 778 #define ext4_find_first_zero_bit ext2_find_first_zero_bit 779 #define ext4_find_next_zero_bit ext2_find_next_zero_bit 780 #define ext4_find_next_bit ext2_find_next_bit 781 782 /* 783 * Maximal mount counts between two filesystem checks 784 */ 785 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 786 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 787 788 /* 789 * Behaviour when detecting errors 790 */ 791 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 792 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 793 #define EXT4_ERRORS_PANIC 3 /* Panic */ 794 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 795 796 /* 797 * Structure of the super block 798 */ 799 struct ext4_super_block { 800 /*00*/ __le32 s_inodes_count; /* Inodes count */ 801 __le32 s_blocks_count_lo; /* Blocks count */ 802 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 803 __le32 s_free_blocks_count_lo; /* Free blocks count */ 804 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 805 __le32 s_first_data_block; /* First Data Block */ 806 __le32 s_log_block_size; /* Block size */ 807 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */ 808 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 809 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */ 810 __le32 s_inodes_per_group; /* # Inodes per group */ 811 __le32 s_mtime; /* Mount time */ 812 /*30*/ __le32 s_wtime; /* Write time */ 813 __le16 s_mnt_count; /* Mount count */ 814 __le16 s_max_mnt_count; /* Maximal mount count */ 815 __le16 s_magic; /* Magic signature */ 816 __le16 s_state; /* File system state */ 817 __le16 s_errors; /* Behaviour when detecting errors */ 818 __le16 s_minor_rev_level; /* minor revision level */ 819 /*40*/ __le32 s_lastcheck; /* time of last check */ 820 __le32 s_checkinterval; /* max. time between checks */ 821 __le32 s_creator_os; /* OS */ 822 __le32 s_rev_level; /* Revision level */ 823 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 824 __le16 s_def_resgid; /* Default gid for reserved blocks */ 825 /* 826 * These fields are for EXT4_DYNAMIC_REV superblocks only. 827 * 828 * Note: the difference between the compatible feature set and 829 * the incompatible feature set is that if there is a bit set 830 * in the incompatible feature set that the kernel doesn't 831 * know about, it should refuse to mount the filesystem. 832 * 833 * e2fsck's requirements are more strict; if it doesn't know 834 * about a feature in either the compatible or incompatible 835 * feature set, it must abort and not try to meddle with 836 * things it doesn't understand... 837 */ 838 __le32 s_first_ino; /* First non-reserved inode */ 839 __le16 s_inode_size; /* size of inode structure */ 840 __le16 s_block_group_nr; /* block group # of this superblock */ 841 __le32 s_feature_compat; /* compatible feature set */ 842 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 843 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 844 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 845 /*78*/ char s_volume_name[16]; /* volume name */ 846 /*88*/ char s_last_mounted[64]; /* directory where last mounted */ 847 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 848 /* 849 * Performance hints. Directory preallocation should only 850 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 851 */ 852 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 853 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 854 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 855 /* 856 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 857 */ 858 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 859 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 860 __le32 s_journal_dev; /* device number of journal file */ 861 __le32 s_last_orphan; /* start of list of inodes to delete */ 862 __le32 s_hash_seed[4]; /* HTREE hash seed */ 863 __u8 s_def_hash_version; /* Default hash version to use */ 864 __u8 s_reserved_char_pad; 865 __le16 s_desc_size; /* size of group descriptor */ 866 /*100*/ __le32 s_default_mount_opts; 867 __le32 s_first_meta_bg; /* First metablock block group */ 868 __le32 s_mkfs_time; /* When the filesystem was created */ 869 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 870 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 871 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 872 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 873 __le32 s_free_blocks_count_hi; /* Free blocks count */ 874 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 875 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 876 __le32 s_flags; /* Miscellaneous flags */ 877 __le16 s_raid_stride; /* RAID stride */ 878 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */ 879 __le64 s_mmp_block; /* Block for multi-mount protection */ 880 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 881 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 882 __u8 s_reserved_char_pad2; 883 __le16 s_reserved_pad; 884 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 885 __u32 s_reserved[160]; /* Padding to the end of the block */ 886 }; 887 888 #ifdef __KERNEL__ 889 890 /* 891 * run-time mount flags 892 */ 893 #define EXT4_MF_MNTDIR_SAMPLED 0x0001 894 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 895 896 /* 897 * fourth extended-fs super-block data in memory 898 */ 899 struct ext4_sb_info { 900 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 901 unsigned long s_inodes_per_block;/* Number of inodes per block */ 902 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 903 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 904 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 905 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 906 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 907 ext4_group_t s_groups_count; /* Number of groups in the fs */ 908 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 909 unsigned long s_overhead_last; /* Last calculated overhead */ 910 unsigned long s_blocks_last; /* Last seen block count */ 911 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 912 struct buffer_head * s_sbh; /* Buffer containing the super block */ 913 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 914 struct buffer_head **s_group_desc; 915 unsigned int s_mount_opt; 916 unsigned int s_mount_flags; 917 ext4_fsblk_t s_sb_block; 918 uid_t s_resuid; 919 gid_t s_resgid; 920 unsigned short s_mount_state; 921 unsigned short s_pad; 922 int s_addr_per_block_bits; 923 int s_desc_per_block_bits; 924 int s_inode_size; 925 int s_first_ino; 926 unsigned int s_inode_readahead_blks; 927 unsigned int s_inode_goal; 928 spinlock_t s_next_gen_lock; 929 u32 s_next_generation; 930 u32 s_hash_seed[4]; 931 int s_def_hash_version; 932 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 933 struct percpu_counter s_freeblocks_counter; 934 struct percpu_counter s_freeinodes_counter; 935 struct percpu_counter s_dirs_counter; 936 struct percpu_counter s_dirtyblocks_counter; 937 struct blockgroup_lock *s_blockgroup_lock; 938 struct proc_dir_entry *s_proc; 939 struct kobject s_kobj; 940 struct completion s_kobj_unregister; 941 942 /* Journaling */ 943 struct inode *s_journal_inode; 944 struct journal_s *s_journal; 945 struct list_head s_orphan; 946 struct mutex s_orphan_lock; 947 struct mutex s_resize_lock; 948 unsigned long s_commit_interval; 949 u32 s_max_batch_time; 950 u32 s_min_batch_time; 951 struct block_device *journal_bdev; 952 #ifdef CONFIG_JBD2_DEBUG 953 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */ 954 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */ 955 #endif 956 #ifdef CONFIG_QUOTA 957 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 958 int s_jquota_fmt; /* Format of quota to use */ 959 #endif 960 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 961 struct rb_root system_blks; 962 963 #ifdef EXTENTS_STATS 964 /* ext4 extents stats */ 965 unsigned long s_ext_min; 966 unsigned long s_ext_max; 967 unsigned long s_depth_max; 968 spinlock_t s_ext_stats_lock; 969 unsigned long s_ext_blocks; 970 unsigned long s_ext_extents; 971 #endif 972 973 /* for buddy allocator */ 974 struct ext4_group_info ***s_group_info; 975 struct inode *s_buddy_cache; 976 long s_blocks_reserved; 977 spinlock_t s_reserve_lock; 978 spinlock_t s_md_lock; 979 tid_t s_last_transaction; 980 unsigned short *s_mb_offsets; 981 unsigned int *s_mb_maxs; 982 983 /* tunables */ 984 unsigned long s_stripe; 985 unsigned int s_mb_stream_request; 986 unsigned int s_mb_max_to_scan; 987 unsigned int s_mb_min_to_scan; 988 unsigned int s_mb_stats; 989 unsigned int s_mb_order2_reqs; 990 unsigned int s_mb_group_prealloc; 991 unsigned int s_max_writeback_mb_bump; 992 /* where last allocation was done - for stream allocation */ 993 unsigned long s_mb_last_group; 994 unsigned long s_mb_last_start; 995 996 /* stats for buddy allocator */ 997 spinlock_t s_mb_pa_lock; 998 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 999 atomic_t s_bal_success; /* we found long enough chunks */ 1000 atomic_t s_bal_allocated; /* in blocks */ 1001 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1002 atomic_t s_bal_goals; /* goal hits */ 1003 atomic_t s_bal_breaks; /* too long searches */ 1004 atomic_t s_bal_2orders; /* 2^order hits */ 1005 spinlock_t s_bal_lock; 1006 unsigned long s_mb_buddies_generated; 1007 unsigned long long s_mb_generation_time; 1008 atomic_t s_mb_lost_chunks; 1009 atomic_t s_mb_preallocated; 1010 atomic_t s_mb_discarded; 1011 atomic_t s_lock_busy; 1012 1013 /* locality groups */ 1014 struct ext4_locality_group *s_locality_groups; 1015 1016 /* for write statistics */ 1017 unsigned long s_sectors_written_start; 1018 u64 s_kbytes_written; 1019 1020 unsigned int s_log_groups_per_flex; 1021 struct flex_groups *s_flex_groups; 1022 1023 /* workqueue for dio unwritten */ 1024 struct workqueue_struct *dio_unwritten_wq; 1025 }; 1026 1027 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1028 { 1029 return sb->s_fs_info; 1030 } 1031 static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1032 { 1033 return container_of(inode, struct ext4_inode_info, vfs_inode); 1034 } 1035 1036 static inline struct timespec ext4_current_time(struct inode *inode) 1037 { 1038 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1039 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1040 } 1041 1042 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1043 { 1044 return ino == EXT4_ROOT_INO || 1045 ino == EXT4_JOURNAL_INO || 1046 ino == EXT4_RESIZE_INO || 1047 (ino >= EXT4_FIRST_INO(sb) && 1048 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1049 } 1050 #else 1051 /* Assume that user mode programs are passing in an ext4fs superblock, not 1052 * a kernel struct super_block. This will allow us to call the feature-test 1053 * macros from user land. */ 1054 #define EXT4_SB(sb) (sb) 1055 #endif 1056 1057 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1058 1059 /* 1060 * Codes for operating systems 1061 */ 1062 #define EXT4_OS_LINUX 0 1063 #define EXT4_OS_HURD 1 1064 #define EXT4_OS_MASIX 2 1065 #define EXT4_OS_FREEBSD 3 1066 #define EXT4_OS_LITES 4 1067 1068 /* 1069 * Revision levels 1070 */ 1071 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1072 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1073 1074 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1075 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1076 1077 #define EXT4_GOOD_OLD_INODE_SIZE 128 1078 1079 /* 1080 * Feature set definitions 1081 */ 1082 1083 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1084 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1085 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1086 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1087 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1088 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1089 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1090 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1091 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1092 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1093 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1094 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1095 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1096 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1097 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1098 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1099 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1100 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1101 1102 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1103 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1104 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1105 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1106 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1107 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1108 1109 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1110 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1111 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1112 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1113 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1114 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1115 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1116 1117 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1118 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1119 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1120 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1121 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1122 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1123 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1124 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1125 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1126 1127 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1128 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1129 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1130 EXT4_FEATURE_INCOMPAT_META_BG| \ 1131 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1132 EXT4_FEATURE_INCOMPAT_64BIT| \ 1133 EXT4_FEATURE_INCOMPAT_FLEX_BG) 1134 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1135 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1136 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1137 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1138 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1139 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1140 EXT4_FEATURE_RO_COMPAT_HUGE_FILE) 1141 1142 /* 1143 * Default values for user and/or group using reserved blocks 1144 */ 1145 #define EXT4_DEF_RESUID 0 1146 #define EXT4_DEF_RESGID 0 1147 1148 #define EXT4_DEF_INODE_READAHEAD_BLKS 32 1149 1150 /* 1151 * Default mount options 1152 */ 1153 #define EXT4_DEFM_DEBUG 0x0001 1154 #define EXT4_DEFM_BSDGROUPS 0x0002 1155 #define EXT4_DEFM_XATTR_USER 0x0004 1156 #define EXT4_DEFM_ACL 0x0008 1157 #define EXT4_DEFM_UID16 0x0010 1158 #define EXT4_DEFM_JMODE 0x0060 1159 #define EXT4_DEFM_JMODE_DATA 0x0020 1160 #define EXT4_DEFM_JMODE_ORDERED 0x0040 1161 #define EXT4_DEFM_JMODE_WBACK 0x0060 1162 1163 /* 1164 * Default journal batch times 1165 */ 1166 #define EXT4_DEF_MIN_BATCH_TIME 0 1167 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1168 1169 /* 1170 * Minimum number of groups in a flexgroup before we separate out 1171 * directories into the first block group of a flexgroup 1172 */ 1173 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1174 1175 /* 1176 * Structure of a directory entry 1177 */ 1178 #define EXT4_NAME_LEN 255 1179 1180 struct ext4_dir_entry { 1181 __le32 inode; /* Inode number */ 1182 __le16 rec_len; /* Directory entry length */ 1183 __le16 name_len; /* Name length */ 1184 char name[EXT4_NAME_LEN]; /* File name */ 1185 }; 1186 1187 /* 1188 * The new version of the directory entry. Since EXT4 structures are 1189 * stored in intel byte order, and the name_len field could never be 1190 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1191 * file_type field. 1192 */ 1193 struct ext4_dir_entry_2 { 1194 __le32 inode; /* Inode number */ 1195 __le16 rec_len; /* Directory entry length */ 1196 __u8 name_len; /* Name length */ 1197 __u8 file_type; 1198 char name[EXT4_NAME_LEN]; /* File name */ 1199 }; 1200 1201 /* 1202 * Ext4 directory file types. Only the low 3 bits are used. The 1203 * other bits are reserved for now. 1204 */ 1205 #define EXT4_FT_UNKNOWN 0 1206 #define EXT4_FT_REG_FILE 1 1207 #define EXT4_FT_DIR 2 1208 #define EXT4_FT_CHRDEV 3 1209 #define EXT4_FT_BLKDEV 4 1210 #define EXT4_FT_FIFO 5 1211 #define EXT4_FT_SOCK 6 1212 #define EXT4_FT_SYMLINK 7 1213 1214 #define EXT4_FT_MAX 8 1215 1216 /* 1217 * EXT4_DIR_PAD defines the directory entries boundaries 1218 * 1219 * NOTE: It must be a multiple of 4 1220 */ 1221 #define EXT4_DIR_PAD 4 1222 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1223 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1224 ~EXT4_DIR_ROUND) 1225 #define EXT4_MAX_REC_LEN ((1<<16)-1) 1226 1227 /* 1228 * Hash Tree Directory indexing 1229 * (c) Daniel Phillips, 2001 1230 */ 1231 1232 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1233 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1234 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL)) 1235 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1236 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1237 1238 /* Legal values for the dx_root hash_version field: */ 1239 1240 #define DX_HASH_LEGACY 0 1241 #define DX_HASH_HALF_MD4 1 1242 #define DX_HASH_TEA 2 1243 #define DX_HASH_LEGACY_UNSIGNED 3 1244 #define DX_HASH_HALF_MD4_UNSIGNED 4 1245 #define DX_HASH_TEA_UNSIGNED 5 1246 1247 #ifdef __KERNEL__ 1248 1249 /* hash info structure used by the directory hash */ 1250 struct dx_hash_info 1251 { 1252 u32 hash; 1253 u32 minor_hash; 1254 int hash_version; 1255 u32 *seed; 1256 }; 1257 1258 #define EXT4_HTREE_EOF 0x7fffffff 1259 1260 /* 1261 * Control parameters used by ext4_htree_next_block 1262 */ 1263 #define HASH_NB_ALWAYS 1 1264 1265 1266 /* 1267 * Describe an inode's exact location on disk and in memory 1268 */ 1269 struct ext4_iloc 1270 { 1271 struct buffer_head *bh; 1272 unsigned long offset; 1273 ext4_group_t block_group; 1274 }; 1275 1276 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1277 { 1278 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1279 } 1280 1281 /* 1282 * This structure is stuffed into the struct file's private_data field 1283 * for directories. It is where we put information so that we can do 1284 * readdir operations in hash tree order. 1285 */ 1286 struct dir_private_info { 1287 struct rb_root root; 1288 struct rb_node *curr_node; 1289 struct fname *extra_fname; 1290 loff_t last_pos; 1291 __u32 curr_hash; 1292 __u32 curr_minor_hash; 1293 __u32 next_hash; 1294 }; 1295 1296 /* calculate the first block number of the group */ 1297 static inline ext4_fsblk_t 1298 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1299 { 1300 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1301 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1302 } 1303 1304 /* 1305 * Special error return code only used by dx_probe() and its callers. 1306 */ 1307 #define ERR_BAD_DX_DIR -75000 1308 1309 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr, 1310 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp); 1311 1312 extern struct proc_dir_entry *ext4_proc_root; 1313 1314 /* 1315 * Function prototypes 1316 */ 1317 1318 /* 1319 * Ok, these declarations are also in <linux/kernel.h> but none of the 1320 * ext4 source programs needs to include it so they are duplicated here. 1321 */ 1322 # define NORET_TYPE /**/ 1323 # define ATTRIB_NORET __attribute__((noreturn)) 1324 # define NORET_AND noreturn, 1325 1326 /* bitmap.c */ 1327 extern unsigned int ext4_count_free(struct buffer_head *, unsigned); 1328 1329 /* balloc.c */ 1330 extern unsigned int ext4_block_group(struct super_block *sb, 1331 ext4_fsblk_t blocknr); 1332 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1333 ext4_fsblk_t blocknr); 1334 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1335 extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1336 ext4_group_t group); 1337 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1338 ext4_fsblk_t goal, unsigned long *count, int *errp); 1339 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1340 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks); 1341 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb, 1342 ext4_fsblk_t block, unsigned long count); 1343 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *); 1344 extern void ext4_check_blocks_bitmap(struct super_block *); 1345 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1346 ext4_group_t block_group, 1347 struct buffer_head ** bh); 1348 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1349 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1350 ext4_group_t block_group); 1351 extern unsigned ext4_init_block_bitmap(struct super_block *sb, 1352 struct buffer_head *bh, 1353 ext4_group_t group, 1354 struct ext4_group_desc *desc); 1355 #define ext4_free_blocks_after_init(sb, group, desc) \ 1356 ext4_init_block_bitmap(sb, NULL, group, desc) 1357 1358 /* dir.c */ 1359 extern int ext4_check_dir_entry(const char *, struct inode *, 1360 struct ext4_dir_entry_2 *, 1361 struct buffer_head *, unsigned int); 1362 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 1363 __u32 minor_hash, 1364 struct ext4_dir_entry_2 *dirent); 1365 extern void ext4_htree_free_dir_info(struct dir_private_info *p); 1366 1367 /* fsync.c */ 1368 extern int ext4_sync_file(struct file *, struct dentry *, int); 1369 1370 /* hash.c */ 1371 extern int ext4fs_dirhash(const char *name, int len, struct 1372 dx_hash_info *hinfo); 1373 1374 /* ialloc.c */ 1375 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int, 1376 const struct qstr *qstr, __u32 goal); 1377 extern void ext4_free_inode(handle_t *, struct inode *); 1378 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 1379 extern unsigned long ext4_count_free_inodes(struct super_block *); 1380 extern unsigned long ext4_count_dirs(struct super_block *); 1381 extern void ext4_check_inodes_bitmap(struct super_block *); 1382 extern unsigned ext4_init_inode_bitmap(struct super_block *sb, 1383 struct buffer_head *bh, 1384 ext4_group_t group, 1385 struct ext4_group_desc *desc); 1386 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 1387 1388 /* mballoc.c */ 1389 extern long ext4_mb_stats; 1390 extern long ext4_mb_max_to_scan; 1391 extern int ext4_mb_init(struct super_block *, int); 1392 extern int ext4_mb_release(struct super_block *); 1393 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 1394 struct ext4_allocation_request *, int *); 1395 extern int ext4_mb_reserve_blocks(struct super_block *, int); 1396 extern void ext4_discard_preallocations(struct inode *); 1397 extern int __init init_ext4_mballoc(void); 1398 extern void exit_ext4_mballoc(void); 1399 extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 1400 struct buffer_head *bh, ext4_fsblk_t block, 1401 unsigned long count, int flags); 1402 extern int ext4_mb_add_groupinfo(struct super_block *sb, 1403 ext4_group_t i, struct ext4_group_desc *desc); 1404 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t); 1405 extern void ext4_mb_put_buddy_cache_lock(struct super_block *, 1406 ext4_group_t, int); 1407 /* inode.c */ 1408 struct buffer_head *ext4_getblk(handle_t *, struct inode *, 1409 ext4_lblk_t, int, int *); 1410 struct buffer_head *ext4_bread(handle_t *, struct inode *, 1411 ext4_lblk_t, int, int *); 1412 int ext4_get_block(struct inode *inode, sector_t iblock, 1413 struct buffer_head *bh_result, int create); 1414 1415 extern struct inode *ext4_iget(struct super_block *, unsigned long); 1416 extern int ext4_write_inode(struct inode *, int); 1417 extern int ext4_setattr(struct dentry *, struct iattr *); 1418 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 1419 struct kstat *stat); 1420 extern void ext4_delete_inode(struct inode *); 1421 extern int ext4_sync_inode(handle_t *, struct inode *); 1422 extern void ext4_dirty_inode(struct inode *); 1423 extern int ext4_change_inode_journal_flag(struct inode *, int); 1424 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 1425 extern int ext4_can_truncate(struct inode *inode); 1426 extern void ext4_truncate(struct inode *); 1427 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 1428 extern void ext4_set_inode_flags(struct inode *); 1429 extern void ext4_get_inode_flags(struct ext4_inode_info *); 1430 extern int ext4_alloc_da_blocks(struct inode *inode); 1431 extern void ext4_set_aops(struct inode *inode); 1432 extern int ext4_writepage_trans_blocks(struct inode *); 1433 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks); 1434 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 1435 extern int ext4_block_truncate_page(handle_t *handle, 1436 struct address_space *mapping, loff_t from); 1437 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 1438 extern qsize_t ext4_get_reserved_space(struct inode *inode); 1439 extern int flush_aio_dio_completed_IO(struct inode *inode); 1440 /* ioctl.c */ 1441 extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 1442 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 1443 1444 /* migrate.c */ 1445 extern int ext4_ext_migrate(struct inode *); 1446 1447 /* namei.c */ 1448 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize); 1449 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize); 1450 extern int ext4_orphan_add(handle_t *, struct inode *); 1451 extern int ext4_orphan_del(handle_t *, struct inode *); 1452 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 1453 __u32 start_minor_hash, __u32 *next_hash); 1454 1455 /* resize.c */ 1456 extern int ext4_group_add(struct super_block *sb, 1457 struct ext4_new_group_data *input); 1458 extern int ext4_group_extend(struct super_block *sb, 1459 struct ext4_super_block *es, 1460 ext4_fsblk_t n_blocks_count); 1461 1462 /* super.c */ 1463 extern void ext4_error(struct super_block *, const char *, const char *, ...) 1464 __attribute__ ((format (printf, 3, 4))); 1465 extern void __ext4_std_error(struct super_block *, const char *, int); 1466 extern void ext4_abort(struct super_block *, const char *, const char *, ...) 1467 __attribute__ ((format (printf, 3, 4))); 1468 extern void ext4_warning(struct super_block *, const char *, const char *, ...) 1469 __attribute__ ((format (printf, 3, 4))); 1470 extern void ext4_msg(struct super_block *, const char *, const char *, ...) 1471 __attribute__ ((format (printf, 3, 4))); 1472 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t, 1473 const char *, const char *, ...) 1474 __attribute__ ((format (printf, 4, 5))); 1475 extern void ext4_update_dynamic_rev(struct super_block *sb); 1476 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 1477 __u32 compat); 1478 extern int ext4_update_rocompat_feature(handle_t *handle, 1479 struct super_block *sb, __u32 rocompat); 1480 extern int ext4_update_incompat_feature(handle_t *handle, 1481 struct super_block *sb, __u32 incompat); 1482 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 1483 struct ext4_group_desc *bg); 1484 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 1485 struct ext4_group_desc *bg); 1486 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 1487 struct ext4_group_desc *bg); 1488 extern __u32 ext4_free_blks_count(struct super_block *sb, 1489 struct ext4_group_desc *bg); 1490 extern __u32 ext4_free_inodes_count(struct super_block *sb, 1491 struct ext4_group_desc *bg); 1492 extern __u32 ext4_used_dirs_count(struct super_block *sb, 1493 struct ext4_group_desc *bg); 1494 extern __u32 ext4_itable_unused_count(struct super_block *sb, 1495 struct ext4_group_desc *bg); 1496 extern void ext4_block_bitmap_set(struct super_block *sb, 1497 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1498 extern void ext4_inode_bitmap_set(struct super_block *sb, 1499 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1500 extern void ext4_inode_table_set(struct super_block *sb, 1501 struct ext4_group_desc *bg, ext4_fsblk_t blk); 1502 extern void ext4_free_blks_set(struct super_block *sb, 1503 struct ext4_group_desc *bg, __u32 count); 1504 extern void ext4_free_inodes_set(struct super_block *sb, 1505 struct ext4_group_desc *bg, __u32 count); 1506 extern void ext4_used_dirs_set(struct super_block *sb, 1507 struct ext4_group_desc *bg, __u32 count); 1508 extern void ext4_itable_unused_set(struct super_block *sb, 1509 struct ext4_group_desc *bg, __u32 count); 1510 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group, 1511 struct ext4_group_desc *gdp); 1512 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group, 1513 struct ext4_group_desc *gdp); 1514 1515 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 1516 { 1517 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 1518 le32_to_cpu(es->s_blocks_count_lo); 1519 } 1520 1521 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 1522 { 1523 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 1524 le32_to_cpu(es->s_r_blocks_count_lo); 1525 } 1526 1527 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 1528 { 1529 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 1530 le32_to_cpu(es->s_free_blocks_count_lo); 1531 } 1532 1533 static inline void ext4_blocks_count_set(struct ext4_super_block *es, 1534 ext4_fsblk_t blk) 1535 { 1536 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 1537 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 1538 } 1539 1540 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 1541 ext4_fsblk_t blk) 1542 { 1543 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 1544 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 1545 } 1546 1547 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 1548 ext4_fsblk_t blk) 1549 { 1550 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 1551 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 1552 } 1553 1554 static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 1555 { 1556 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 1557 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 1558 le32_to_cpu(raw_inode->i_size_lo); 1559 else 1560 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 1561 } 1562 1563 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 1564 { 1565 raw_inode->i_size_lo = cpu_to_le32(i_size); 1566 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 1567 } 1568 1569 static inline 1570 struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 1571 ext4_group_t group) 1572 { 1573 struct ext4_group_info ***grp_info; 1574 long indexv, indexh; 1575 grp_info = EXT4_SB(sb)->s_group_info; 1576 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 1577 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 1578 return grp_info[indexv][indexh]; 1579 } 1580 1581 /* 1582 * Reading s_groups_count requires using smp_rmb() afterwards. See 1583 * the locking protocol documented in the comments of ext4_group_add() 1584 * in resize.c 1585 */ 1586 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 1587 { 1588 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 1589 1590 smp_rmb(); 1591 return ngroups; 1592 } 1593 1594 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 1595 ext4_group_t block_group) 1596 { 1597 return block_group >> sbi->s_log_groups_per_flex; 1598 } 1599 1600 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 1601 { 1602 return 1 << sbi->s_log_groups_per_flex; 1603 } 1604 1605 #define ext4_std_error(sb, errno) \ 1606 do { \ 1607 if ((errno)) \ 1608 __ext4_std_error((sb), __func__, (errno)); \ 1609 } while (0) 1610 1611 #ifdef CONFIG_SMP 1612 /* Each CPU can accumulate percpu_counter_batch blocks in their local 1613 * counters. So we need to make sure we have free blocks more 1614 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 1615 */ 1616 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 1617 #else 1618 #define EXT4_FREEBLOCKS_WATERMARK 0 1619 #endif 1620 1621 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 1622 { 1623 /* 1624 * XXX: replace with spinlock if seen contended -bzzz 1625 */ 1626 down_write(&EXT4_I(inode)->i_data_sem); 1627 if (newsize > EXT4_I(inode)->i_disksize) 1628 EXT4_I(inode)->i_disksize = newsize; 1629 up_write(&EXT4_I(inode)->i_data_sem); 1630 return ; 1631 } 1632 1633 struct ext4_group_info { 1634 unsigned long bb_state; 1635 struct rb_root bb_free_root; 1636 ext4_grpblk_t bb_first_free; /* first free block */ 1637 ext4_grpblk_t bb_free; /* total free blocks */ 1638 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 1639 struct list_head bb_prealloc_list; 1640 #ifdef DOUBLE_CHECK 1641 void *bb_bitmap; 1642 #endif 1643 struct rw_semaphore alloc_sem; 1644 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 1645 * regions, index is order. 1646 * bb_counters[3] = 5 means 1647 * 5 free 8-block regions. */ 1648 }; 1649 1650 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0 1651 1652 #define EXT4_MB_GRP_NEED_INIT(grp) \ 1653 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 1654 1655 #define EXT4_MAX_CONTENTION 8 1656 #define EXT4_CONTENTION_THRESHOLD 2 1657 1658 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 1659 ext4_group_t group) 1660 { 1661 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 1662 } 1663 1664 /* 1665 * Returns true if the filesystem is busy enough that attempts to 1666 * access the block group locks has run into contention. 1667 */ 1668 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 1669 { 1670 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 1671 } 1672 1673 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 1674 { 1675 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 1676 if (spin_trylock(lock)) 1677 /* 1678 * We're able to grab the lock right away, so drop the 1679 * lock contention counter. 1680 */ 1681 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 1682 else { 1683 /* 1684 * The lock is busy, so bump the contention counter, 1685 * and then wait on the spin lock. 1686 */ 1687 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 1688 EXT4_MAX_CONTENTION); 1689 spin_lock(lock); 1690 } 1691 } 1692 1693 static inline void ext4_unlock_group(struct super_block *sb, 1694 ext4_group_t group) 1695 { 1696 spin_unlock(ext4_group_lock_ptr(sb, group)); 1697 } 1698 1699 /* 1700 * Inodes and files operations 1701 */ 1702 1703 /* dir.c */ 1704 extern const struct file_operations ext4_dir_operations; 1705 1706 /* file.c */ 1707 extern const struct inode_operations ext4_file_inode_operations; 1708 extern const struct file_operations ext4_file_operations; 1709 1710 /* namei.c */ 1711 extern const struct inode_operations ext4_dir_inode_operations; 1712 extern const struct inode_operations ext4_special_inode_operations; 1713 extern struct dentry *ext4_get_parent(struct dentry *child); 1714 1715 /* symlink.c */ 1716 extern const struct inode_operations ext4_symlink_inode_operations; 1717 extern const struct inode_operations ext4_fast_symlink_inode_operations; 1718 1719 /* block_validity */ 1720 extern void ext4_release_system_zone(struct super_block *sb); 1721 extern int ext4_setup_system_zone(struct super_block *sb); 1722 extern int __init init_ext4_system_zone(void); 1723 extern void exit_ext4_system_zone(void); 1724 extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 1725 ext4_fsblk_t start_blk, 1726 unsigned int count); 1727 1728 /* extents.c */ 1729 extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 1730 extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 1731 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, 1732 int chunk); 1733 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, 1734 ext4_lblk_t iblock, unsigned int max_blocks, 1735 struct buffer_head *bh_result, int flags); 1736 extern void ext4_ext_truncate(struct inode *); 1737 extern void ext4_ext_init(struct super_block *); 1738 extern void ext4_ext_release(struct super_block *); 1739 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset, 1740 loff_t len); 1741 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset, 1742 loff_t len); 1743 extern int ext4_get_blocks(handle_t *handle, struct inode *inode, 1744 sector_t block, unsigned int max_blocks, 1745 struct buffer_head *bh, int flags); 1746 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 1747 __u64 start, __u64 len); 1748 /* move_extent.c */ 1749 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 1750 __u64 start_orig, __u64 start_donor, 1751 __u64 len, __u64 *moved_len); 1752 1753 1754 /* 1755 * Add new method to test wether block and inode bitmaps are properly 1756 * initialized. With uninit_bg reading the block from disk is not enough 1757 * to mark the bitmap uptodate. We need to also zero-out the bitmap 1758 */ 1759 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart 1760 1761 static inline int bitmap_uptodate(struct buffer_head *bh) 1762 { 1763 return (buffer_uptodate(bh) && 1764 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 1765 } 1766 static inline void set_bitmap_uptodate(struct buffer_head *bh) 1767 { 1768 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 1769 } 1770 1771 #endif /* __KERNEL__ */ 1772 1773 #endif /* _EXT4_H */ 1774