Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
at v3.14-rc7 2852 lines 103 kB view raw
1/* 2 * ext4.h 3 * 4 * Copyright (C) 1992, 1993, 1994, 1995 5 * Remy Card (card@masi.ibp.fr) 6 * Laboratoire MASI - Institut Blaise Pascal 7 * Universite Pierre et Marie Curie (Paris VI) 8 * 9 * from 10 * 11 * linux/include/linux/minix_fs.h 12 * 13 * Copyright (C) 1991, 1992 Linus Torvalds 14 */ 15 16#ifndef _EXT4_H 17#define _EXT4_H 18 19#include <linux/types.h> 20#include <linux/blkdev.h> 21#include <linux/magic.h> 22#include <linux/jbd2.h> 23#include <linux/quota.h> 24#include <linux/rwsem.h> 25#include <linux/rbtree.h> 26#include <linux/seqlock.h> 27#include <linux/mutex.h> 28#include <linux/timer.h> 29#include <linux/wait.h> 30#include <linux/blockgroup_lock.h> 31#include <linux/percpu_counter.h> 32#include <linux/ratelimit.h> 33#include <crypto/hash.h> 34#ifdef __KERNEL__ 35#include <linux/compat.h> 36#endif 37 38/* 39 * The fourth extended filesystem constants/structures 40 */ 41 42/* 43 * Define EXT4FS_DEBUG to produce debug messages 44 */ 45#undef EXT4FS_DEBUG 46 47/* 48 * Debug code 49 */ 50#ifdef EXT4FS_DEBUG 51#define ext4_debug(f, a...) \ 52 do { \ 53 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 54 __FILE__, __LINE__, __func__); \ 55 printk(KERN_DEBUG f, ## a); \ 56 } while (0) 57#else 58#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 59#endif 60 61/* 62 * Turn on EXT_DEBUG to get lots of info about extents operations. 63 */ 64#define EXT_DEBUG__ 65#ifdef EXT_DEBUG 66#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__) 67#else 68#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 69#endif 70 71#define EXT4_ERROR_INODE(inode, fmt, a...) \ 72 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 73 74#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \ 75 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a) 76 77#define EXT4_ERROR_FILE(file, block, fmt, a...) \ 78 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 79 80/* data type for block offset of block group */ 81typedef int ext4_grpblk_t; 82 83/* data type for filesystem-wide blocks number */ 84typedef unsigned long long ext4_fsblk_t; 85 86/* data type for file logical block number */ 87typedef __u32 ext4_lblk_t; 88 89/* data type for block group number */ 90typedef unsigned int ext4_group_t; 91 92/* 93 * Flags used in mballoc's allocation_context flags field. 94 * 95 * Also used to show what's going on for debugging purposes when the 96 * flag field is exported via the traceport interface 97 */ 98 99/* prefer goal again. length */ 100#define EXT4_MB_HINT_MERGE 0x0001 101/* blocks already reserved */ 102#define EXT4_MB_HINT_RESERVED 0x0002 103/* metadata is being allocated */ 104#define EXT4_MB_HINT_METADATA 0x0004 105/* first blocks in the file */ 106#define EXT4_MB_HINT_FIRST 0x0008 107/* search for the best chunk */ 108#define EXT4_MB_HINT_BEST 0x0010 109/* data is being allocated */ 110#define EXT4_MB_HINT_DATA 0x0020 111/* don't preallocate (for tails) */ 112#define EXT4_MB_HINT_NOPREALLOC 0x0040 113/* allocate for locality group */ 114#define EXT4_MB_HINT_GROUP_ALLOC 0x0080 115/* allocate goal blocks or none */ 116#define EXT4_MB_HINT_GOAL_ONLY 0x0100 117/* goal is meaningful */ 118#define EXT4_MB_HINT_TRY_GOAL 0x0200 119/* blocks already pre-reserved by delayed allocation */ 120#define EXT4_MB_DELALLOC_RESERVED 0x0400 121/* We are doing stream allocation */ 122#define EXT4_MB_STREAM_ALLOC 0x0800 123/* Use reserved root blocks if needed */ 124#define EXT4_MB_USE_ROOT_BLOCKS 0x1000 125/* Use blocks from reserved pool */ 126#define EXT4_MB_USE_RESERVED 0x2000 127 128struct ext4_allocation_request { 129 /* target inode for block we're allocating */ 130 struct inode *inode; 131 /* how many blocks we want to allocate */ 132 unsigned int len; 133 /* logical block in target inode */ 134 ext4_lblk_t logical; 135 /* the closest logical allocated block to the left */ 136 ext4_lblk_t lleft; 137 /* the closest logical allocated block to the right */ 138 ext4_lblk_t lright; 139 /* phys. target (a hint) */ 140 ext4_fsblk_t goal; 141 /* phys. block for the closest logical allocated block to the left */ 142 ext4_fsblk_t pleft; 143 /* phys. block for the closest logical allocated block to the right */ 144 ext4_fsblk_t pright; 145 /* flags. see above EXT4_MB_HINT_* */ 146 unsigned int flags; 147}; 148 149/* 150 * Logical to physical block mapping, used by ext4_map_blocks() 151 * 152 * This structure is used to pass requests into ext4_map_blocks() as 153 * well as to store the information returned by ext4_map_blocks(). It 154 * takes less room on the stack than a struct buffer_head. 155 */ 156#define EXT4_MAP_NEW (1 << BH_New) 157#define EXT4_MAP_MAPPED (1 << BH_Mapped) 158#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) 159#define EXT4_MAP_BOUNDARY (1 << BH_Boundary) 160#define EXT4_MAP_UNINIT (1 << BH_Uninit) 161/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of 162 * ext4_map_blocks wants to know whether or not the underlying cluster has 163 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that 164 * the requested mapping was from previously mapped (or delayed allocated) 165 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster 166 * should never appear on buffer_head's state flags. 167 */ 168#define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) 169#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 170 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 171 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) 172 173struct ext4_map_blocks { 174 ext4_fsblk_t m_pblk; 175 ext4_lblk_t m_lblk; 176 unsigned int m_len; 177 unsigned int m_flags; 178}; 179 180/* 181 * Flags for ext4_io_end->flags 182 */ 183#define EXT4_IO_END_UNWRITTEN 0x0001 184 185/* 186 * For converting uninitialized extents on a work queue. 'handle' is used for 187 * buffered writeback. 188 */ 189typedef struct ext4_io_end { 190 struct list_head list; /* per-file finished IO list */ 191 handle_t *handle; /* handle reserved for extent 192 * conversion */ 193 struct inode *inode; /* file being written to */ 194 struct bio *bio; /* Linked list of completed 195 * bios covering the extent */ 196 unsigned int flag; /* unwritten or not */ 197 loff_t offset; /* offset in the file */ 198 ssize_t size; /* size of the extent */ 199 atomic_t count; /* reference counter */ 200} ext4_io_end_t; 201 202struct ext4_io_submit { 203 int io_op; 204 struct bio *io_bio; 205 ext4_io_end_t *io_end; 206 sector_t io_next_block; 207}; 208 209/* 210 * Special inodes numbers 211 */ 212#define EXT4_BAD_INO 1 /* Bad blocks inode */ 213#define EXT4_ROOT_INO 2 /* Root inode */ 214#define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 215#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 216#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 217#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 218#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 219#define EXT4_JOURNAL_INO 8 /* Journal inode */ 220 221/* First non-reserved inode for old ext4 filesystems */ 222#define EXT4_GOOD_OLD_FIRST_INO 11 223 224/* 225 * Maximal count of links to a file 226 */ 227#define EXT4_LINK_MAX 65000 228 229/* 230 * Macro-instructions used to manage several block sizes 231 */ 232#define EXT4_MIN_BLOCK_SIZE 1024 233#define EXT4_MAX_BLOCK_SIZE 65536 234#define EXT4_MIN_BLOCK_LOG_SIZE 10 235#define EXT4_MAX_BLOCK_LOG_SIZE 16 236#ifdef __KERNEL__ 237# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 238#else 239# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 240#endif 241#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 242#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 243 EXT4_SB(s)->s_cluster_bits) 244#ifdef __KERNEL__ 245# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 246# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 247#else 248# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 249#endif 250#ifdef __KERNEL__ 251#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 252#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 253#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 254#else 255#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 256 EXT4_GOOD_OLD_INODE_SIZE : \ 257 (s)->s_inode_size) 258#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 259 EXT4_GOOD_OLD_FIRST_INO : \ 260 (s)->s_first_ino) 261#endif 262#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 263 264/* Translate a block number to a cluster number */ 265#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 266/* Translate a cluster number to a block number */ 267#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 268/* Translate # of blks to # of clusters */ 269#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 270 (sbi)->s_cluster_bits) 271/* Mask out the low bits to get the starting block of the cluster */ 272#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 273 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 274#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 275 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 276/* Get the cluster offset */ 277#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 278 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 279#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 280 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 281 282/* 283 * Structure of a blocks group descriptor 284 */ 285struct ext4_group_desc 286{ 287 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 288 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 289 __le32 bg_inode_table_lo; /* Inodes table block */ 290 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 291 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 292 __le16 bg_used_dirs_count_lo; /* Directories count */ 293 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 294 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 295 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 296 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 297 __le16 bg_itable_unused_lo; /* Unused inodes count */ 298 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 299 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 300 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 301 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 302 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 303 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 304 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 305 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 306 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 307 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 308 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 309 __u32 bg_reserved; 310}; 311 312#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 313 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 314 sizeof(__le16)) 315#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 316 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 317 sizeof(__le16)) 318 319/* 320 * Structure of a flex block group info 321 */ 322 323struct flex_groups { 324 atomic64_t free_clusters; 325 atomic_t free_inodes; 326 atomic_t used_dirs; 327}; 328 329#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 330#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 331#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 332 333/* 334 * Macro-instructions used to manage group descriptors 335 */ 336#define EXT4_MIN_DESC_SIZE 32 337#define EXT4_MIN_DESC_SIZE_64BIT 64 338#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 339#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 340#ifdef __KERNEL__ 341# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 342# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 343# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 344# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 345# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 346#else 347# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 348# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 349# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 350#endif 351 352/* 353 * Constants relative to the data blocks 354 */ 355#define EXT4_NDIR_BLOCKS 12 356#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 357#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 358#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 359#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 360 361/* 362 * Inode flags 363 */ 364#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 365#define EXT4_UNRM_FL 0x00000002 /* Undelete */ 366#define EXT4_COMPR_FL 0x00000004 /* Compress file */ 367#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 368#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 369#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 370#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 371#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 372/* Reserved for compression usage... */ 373#define EXT4_DIRTY_FL 0x00000100 374#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 375#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 376#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */ 377/* End compression flags --- maybe not all used */ 378#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 379#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 380#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 381#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 382#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 383#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 384#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 385#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 386#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 387#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 388#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 389#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 390 391#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */ 392#define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */ 393 394/* Flags that should be inherited by new inodes from their parent. */ 395#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 396 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 397 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 398 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL) 399 400/* Flags that are appropriate for regular files (all but dir-specific ones). */ 401#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL)) 402 403/* Flags that are appropriate for non-directories/regular files. */ 404#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 405 406/* Mask out flags that are inappropriate for the given type of inode. */ 407static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 408{ 409 if (S_ISDIR(mode)) 410 return flags; 411 else if (S_ISREG(mode)) 412 return flags & EXT4_REG_FLMASK; 413 else 414 return flags & EXT4_OTHER_FLMASK; 415} 416 417/* 418 * Inode flags used for atomic set/get 419 */ 420enum { 421 EXT4_INODE_SECRM = 0, /* Secure deletion */ 422 EXT4_INODE_UNRM = 1, /* Undelete */ 423 EXT4_INODE_COMPR = 2, /* Compress file */ 424 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 425 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 426 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 427 EXT4_INODE_NODUMP = 6, /* do not dump file */ 428 EXT4_INODE_NOATIME = 7, /* do not update atime */ 429/* Reserved for compression usage... */ 430 EXT4_INODE_DIRTY = 8, 431 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 432 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 433 EXT4_INODE_ECOMPR = 11, /* Compression error */ 434/* End compression flags --- maybe not all used */ 435 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 436 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 437 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 438 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 439 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 440 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 441 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 442 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 443 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 444 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */ 445 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 446 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 447}; 448 449/* 450 * Since it's pretty easy to mix up bit numbers and hex values, we use a 451 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 452 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 453 * any extra space in the compiled kernel image, otherwise, the build will fail. 454 * It's important that these values are the same, since we are using 455 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 456 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 457 * values found in ext2, ext3 and ext4 filesystems, and of course the values 458 * defined in e2fsprogs. 459 * 460 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 461 */ 462#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG)) 463#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 464 465static inline void ext4_check_flag_values(void) 466{ 467 CHECK_FLAG_VALUE(SECRM); 468 CHECK_FLAG_VALUE(UNRM); 469 CHECK_FLAG_VALUE(COMPR); 470 CHECK_FLAG_VALUE(SYNC); 471 CHECK_FLAG_VALUE(IMMUTABLE); 472 CHECK_FLAG_VALUE(APPEND); 473 CHECK_FLAG_VALUE(NODUMP); 474 CHECK_FLAG_VALUE(NOATIME); 475 CHECK_FLAG_VALUE(DIRTY); 476 CHECK_FLAG_VALUE(COMPRBLK); 477 CHECK_FLAG_VALUE(NOCOMPR); 478 CHECK_FLAG_VALUE(ECOMPR); 479 CHECK_FLAG_VALUE(INDEX); 480 CHECK_FLAG_VALUE(IMAGIC); 481 CHECK_FLAG_VALUE(JOURNAL_DATA); 482 CHECK_FLAG_VALUE(NOTAIL); 483 CHECK_FLAG_VALUE(DIRSYNC); 484 CHECK_FLAG_VALUE(TOPDIR); 485 CHECK_FLAG_VALUE(HUGE_FILE); 486 CHECK_FLAG_VALUE(EXTENTS); 487 CHECK_FLAG_VALUE(EA_INODE); 488 CHECK_FLAG_VALUE(EOFBLOCKS); 489 CHECK_FLAG_VALUE(INLINE_DATA); 490 CHECK_FLAG_VALUE(RESERVED); 491} 492 493/* Used to pass group descriptor data when online resize is done */ 494struct ext4_new_group_input { 495 __u32 group; /* Group number for this data */ 496 __u64 block_bitmap; /* Absolute block number of block bitmap */ 497 __u64 inode_bitmap; /* Absolute block number of inode bitmap */ 498 __u64 inode_table; /* Absolute block number of inode table start */ 499 __u32 blocks_count; /* Total number of blocks in this group */ 500 __u16 reserved_blocks; /* Number of reserved blocks in this group */ 501 __u16 unused; 502}; 503 504#if defined(__KERNEL__) && defined(CONFIG_COMPAT) 505struct compat_ext4_new_group_input { 506 u32 group; 507 compat_u64 block_bitmap; 508 compat_u64 inode_bitmap; 509 compat_u64 inode_table; 510 u32 blocks_count; 511 u16 reserved_blocks; 512 u16 unused; 513}; 514#endif 515 516/* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 517struct ext4_new_group_data { 518 __u32 group; 519 __u64 block_bitmap; 520 __u64 inode_bitmap; 521 __u64 inode_table; 522 __u32 blocks_count; 523 __u16 reserved_blocks; 524 __u16 unused; 525 __u32 free_blocks_count; 526}; 527 528/* Indexes used to index group tables in ext4_new_group_data */ 529enum { 530 BLOCK_BITMAP = 0, /* block bitmap */ 531 INODE_BITMAP, /* inode bitmap */ 532 INODE_TABLE, /* inode tables */ 533 GROUP_TABLE_COUNT, 534}; 535 536/* 537 * Flags used by ext4_map_blocks() 538 */ 539 /* Allocate any needed blocks and/or convert an unitialized 540 extent to be an initialized ext4 */ 541#define EXT4_GET_BLOCKS_CREATE 0x0001 542 /* Request the creation of an unitialized extent */ 543#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002 544#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\ 545 EXT4_GET_BLOCKS_CREATE) 546 /* Caller is from the delayed allocation writeout path 547 * finally doing the actual allocation of delayed blocks */ 548#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 549 /* caller is from the direct IO path, request to creation of an 550 unitialized extents if not allocated, split the uninitialized 551 extent if blocks has been preallocated already*/ 552#define EXT4_GET_BLOCKS_PRE_IO 0x0008 553#define EXT4_GET_BLOCKS_CONVERT 0x0010 554#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 555 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 556 /* Convert extent to initialized after IO complete */ 557#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\ 558 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) 559 /* Eventual metadata allocation (due to growing extent tree) 560 * should not fail, so try to use reserved blocks for that.*/ 561#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 562 /* Don't normalize allocation size (used for fallocate) */ 563#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 564 /* Request will not result in inode size update (user for fallocate) */ 565#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 566 /* Do not take i_data_sem locking in ext4_map_blocks */ 567#define EXT4_GET_BLOCKS_NO_LOCK 0x0100 568 /* Do not put hole in extent cache */ 569#define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200 570 571/* 572 * The bit position of these flags must not overlap with any of the 573 * EXT4_GET_BLOCKS_*. They are used by ext4_ext_find_extent(), 574 * read_extent_tree_block(), ext4_split_extent_at(), 575 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 576 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 577 * caching the extents when reading from the extent tree while a 578 * truncate or punch hole operation is in progress. 579 */ 580#define EXT4_EX_NOCACHE 0x0400 581#define EXT4_EX_FORCE_CACHE 0x0800 582 583/* 584 * Flags used by ext4_free_blocks 585 */ 586#define EXT4_FREE_BLOCKS_METADATA 0x0001 587#define EXT4_FREE_BLOCKS_FORGET 0x0002 588#define EXT4_FREE_BLOCKS_VALIDATED 0x0004 589#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 590#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 591#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 592#define EXT4_FREE_BLOCKS_RESERVE 0x0040 593 594/* 595 * ioctl commands 596 */ 597#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS 598#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS 599#define EXT4_IOC_GETVERSION _IOR('f', 3, long) 600#define EXT4_IOC_SETVERSION _IOW('f', 4, long) 601#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION 602#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION 603#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long) 604#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long) 605#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long) 606#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input) 607#define EXT4_IOC_MIGRATE _IO('f', 9) 608 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */ 609 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */ 610#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12) 611#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent) 612#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64) 613#define EXT4_IOC_SWAP_BOOT _IO('f', 17) 614#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18) 615 616#if defined(__KERNEL__) && defined(CONFIG_COMPAT) 617/* 618 * ioctl commands in 32 bit emulation 619 */ 620#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS 621#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS 622#define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 623#define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 624#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 625#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 626#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 627#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 628#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 629#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 630#endif 631 632/* Max physical block we can address w/o extents */ 633#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 634 635/* 636 * Structure of an inode on the disk 637 */ 638struct ext4_inode { 639 __le16 i_mode; /* File mode */ 640 __le16 i_uid; /* Low 16 bits of Owner Uid */ 641 __le32 i_size_lo; /* Size in bytes */ 642 __le32 i_atime; /* Access time */ 643 __le32 i_ctime; /* Inode Change time */ 644 __le32 i_mtime; /* Modification time */ 645 __le32 i_dtime; /* Deletion Time */ 646 __le16 i_gid; /* Low 16 bits of Group Id */ 647 __le16 i_links_count; /* Links count */ 648 __le32 i_blocks_lo; /* Blocks count */ 649 __le32 i_flags; /* File flags */ 650 union { 651 struct { 652 __le32 l_i_version; 653 } linux1; 654 struct { 655 __u32 h_i_translator; 656 } hurd1; 657 struct { 658 __u32 m_i_reserved1; 659 } masix1; 660 } osd1; /* OS dependent 1 */ 661 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 662 __le32 i_generation; /* File version (for NFS) */ 663 __le32 i_file_acl_lo; /* File ACL */ 664 __le32 i_size_high; 665 __le32 i_obso_faddr; /* Obsoleted fragment address */ 666 union { 667 struct { 668 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 669 __le16 l_i_file_acl_high; 670 __le16 l_i_uid_high; /* these 2 fields */ 671 __le16 l_i_gid_high; /* were reserved2[0] */ 672 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 673 __le16 l_i_reserved; 674 } linux2; 675 struct { 676 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 677 __u16 h_i_mode_high; 678 __u16 h_i_uid_high; 679 __u16 h_i_gid_high; 680 __u32 h_i_author; 681 } hurd2; 682 struct { 683 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 684 __le16 m_i_file_acl_high; 685 __u32 m_i_reserved2[2]; 686 } masix2; 687 } osd2; /* OS dependent 2 */ 688 __le16 i_extra_isize; 689 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 690 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 691 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 692 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 693 __le32 i_crtime; /* File Creation time */ 694 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 695 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 696}; 697 698struct move_extent { 699 __u32 reserved; /* should be zero */ 700 __u32 donor_fd; /* donor file descriptor */ 701 __u64 orig_start; /* logical start offset in block for orig */ 702 __u64 donor_start; /* logical start offset in block for donor */ 703 __u64 len; /* block length to be moved */ 704 __u64 moved_len; /* moved block length */ 705}; 706 707#define EXT4_EPOCH_BITS 2 708#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 709#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 710 711/* 712 * Extended fields will fit into an inode if the filesystem was formatted 713 * with large inodes (-I 256 or larger) and there are not currently any EAs 714 * consuming all of the available space. For new inodes we always reserve 715 * enough space for the kernel's known extended fields, but for inodes 716 * created with an old kernel this might not have been the case. None of 717 * the extended inode fields is critical for correct filesystem operation. 718 * This macro checks if a certain field fits in the inode. Note that 719 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 720 */ 721#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 722 ((offsetof(typeof(*ext4_inode), field) + \ 723 sizeof((ext4_inode)->field)) \ 724 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 725 (einode)->i_extra_isize)) \ 726 727static inline __le32 ext4_encode_extra_time(struct timespec *time) 728{ 729 return cpu_to_le32((sizeof(time->tv_sec) > 4 ? 730 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) | 731 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK)); 732} 733 734static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra) 735{ 736 if (sizeof(time->tv_sec) > 4) 737 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) 738 << 32; 739 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 740} 741 742#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \ 743do { \ 744 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \ 745 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 746 (raw_inode)->xtime ## _extra = \ 747 ext4_encode_extra_time(&(inode)->xtime); \ 748} while (0) 749 750#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 751do { \ 752 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 753 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \ 754 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 755 (raw_inode)->xtime ## _extra = \ 756 ext4_encode_extra_time(&(einode)->xtime); \ 757} while (0) 758 759#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \ 760do { \ 761 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \ 762 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \ 763 ext4_decode_extra_time(&(inode)->xtime, \ 764 raw_inode->xtime ## _extra); \ 765 else \ 766 (inode)->xtime.tv_nsec = 0; \ 767} while (0) 768 769#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 770do { \ 771 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 772 (einode)->xtime.tv_sec = \ 773 (signed)le32_to_cpu((raw_inode)->xtime); \ 774 else \ 775 (einode)->xtime.tv_sec = 0; \ 776 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \ 777 ext4_decode_extra_time(&(einode)->xtime, \ 778 raw_inode->xtime ## _extra); \ 779 else \ 780 (einode)->xtime.tv_nsec = 0; \ 781} while (0) 782 783#define i_disk_version osd1.linux1.l_i_version 784 785#if defined(__KERNEL__) || defined(__linux__) 786#define i_reserved1 osd1.linux1.l_i_reserved1 787#define i_file_acl_high osd2.linux2.l_i_file_acl_high 788#define i_blocks_high osd2.linux2.l_i_blocks_high 789#define i_uid_low i_uid 790#define i_gid_low i_gid 791#define i_uid_high osd2.linux2.l_i_uid_high 792#define i_gid_high osd2.linux2.l_i_gid_high 793#define i_checksum_lo osd2.linux2.l_i_checksum_lo 794 795#elif defined(__GNU__) 796 797#define i_translator osd1.hurd1.h_i_translator 798#define i_uid_high osd2.hurd2.h_i_uid_high 799#define i_gid_high osd2.hurd2.h_i_gid_high 800#define i_author osd2.hurd2.h_i_author 801 802#elif defined(__masix__) 803 804#define i_reserved1 osd1.masix1.m_i_reserved1 805#define i_file_acl_high osd2.masix2.m_i_file_acl_high 806#define i_reserved2 osd2.masix2.m_i_reserved2 807 808#endif /* defined(__KERNEL__) || defined(__linux__) */ 809 810#include "extents_status.h" 811 812/* 813 * fourth extended file system inode data in memory 814 */ 815struct ext4_inode_info { 816 __le32 i_data[15]; /* unconverted */ 817 __u32 i_dtime; 818 ext4_fsblk_t i_file_acl; 819 820 /* 821 * i_block_group is the number of the block group which contains 822 * this file's inode. Constant across the lifetime of the inode, 823 * it is ued for making block allocation decisions - we try to 824 * place a file's data blocks near its inode block, and new inodes 825 * near to their parent directory's inode. 826 */ 827 ext4_group_t i_block_group; 828 ext4_lblk_t i_dir_start_lookup; 829#if (BITS_PER_LONG < 64) 830 unsigned long i_state_flags; /* Dynamic state flags */ 831#endif 832 unsigned long i_flags; 833 834 /* 835 * Extended attributes can be read independently of the main file 836 * data. Taking i_mutex even when reading would cause contention 837 * between readers of EAs and writers of regular file data, so 838 * instead we synchronize on xattr_sem when reading or changing 839 * EAs. 840 */ 841 struct rw_semaphore xattr_sem; 842 843 struct list_head i_orphan; /* unlinked but open inodes */ 844 845 /* 846 * i_disksize keeps track of what the inode size is ON DISK, not 847 * in memory. During truncate, i_size is set to the new size by 848 * the VFS prior to calling ext4_truncate(), but the filesystem won't 849 * set i_disksize to 0 until the truncate is actually under way. 850 * 851 * The intent is that i_disksize always represents the blocks which 852 * are used by this file. This allows recovery to restart truncate 853 * on orphans if we crash during truncate. We actually write i_disksize 854 * into the on-disk inode when writing inodes out, instead of i_size. 855 * 856 * The only time when i_disksize and i_size may be different is when 857 * a truncate is in progress. The only things which change i_disksize 858 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 859 */ 860 loff_t i_disksize; 861 862 /* 863 * i_data_sem is for serialising ext4_truncate() against 864 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 865 * data tree are chopped off during truncate. We can't do that in 866 * ext4 because whenever we perform intermediate commits during 867 * truncate, the inode and all the metadata blocks *must* be in a 868 * consistent state which allows truncation of the orphans to restart 869 * during recovery. Hence we must fix the get_block-vs-truncate race 870 * by other means, so we have i_data_sem. 871 */ 872 struct rw_semaphore i_data_sem; 873 struct inode vfs_inode; 874 struct jbd2_inode *jinode; 875 876 /* 877 * File creation time. Its function is same as that of 878 * struct timespec i_{a,c,m}time in the generic inode. 879 */ 880 struct timespec i_crtime; 881 882 /* mballoc */ 883 struct list_head i_prealloc_list; 884 spinlock_t i_prealloc_lock; 885 886 /* extents status tree */ 887 struct ext4_es_tree i_es_tree; 888 rwlock_t i_es_lock; 889 struct list_head i_es_lru; 890 unsigned int i_es_lru_nr; /* protected by i_es_lock */ 891 unsigned long i_touch_when; /* jiffies of last accessing */ 892 893 /* ialloc */ 894 ext4_group_t i_last_alloc_group; 895 896 /* allocation reservation info for delalloc */ 897 /* In case of bigalloc, these refer to clusters rather than blocks */ 898 unsigned int i_reserved_data_blocks; 899 unsigned int i_reserved_meta_blocks; 900 unsigned int i_allocated_meta_blocks; 901 ext4_lblk_t i_da_metadata_calc_last_lblock; 902 int i_da_metadata_calc_len; 903 904 /* on-disk additional length */ 905 __u16 i_extra_isize; 906 907 /* Indicate the inline data space. */ 908 u16 i_inline_off; 909 u16 i_inline_size; 910 911#ifdef CONFIG_QUOTA 912 /* quota space reservation, managed internally by quota code */ 913 qsize_t i_reserved_quota; 914#endif 915 916 /* Lock protecting lists below */ 917 spinlock_t i_completed_io_lock; 918 /* 919 * Completed IOs that need unwritten extents handling and have 920 * transaction reserved 921 */ 922 struct list_head i_rsv_conversion_list; 923 /* 924 * Completed IOs that need unwritten extents handling and don't have 925 * transaction reserved 926 */ 927 atomic_t i_ioend_count; /* Number of outstanding io_end structs */ 928 atomic_t i_unwritten; /* Nr. of inflight conversions pending */ 929 struct work_struct i_rsv_conversion_work; 930 931 spinlock_t i_block_reservation_lock; 932 933 /* 934 * Transactions that contain inode's metadata needed to complete 935 * fsync and fdatasync, respectively. 936 */ 937 tid_t i_sync_tid; 938 tid_t i_datasync_tid; 939 940 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 941 __u32 i_csum_seed; 942}; 943 944/* 945 * File system states 946 */ 947#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 948#define EXT4_ERROR_FS 0x0002 /* Errors detected */ 949#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 950 951/* 952 * Misc. filesystem flags 953 */ 954#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 955#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 956#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 957 958/* 959 * Mount flags set via mount options or defaults 960 */ 961#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 962#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 963#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 964#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 965#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 966#define EXT4_MOUNT_ERRORS_MASK 0x00070 967#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 968#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 969#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 970#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 971#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 972#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 973#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 974#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 975#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 976#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 977#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 978#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 979#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */ 980#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */ 981#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */ 982#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 983#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 984#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 985#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 986#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 987#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 988#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 989#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 990 991/* 992 * Mount flags set either automatically (could not be set by mount option) 993 * based on per file system feature or property or in special cases such as 994 * distinguishing between explicit mount option definition and default. 995 */ 996#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 997 specified delalloc */ 998#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 999 size of blocksize * 8 1000 blocks */ 1001 1002#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1003 ~EXT4_MOUNT_##opt 1004#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1005 EXT4_MOUNT_##opt 1006#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1007 EXT4_MOUNT_##opt) 1008 1009#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1010 ~EXT4_MOUNT2_##opt 1011#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1012 EXT4_MOUNT2_##opt 1013#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1014 EXT4_MOUNT2_##opt) 1015 1016#define ext4_test_and_set_bit __test_and_set_bit_le 1017#define ext4_set_bit __set_bit_le 1018#define ext4_set_bit_atomic ext2_set_bit_atomic 1019#define ext4_test_and_clear_bit __test_and_clear_bit_le 1020#define ext4_clear_bit __clear_bit_le 1021#define ext4_clear_bit_atomic ext2_clear_bit_atomic 1022#define ext4_test_bit test_bit_le 1023#define ext4_find_next_zero_bit find_next_zero_bit_le 1024#define ext4_find_next_bit find_next_bit_le 1025 1026extern void ext4_set_bits(void *bm, int cur, int len); 1027 1028/* 1029 * Maximal mount counts between two filesystem checks 1030 */ 1031#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1032#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1033 1034/* 1035 * Behaviour when detecting errors 1036 */ 1037#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1038#define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1039#define EXT4_ERRORS_PANIC 3 /* Panic */ 1040#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1041 1042/* Metadata checksum algorithm codes */ 1043#define EXT4_CRC32C_CHKSUM 1 1044 1045/* 1046 * Structure of the super block 1047 */ 1048struct ext4_super_block { 1049/*00*/ __le32 s_inodes_count; /* Inodes count */ 1050 __le32 s_blocks_count_lo; /* Blocks count */ 1051 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1052 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1053/*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1054 __le32 s_first_data_block; /* First Data Block */ 1055 __le32 s_log_block_size; /* Block size */ 1056 __le32 s_log_cluster_size; /* Allocation cluster size */ 1057/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1058 __le32 s_clusters_per_group; /* # Clusters per group */ 1059 __le32 s_inodes_per_group; /* # Inodes per group */ 1060 __le32 s_mtime; /* Mount time */ 1061/*30*/ __le32 s_wtime; /* Write time */ 1062 __le16 s_mnt_count; /* Mount count */ 1063 __le16 s_max_mnt_count; /* Maximal mount count */ 1064 __le16 s_magic; /* Magic signature */ 1065 __le16 s_state; /* File system state */ 1066 __le16 s_errors; /* Behaviour when detecting errors */ 1067 __le16 s_minor_rev_level; /* minor revision level */ 1068/*40*/ __le32 s_lastcheck; /* time of last check */ 1069 __le32 s_checkinterval; /* max. time between checks */ 1070 __le32 s_creator_os; /* OS */ 1071 __le32 s_rev_level; /* Revision level */ 1072/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1073 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1074 /* 1075 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1076 * 1077 * Note: the difference between the compatible feature set and 1078 * the incompatible feature set is that if there is a bit set 1079 * in the incompatible feature set that the kernel doesn't 1080 * know about, it should refuse to mount the filesystem. 1081 * 1082 * e2fsck's requirements are more strict; if it doesn't know 1083 * about a feature in either the compatible or incompatible 1084 * feature set, it must abort and not try to meddle with 1085 * things it doesn't understand... 1086 */ 1087 __le32 s_first_ino; /* First non-reserved inode */ 1088 __le16 s_inode_size; /* size of inode structure */ 1089 __le16 s_block_group_nr; /* block group # of this superblock */ 1090 __le32 s_feature_compat; /* compatible feature set */ 1091/*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1092 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1093/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1094/*78*/ char s_volume_name[16]; /* volume name */ 1095/*88*/ char s_last_mounted[64]; /* directory where last mounted */ 1096/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1097 /* 1098 * Performance hints. Directory preallocation should only 1099 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1100 */ 1101 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1102 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1103 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1104 /* 1105 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1106 */ 1107/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1108/*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1109 __le32 s_journal_dev; /* device number of journal file */ 1110 __le32 s_last_orphan; /* start of list of inodes to delete */ 1111 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1112 __u8 s_def_hash_version; /* Default hash version to use */ 1113 __u8 s_jnl_backup_type; 1114 __le16 s_desc_size; /* size of group descriptor */ 1115/*100*/ __le32 s_default_mount_opts; 1116 __le32 s_first_meta_bg; /* First metablock block group */ 1117 __le32 s_mkfs_time; /* When the filesystem was created */ 1118 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1119 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */ 1120/*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1121 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1122 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1123 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1124 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1125 __le32 s_flags; /* Miscellaneous flags */ 1126 __le16 s_raid_stride; /* RAID stride */ 1127 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1128 __le64 s_mmp_block; /* Block for multi-mount protection */ 1129 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1130 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1131 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1132 __le16 s_reserved_pad; 1133 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1134 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1135 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1136 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1137 snapshot's future use */ 1138 __le32 s_snapshot_list; /* inode number of the head of the 1139 on-disk snapshot list */ 1140#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1141 __le32 s_error_count; /* number of fs errors */ 1142 __le32 s_first_error_time; /* first time an error happened */ 1143 __le32 s_first_error_ino; /* inode involved in first error */ 1144 __le64 s_first_error_block; /* block involved of first error */ 1145 __u8 s_first_error_func[32]; /* function where the error happened */ 1146 __le32 s_first_error_line; /* line number where error happened */ 1147 __le32 s_last_error_time; /* most recent time of an error */ 1148 __le32 s_last_error_ino; /* inode involved in last error */ 1149 __le32 s_last_error_line; /* line number where error happened */ 1150 __le64 s_last_error_block; /* block involved of last error */ 1151 __u8 s_last_error_func[32]; /* function where the error happened */ 1152#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1153 __u8 s_mount_opts[64]; 1154 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1155 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1156 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1157 __le32 s_reserved[108]; /* Padding to the end of the block */ 1158 __le32 s_checksum; /* crc32c(superblock) */ 1159}; 1160 1161#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1162 1163#ifdef __KERNEL__ 1164 1165/* 1166 * run-time mount flags 1167 */ 1168#define EXT4_MF_MNTDIR_SAMPLED 0x0001 1169#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */ 1170 1171/* 1172 * fourth extended-fs super-block data in memory 1173 */ 1174struct ext4_sb_info { 1175 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1176 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1177 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1178 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1179 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1180 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1181 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1182 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1183 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1184 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1185 unsigned long s_overhead; /* # of fs overhead clusters */ 1186 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1187 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1188 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1189 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1190 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1191 struct buffer_head **s_group_desc; 1192 unsigned int s_mount_opt; 1193 unsigned int s_mount_opt2; 1194 unsigned int s_mount_flags; 1195 unsigned int s_def_mount_opt; 1196 ext4_fsblk_t s_sb_block; 1197 atomic64_t s_resv_clusters; 1198 kuid_t s_resuid; 1199 kgid_t s_resgid; 1200 unsigned short s_mount_state; 1201 unsigned short s_pad; 1202 int s_addr_per_block_bits; 1203 int s_desc_per_block_bits; 1204 int s_inode_size; 1205 int s_first_ino; 1206 unsigned int s_inode_readahead_blks; 1207 unsigned int s_inode_goal; 1208 spinlock_t s_next_gen_lock; 1209 u32 s_next_generation; 1210 u32 s_hash_seed[4]; 1211 int s_def_hash_version; 1212 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ 1213 struct percpu_counter s_freeclusters_counter; 1214 struct percpu_counter s_freeinodes_counter; 1215 struct percpu_counter s_dirs_counter; 1216 struct percpu_counter s_dirtyclusters_counter; 1217 struct blockgroup_lock *s_blockgroup_lock; 1218 struct proc_dir_entry *s_proc; 1219 struct kobject s_kobj; 1220 struct completion s_kobj_unregister; 1221 struct super_block *s_sb; 1222 1223 /* Journaling */ 1224 struct journal_s *s_journal; 1225 struct list_head s_orphan; 1226 struct mutex s_orphan_lock; 1227 unsigned long s_resize_flags; /* Flags indicating if there 1228 is a resizer */ 1229 unsigned long s_commit_interval; 1230 u32 s_max_batch_time; 1231 u32 s_min_batch_time; 1232 struct block_device *journal_bdev; 1233#ifdef CONFIG_QUOTA 1234 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */ 1235 int s_jquota_fmt; /* Format of quota to use */ 1236#endif 1237 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1238 struct rb_root system_blks; 1239 1240#ifdef EXTENTS_STATS 1241 /* ext4 extents stats */ 1242 unsigned long s_ext_min; 1243 unsigned long s_ext_max; 1244 unsigned long s_depth_max; 1245 spinlock_t s_ext_stats_lock; 1246 unsigned long s_ext_blocks; 1247 unsigned long s_ext_extents; 1248#endif 1249 1250 /* for buddy allocator */ 1251 struct ext4_group_info ***s_group_info; 1252 struct inode *s_buddy_cache; 1253 spinlock_t s_md_lock; 1254 unsigned short *s_mb_offsets; 1255 unsigned int *s_mb_maxs; 1256 unsigned int s_group_info_size; 1257 1258 /* tunables */ 1259 unsigned long s_stripe; 1260 unsigned int s_mb_stream_request; 1261 unsigned int s_mb_max_to_scan; 1262 unsigned int s_mb_min_to_scan; 1263 unsigned int s_mb_stats; 1264 unsigned int s_mb_order2_reqs; 1265 unsigned int s_mb_group_prealloc; 1266 unsigned int s_max_dir_size_kb; 1267 /* where last allocation was done - for stream allocation */ 1268 unsigned long s_mb_last_group; 1269 unsigned long s_mb_last_start; 1270 1271 /* stats for buddy allocator */ 1272 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1273 atomic_t s_bal_success; /* we found long enough chunks */ 1274 atomic_t s_bal_allocated; /* in blocks */ 1275 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1276 atomic_t s_bal_goals; /* goal hits */ 1277 atomic_t s_bal_breaks; /* too long searches */ 1278 atomic_t s_bal_2orders; /* 2^order hits */ 1279 spinlock_t s_bal_lock; 1280 unsigned long s_mb_buddies_generated; 1281 unsigned long long s_mb_generation_time; 1282 atomic_t s_mb_lost_chunks; 1283 atomic_t s_mb_preallocated; 1284 atomic_t s_mb_discarded; 1285 atomic_t s_lock_busy; 1286 1287 /* locality groups */ 1288 struct ext4_locality_group __percpu *s_locality_groups; 1289 1290 /* for write statistics */ 1291 unsigned long s_sectors_written_start; 1292 u64 s_kbytes_written; 1293 1294 /* the size of zero-out chunk */ 1295 unsigned int s_extent_max_zeroout_kb; 1296 1297 unsigned int s_log_groups_per_flex; 1298 struct flex_groups *s_flex_groups; 1299 ext4_group_t s_flex_groups_allocated; 1300 1301 /* workqueue for reserved extent conversions (buffered io) */ 1302 struct workqueue_struct *rsv_conversion_wq; 1303 1304 /* timer for periodic error stats printing */ 1305 struct timer_list s_err_report; 1306 1307 /* Lazy inode table initialization info */ 1308 struct ext4_li_request *s_li_request; 1309 /* Wait multiplier for lazy initialization thread */ 1310 unsigned int s_li_wait_mult; 1311 1312 /* Kernel thread for multiple mount protection */ 1313 struct task_struct *s_mmp_tsk; 1314 1315 /* record the last minlen when FITRIM is called. */ 1316 atomic_t s_last_trim_minblks; 1317 1318 /* Reference to checksum algorithm driver via cryptoapi */ 1319 struct crypto_shash *s_chksum_driver; 1320 1321 /* Precomputed FS UUID checksum for seeding other checksums */ 1322 __u32 s_csum_seed; 1323 1324 /* Reclaim extents from extent status tree */ 1325 struct shrinker s_es_shrinker; 1326 struct list_head s_es_lru; 1327 unsigned long s_es_last_sorted; 1328 struct percpu_counter s_extent_cache_cnt; 1329 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp; 1330 1331 /* Ratelimit ext4 messages. */ 1332 struct ratelimit_state s_err_ratelimit_state; 1333 struct ratelimit_state s_warning_ratelimit_state; 1334 struct ratelimit_state s_msg_ratelimit_state; 1335}; 1336 1337static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1338{ 1339 return sb->s_fs_info; 1340} 1341static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1342{ 1343 return container_of(inode, struct ext4_inode_info, vfs_inode); 1344} 1345 1346static inline struct timespec ext4_current_time(struct inode *inode) 1347{ 1348 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ? 1349 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC; 1350} 1351 1352static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1353{ 1354 return ino == EXT4_ROOT_INO || 1355 ino == EXT4_USR_QUOTA_INO || 1356 ino == EXT4_GRP_QUOTA_INO || 1357 ino == EXT4_BOOT_LOADER_INO || 1358 ino == EXT4_JOURNAL_INO || 1359 ino == EXT4_RESIZE_INO || 1360 (ino >= EXT4_FIRST_INO(sb) && 1361 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1362} 1363 1364static inline void ext4_set_io_unwritten_flag(struct inode *inode, 1365 struct ext4_io_end *io_end) 1366{ 1367 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { 1368 io_end->flag |= EXT4_IO_END_UNWRITTEN; 1369 atomic_inc(&EXT4_I(inode)->i_unwritten); 1370 } 1371} 1372 1373static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode) 1374{ 1375 return inode->i_private; 1376} 1377 1378static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io) 1379{ 1380 inode->i_private = io; 1381} 1382 1383/* 1384 * Inode dynamic state flags 1385 */ 1386enum { 1387 EXT4_STATE_JDATA, /* journaled data exists */ 1388 EXT4_STATE_NEW, /* inode is newly created */ 1389 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1390 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1391 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1392 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1393 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/ 1394 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1395 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */ 1396 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read 1397 nolocking */ 1398 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1399 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */ 1400 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1401}; 1402 1403#define EXT4_INODE_BIT_FNS(name, field, offset) \ 1404static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1405{ \ 1406 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1407} \ 1408static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1409{ \ 1410 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1411} \ 1412static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1413{ \ 1414 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1415} 1416 1417/* Add these declarations here only so that these functions can be 1418 * found by name. Otherwise, they are very hard to locate. */ 1419static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1420static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1421static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1422EXT4_INODE_BIT_FNS(flag, flags, 0) 1423 1424/* Add these declarations here only so that these functions can be 1425 * found by name. Otherwise, they are very hard to locate. */ 1426static inline int ext4_test_inode_state(struct inode *inode, int bit); 1427static inline void ext4_set_inode_state(struct inode *inode, int bit); 1428static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1429#if (BITS_PER_LONG < 64) 1430EXT4_INODE_BIT_FNS(state, state_flags, 0) 1431 1432static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1433{ 1434 (ei)->i_state_flags = 0; 1435} 1436#else 1437EXT4_INODE_BIT_FNS(state, flags, 32) 1438 1439static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1440{ 1441 /* We depend on the fact that callers will set i_flags */ 1442} 1443#endif 1444#else 1445/* Assume that user mode programs are passing in an ext4fs superblock, not 1446 * a kernel struct super_block. This will allow us to call the feature-test 1447 * macros from user land. */ 1448#define EXT4_SB(sb) (sb) 1449#endif 1450 1451#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 1452 1453/* 1454 * Codes for operating systems 1455 */ 1456#define EXT4_OS_LINUX 0 1457#define EXT4_OS_HURD 1 1458#define EXT4_OS_MASIX 2 1459#define EXT4_OS_FREEBSD 3 1460#define EXT4_OS_LITES 4 1461 1462/* 1463 * Revision levels 1464 */ 1465#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 1466#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 1467 1468#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV 1469#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 1470 1471#define EXT4_GOOD_OLD_INODE_SIZE 128 1472 1473/* 1474 * Feature set definitions 1475 */ 1476 1477#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \ 1478 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0) 1479#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \ 1480 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0) 1481#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \ 1482 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0) 1483#define EXT4_SET_COMPAT_FEATURE(sb,mask) \ 1484 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask) 1485#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \ 1486 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask) 1487#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \ 1488 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask) 1489#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \ 1490 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask) 1491#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \ 1492 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask) 1493#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \ 1494 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask) 1495 1496#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 1497#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 1498#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 1499#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 1500#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 1501#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 1502 1503#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 1504#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 1505#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 1506#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 1507#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 1508#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 1509#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 1510#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 1511#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 1512/* 1513 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 1514 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 1515 * all other data structures' checksums. However, the METADATA_CSUM and 1516 * GDT_CSUM bits are mutually exclusive. 1517 */ 1518#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 1519 1520#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 1521#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 1522#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 1523#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 1524#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 1525#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 1526#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 1527#define EXT4_FEATURE_INCOMPAT_MMP 0x0100 1528#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 1529#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 1530#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 1531#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */ 1532#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 1533#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 1534 1535#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1536#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1537 EXT4_FEATURE_INCOMPAT_META_BG) 1538#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1539 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1540 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1541 1542#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 1543#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1544 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1545 EXT4_FEATURE_INCOMPAT_META_BG) 1546#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1547 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1548 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 1549 1550#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR 1551#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 1552 EXT4_FEATURE_INCOMPAT_RECOVER| \ 1553 EXT4_FEATURE_INCOMPAT_META_BG| \ 1554 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 1555 EXT4_FEATURE_INCOMPAT_64BIT| \ 1556 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 1557 EXT4_FEATURE_INCOMPAT_MMP | \ 1558 EXT4_FEATURE_INCOMPAT_INLINE_DATA) 1559#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 1560 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 1561 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 1562 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 1563 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 1564 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 1565 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 1566 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 1567 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 1568 EXT4_FEATURE_RO_COMPAT_QUOTA) 1569 1570/* 1571 * Default values for user and/or group using reserved blocks 1572 */ 1573#define EXT4_DEF_RESUID 0 1574#define EXT4_DEF_RESGID 0 1575 1576#define EXT4_DEF_INODE_READAHEAD_BLKS 32 1577 1578/* 1579 * Default mount options 1580 */ 1581#define EXT4_DEFM_DEBUG 0x0001 1582#define EXT4_DEFM_BSDGROUPS 0x0002 1583#define EXT4_DEFM_XATTR_USER 0x0004 1584#define EXT4_DEFM_ACL 0x0008 1585#define EXT4_DEFM_UID16 0x0010 1586#define EXT4_DEFM_JMODE 0x0060 1587#define EXT4_DEFM_JMODE_DATA 0x0020 1588#define EXT4_DEFM_JMODE_ORDERED 0x0040 1589#define EXT4_DEFM_JMODE_WBACK 0x0060 1590#define EXT4_DEFM_NOBARRIER 0x0100 1591#define EXT4_DEFM_BLOCK_VALIDITY 0x0200 1592#define EXT4_DEFM_DISCARD 0x0400 1593#define EXT4_DEFM_NODELALLOC 0x0800 1594 1595/* 1596 * Default journal batch times 1597 */ 1598#define EXT4_DEF_MIN_BATCH_TIME 0 1599#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 1600 1601/* 1602 * Minimum number of groups in a flexgroup before we separate out 1603 * directories into the first block group of a flexgroup 1604 */ 1605#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 1606 1607/* 1608 * Structure of a directory entry 1609 */ 1610#define EXT4_NAME_LEN 255 1611 1612struct ext4_dir_entry { 1613 __le32 inode; /* Inode number */ 1614 __le16 rec_len; /* Directory entry length */ 1615 __le16 name_len; /* Name length */ 1616 char name[EXT4_NAME_LEN]; /* File name */ 1617}; 1618 1619/* 1620 * The new version of the directory entry. Since EXT4 structures are 1621 * stored in intel byte order, and the name_len field could never be 1622 * bigger than 255 chars, it's safe to reclaim the extra byte for the 1623 * file_type field. 1624 */ 1625struct ext4_dir_entry_2 { 1626 __le32 inode; /* Inode number */ 1627 __le16 rec_len; /* Directory entry length */ 1628 __u8 name_len; /* Name length */ 1629 __u8 file_type; 1630 char name[EXT4_NAME_LEN]; /* File name */ 1631}; 1632 1633/* 1634 * This is a bogus directory entry at the end of each leaf block that 1635 * records checksums. 1636 */ 1637struct ext4_dir_entry_tail { 1638 __le32 det_reserved_zero1; /* Pretend to be unused */ 1639 __le16 det_rec_len; /* 12 */ 1640 __u8 det_reserved_zero2; /* Zero name length */ 1641 __u8 det_reserved_ft; /* 0xDE, fake file type */ 1642 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 1643}; 1644 1645#define EXT4_DIRENT_TAIL(block, blocksize) \ 1646 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 1647 ((blocksize) - \ 1648 sizeof(struct ext4_dir_entry_tail)))) 1649 1650/* 1651 * Ext4 directory file types. Only the low 3 bits are used. The 1652 * other bits are reserved for now. 1653 */ 1654#define EXT4_FT_UNKNOWN 0 1655#define EXT4_FT_REG_FILE 1 1656#define EXT4_FT_DIR 2 1657#define EXT4_FT_CHRDEV 3 1658#define EXT4_FT_BLKDEV 4 1659#define EXT4_FT_FIFO 5 1660#define EXT4_FT_SOCK 6 1661#define EXT4_FT_SYMLINK 7 1662 1663#define EXT4_FT_MAX 8 1664 1665#define EXT4_FT_DIR_CSUM 0xDE 1666 1667/* 1668 * EXT4_DIR_PAD defines the directory entries boundaries 1669 * 1670 * NOTE: It must be a multiple of 4 1671 */ 1672#define EXT4_DIR_PAD 4 1673#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 1674#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \ 1675 ~EXT4_DIR_ROUND) 1676#define EXT4_MAX_REC_LEN ((1<<16)-1) 1677 1678/* 1679 * If we ever get support for fs block sizes > page_size, we'll need 1680 * to remove the #if statements in the next two functions... 1681 */ 1682static inline unsigned int 1683ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 1684{ 1685 unsigned len = le16_to_cpu(dlen); 1686 1687#if (PAGE_CACHE_SIZE >= 65536) 1688 if (len == EXT4_MAX_REC_LEN || len == 0) 1689 return blocksize; 1690 return (len & 65532) | ((len & 3) << 16); 1691#else 1692 return len; 1693#endif 1694} 1695 1696static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 1697{ 1698 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)) 1699 BUG(); 1700#if (PAGE_CACHE_SIZE >= 65536) 1701 if (len < 65536) 1702 return cpu_to_le16(len); 1703 if (len == blocksize) { 1704 if (blocksize == 65536) 1705 return cpu_to_le16(EXT4_MAX_REC_LEN); 1706 else 1707 return cpu_to_le16(0); 1708 } 1709 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 1710#else 1711 return cpu_to_le16(len); 1712#endif 1713} 1714 1715/* 1716 * Hash Tree Directory indexing 1717 * (c) Daniel Phillips, 2001 1718 */ 1719 1720#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \ 1721 EXT4_FEATURE_COMPAT_DIR_INDEX) && \ 1722 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 1723#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX) 1724#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 1725 1726/* Legal values for the dx_root hash_version field: */ 1727 1728#define DX_HASH_LEGACY 0 1729#define DX_HASH_HALF_MD4 1 1730#define DX_HASH_TEA 2 1731#define DX_HASH_LEGACY_UNSIGNED 3 1732#define DX_HASH_HALF_MD4_UNSIGNED 4 1733#define DX_HASH_TEA_UNSIGNED 5 1734 1735static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc, 1736 const void *address, unsigned int length) 1737{ 1738 struct { 1739 struct shash_desc shash; 1740 char ctx[4]; 1741 } desc; 1742 int err; 1743 1744 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx)); 1745 1746 desc.shash.tfm = sbi->s_chksum_driver; 1747 desc.shash.flags = 0; 1748 *(u32 *)desc.ctx = crc; 1749 1750 err = crypto_shash_update(&desc.shash, address, length); 1751 BUG_ON(err); 1752 1753 return *(u32 *)desc.ctx; 1754} 1755 1756#ifdef __KERNEL__ 1757 1758/* hash info structure used by the directory hash */ 1759struct dx_hash_info 1760{ 1761 u32 hash; 1762 u32 minor_hash; 1763 int hash_version; 1764 u32 *seed; 1765}; 1766 1767 1768/* 32 and 64 bit signed EOF for dx directories */ 1769#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 1770#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 1771 1772 1773/* 1774 * Control parameters used by ext4_htree_next_block 1775 */ 1776#define HASH_NB_ALWAYS 1 1777 1778 1779/* 1780 * Describe an inode's exact location on disk and in memory 1781 */ 1782struct ext4_iloc 1783{ 1784 struct buffer_head *bh; 1785 unsigned long offset; 1786 ext4_group_t block_group; 1787}; 1788 1789static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 1790{ 1791 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 1792} 1793 1794/* 1795 * This structure is stuffed into the struct file's private_data field 1796 * for directories. It is where we put information so that we can do 1797 * readdir operations in hash tree order. 1798 */ 1799struct dir_private_info { 1800 struct rb_root root; 1801 struct rb_node *curr_node; 1802 struct fname *extra_fname; 1803 loff_t last_pos; 1804 __u32 curr_hash; 1805 __u32 curr_minor_hash; 1806 __u32 next_hash; 1807}; 1808 1809/* calculate the first block number of the group */ 1810static inline ext4_fsblk_t 1811ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 1812{ 1813 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 1814 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 1815} 1816 1817/* 1818 * Special error return code only used by dx_probe() and its callers. 1819 */ 1820#define ERR_BAD_DX_DIR -75000 1821 1822/* 1823 * Timeout and state flag for lazy initialization inode thread. 1824 */ 1825#define EXT4_DEF_LI_WAIT_MULT 10 1826#define EXT4_DEF_LI_MAX_START_DELAY 5 1827#define EXT4_LAZYINIT_QUIT 0x0001 1828#define EXT4_LAZYINIT_RUNNING 0x0002 1829 1830/* 1831 * Lazy inode table initialization info 1832 */ 1833struct ext4_lazy_init { 1834 unsigned long li_state; 1835 struct list_head li_request_list; 1836 struct mutex li_list_mtx; 1837}; 1838 1839struct ext4_li_request { 1840 struct super_block *lr_super; 1841 struct ext4_sb_info *lr_sbi; 1842 ext4_group_t lr_next_group; 1843 struct list_head lr_request; 1844 unsigned long lr_next_sched; 1845 unsigned long lr_timeout; 1846}; 1847 1848struct ext4_features { 1849 struct kobject f_kobj; 1850 struct completion f_kobj_unregister; 1851}; 1852 1853/* 1854 * This structure will be used for multiple mount protection. It will be 1855 * written into the block number saved in the s_mmp_block field in the 1856 * superblock. Programs that check MMP should assume that if 1857 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 1858 * to use the filesystem, regardless of how old the timestamp is. 1859 */ 1860#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 1861#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 1862#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 1863#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 1864 1865struct mmp_struct { 1866 __le32 mmp_magic; /* Magic number for MMP */ 1867 __le32 mmp_seq; /* Sequence no. updated periodically */ 1868 1869 /* 1870 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 1871 * purposes and do not affect the correctness of the algorithm 1872 */ 1873 __le64 mmp_time; /* Time last updated */ 1874 char mmp_nodename[64]; /* Node which last updated MMP block */ 1875 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 1876 1877 /* 1878 * mmp_check_interval is used to verify if the MMP block has been 1879 * updated on the block device. The value is updated based on the 1880 * maximum time to write the MMP block during an update cycle. 1881 */ 1882 __le16 mmp_check_interval; 1883 1884 __le16 mmp_pad1; 1885 __le32 mmp_pad2[226]; 1886 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 1887}; 1888 1889/* arguments passed to the mmp thread */ 1890struct mmpd_data { 1891 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 1892 struct super_block *sb; /* super block of the fs */ 1893}; 1894 1895/* 1896 * Check interval multiplier 1897 * The MMP block is written every update interval and initially checked every 1898 * update interval x the multiplier (the value is then adapted based on the 1899 * write latency). The reason is that writes can be delayed under load and we 1900 * don't want readers to incorrectly assume that the filesystem is no longer 1901 * in use. 1902 */ 1903#define EXT4_MMP_CHECK_MULT 2UL 1904 1905/* 1906 * Minimum interval for MMP checking in seconds. 1907 */ 1908#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 1909 1910/* 1911 * Maximum interval for MMP checking in seconds. 1912 */ 1913#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 1914 1915/* 1916 * Function prototypes 1917 */ 1918 1919/* 1920 * Ok, these declarations are also in <linux/kernel.h> but none of the 1921 * ext4 source programs needs to include it so they are duplicated here. 1922 */ 1923# define NORET_TYPE /**/ 1924# define ATTRIB_NORET __attribute__((noreturn)) 1925# define NORET_AND noreturn, 1926 1927/* bitmap.c */ 1928extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 1929void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1930 struct ext4_group_desc *gdp, 1931 struct buffer_head *bh, int sz); 1932int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1933 struct ext4_group_desc *gdp, 1934 struct buffer_head *bh, int sz); 1935void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group, 1936 struct ext4_group_desc *gdp, 1937 struct buffer_head *bh); 1938int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, 1939 struct ext4_group_desc *gdp, 1940 struct buffer_head *bh); 1941 1942/* balloc.c */ 1943extern void ext4_get_group_no_and_offset(struct super_block *sb, 1944 ext4_fsblk_t blocknr, 1945 ext4_group_t *blockgrpp, 1946 ext4_grpblk_t *offsetp); 1947extern ext4_group_t ext4_get_group_number(struct super_block *sb, 1948 ext4_fsblk_t block); 1949 1950extern void ext4_validate_block_bitmap(struct super_block *sb, 1951 struct ext4_group_desc *desc, 1952 ext4_group_t block_group, 1953 struct buffer_head *bh); 1954extern unsigned int ext4_block_group(struct super_block *sb, 1955 ext4_fsblk_t blocknr); 1956extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb, 1957 ext4_fsblk_t blocknr); 1958extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 1959extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 1960 ext4_group_t group); 1961extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 1962 ext4_fsblk_t goal, 1963 unsigned int flags, 1964 unsigned long *count, 1965 int *errp); 1966extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 1967 s64 nclusters, unsigned int flags); 1968extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 1969extern void ext4_check_blocks_bitmap(struct super_block *); 1970extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 1971 ext4_group_t block_group, 1972 struct buffer_head ** bh); 1973extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 1974 1975extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 1976 ext4_group_t block_group); 1977extern int ext4_wait_block_bitmap(struct super_block *sb, 1978 ext4_group_t block_group, 1979 struct buffer_head *bh); 1980extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 1981 ext4_group_t block_group); 1982extern void ext4_init_block_bitmap(struct super_block *sb, 1983 struct buffer_head *bh, 1984 ext4_group_t group, 1985 struct ext4_group_desc *desc); 1986extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 1987 ext4_group_t block_group, 1988 struct ext4_group_desc *gdp); 1989extern unsigned ext4_num_overhead_clusters(struct super_block *sb, 1990 ext4_group_t block_group, 1991 struct ext4_group_desc *gdp); 1992ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 1993 1994/* dir.c */ 1995extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 1996 struct file *, 1997 struct ext4_dir_entry_2 *, 1998 struct buffer_head *, char *, int, 1999 unsigned int); 2000#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2001 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2002 (de), (bh), (buf), (size), (offset))) 2003extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2004 __u32 minor_hash, 2005 struct ext4_dir_entry_2 *dirent); 2006extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2007extern int ext4_find_dest_de(struct inode *dir, struct inode *inode, 2008 struct buffer_head *bh, 2009 void *buf, int buf_size, 2010 const char *name, int namelen, 2011 struct ext4_dir_entry_2 **dest_de); 2012void ext4_insert_dentry(struct inode *inode, 2013 struct ext4_dir_entry_2 *de, 2014 int buf_size, 2015 const char *name, int namelen); 2016static inline void ext4_update_dx_flag(struct inode *inode) 2017{ 2018 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb, 2019 EXT4_FEATURE_COMPAT_DIR_INDEX)) 2020 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2021} 2022static unsigned char ext4_filetype_table[] = { 2023 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2024}; 2025 2026static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2027{ 2028 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) || 2029 (filetype >= EXT4_FT_MAX)) 2030 return DT_UNKNOWN; 2031 2032 return ext4_filetype_table[filetype]; 2033} 2034 2035/* fsync.c */ 2036extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2037 2038/* hash.c */ 2039extern int ext4fs_dirhash(const char *name, int len, struct 2040 dx_hash_info *hinfo); 2041 2042/* ialloc.c */ 2043extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t, 2044 const struct qstr *qstr, __u32 goal, 2045 uid_t *owner, int handle_type, 2046 unsigned int line_no, int nblocks); 2047 2048#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \ 2049 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \ 2050 0, 0, 0) 2051#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \ 2052 type, nblocks) \ 2053 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \ 2054 (type), __LINE__, (nblocks)) 2055 2056 2057extern void ext4_free_inode(handle_t *, struct inode *); 2058extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2059extern unsigned long ext4_count_free_inodes(struct super_block *); 2060extern unsigned long ext4_count_dirs(struct super_block *); 2061extern void ext4_check_inodes_bitmap(struct super_block *); 2062extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2063extern int ext4_init_inode_table(struct super_block *sb, 2064 ext4_group_t group, int barrier); 2065extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2066 2067/* mballoc.c */ 2068extern long ext4_mb_stats; 2069extern long ext4_mb_max_to_scan; 2070extern int ext4_mb_init(struct super_block *); 2071extern int ext4_mb_release(struct super_block *); 2072extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2073 struct ext4_allocation_request *, int *); 2074extern int ext4_mb_reserve_blocks(struct super_block *, int); 2075extern void ext4_discard_preallocations(struct inode *); 2076extern int __init ext4_init_mballoc(void); 2077extern void ext4_exit_mballoc(void); 2078extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2079 struct buffer_head *bh, ext4_fsblk_t block, 2080 unsigned long count, int flags); 2081extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2082 ext4_group_t ngroups); 2083extern int ext4_mb_add_groupinfo(struct super_block *sb, 2084 ext4_group_t i, struct ext4_group_desc *desc); 2085extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 2086 ext4_fsblk_t block, unsigned long count); 2087extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 2088 2089/* inode.c */ 2090struct buffer_head *ext4_getblk(handle_t *, struct inode *, 2091 ext4_lblk_t, int, int *); 2092struct buffer_head *ext4_bread(handle_t *, struct inode *, 2093 ext4_lblk_t, int, int *); 2094int ext4_get_block_write(struct inode *inode, sector_t iblock, 2095 struct buffer_head *bh_result, int create); 2096int ext4_get_block(struct inode *inode, sector_t iblock, 2097 struct buffer_head *bh_result, int create); 2098int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 2099 struct buffer_head *bh, int create); 2100int ext4_walk_page_buffers(handle_t *handle, 2101 struct buffer_head *head, 2102 unsigned from, 2103 unsigned to, 2104 int *partial, 2105 int (*fn)(handle_t *handle, 2106 struct buffer_head *bh)); 2107int do_journal_get_write_access(handle_t *handle, 2108 struct buffer_head *bh); 2109#define FALL_BACK_TO_NONDELALLOC 1 2110#define CONVERT_INLINE_DATA 2 2111 2112extern struct inode *ext4_iget(struct super_block *, unsigned long); 2113extern int ext4_write_inode(struct inode *, struct writeback_control *); 2114extern int ext4_setattr(struct dentry *, struct iattr *); 2115extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry, 2116 struct kstat *stat); 2117extern void ext4_evict_inode(struct inode *); 2118extern void ext4_clear_inode(struct inode *); 2119extern int ext4_sync_inode(handle_t *, struct inode *); 2120extern void ext4_dirty_inode(struct inode *, int); 2121extern int ext4_change_inode_journal_flag(struct inode *, int); 2122extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 2123extern int ext4_inode_attach_jinode(struct inode *inode); 2124extern int ext4_can_truncate(struct inode *inode); 2125extern void ext4_truncate(struct inode *); 2126extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length); 2127extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks); 2128extern void ext4_set_inode_flags(struct inode *); 2129extern void ext4_get_inode_flags(struct ext4_inode_info *); 2130extern int ext4_alloc_da_blocks(struct inode *inode); 2131extern void ext4_set_aops(struct inode *inode); 2132extern int ext4_writepage_trans_blocks(struct inode *); 2133extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 2134extern int ext4_block_truncate_page(handle_t *handle, 2135 struct address_space *mapping, loff_t from); 2136extern int ext4_block_zero_page_range(handle_t *handle, 2137 struct address_space *mapping, loff_t from, loff_t length); 2138extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 2139 loff_t lstart, loff_t lend); 2140extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); 2141extern qsize_t *ext4_get_reserved_space(struct inode *inode); 2142extern void ext4_da_update_reserve_space(struct inode *inode, 2143 int used, int quota_claim); 2144 2145/* indirect.c */ 2146extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 2147 struct ext4_map_blocks *map, int flags); 2148extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb, 2149 const struct iovec *iov, loff_t offset, 2150 unsigned long nr_segs); 2151extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock); 2152extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 2153extern void ext4_ind_truncate(handle_t *, struct inode *inode); 2154extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode, 2155 ext4_lblk_t first, ext4_lblk_t stop); 2156 2157/* ioctl.c */ 2158extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 2159extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 2160 2161/* migrate.c */ 2162extern int ext4_ext_migrate(struct inode *); 2163extern int ext4_ind_migrate(struct inode *inode); 2164 2165/* namei.c */ 2166extern int ext4_dirent_csum_verify(struct inode *inode, 2167 struct ext4_dir_entry *dirent); 2168extern int ext4_orphan_add(handle_t *, struct inode *); 2169extern int ext4_orphan_del(handle_t *, struct inode *); 2170extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 2171 __u32 start_minor_hash, __u32 *next_hash); 2172extern int search_dir(struct buffer_head *bh, 2173 char *search_buf, 2174 int buf_size, 2175 struct inode *dir, 2176 const struct qstr *d_name, 2177 unsigned int offset, 2178 struct ext4_dir_entry_2 **res_dir); 2179extern int ext4_generic_delete_entry(handle_t *handle, 2180 struct inode *dir, 2181 struct ext4_dir_entry_2 *de_del, 2182 struct buffer_head *bh, 2183 void *entry_buf, 2184 int buf_size, 2185 int csum_size); 2186 2187/* resize.c */ 2188extern int ext4_group_add(struct super_block *sb, 2189 struct ext4_new_group_data *input); 2190extern int ext4_group_extend(struct super_block *sb, 2191 struct ext4_super_block *es, 2192 ext4_fsblk_t n_blocks_count); 2193extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 2194 2195/* super.c */ 2196extern int ext4_calculate_overhead(struct super_block *sb); 2197extern int ext4_superblock_csum_verify(struct super_block *sb, 2198 struct ext4_super_block *es); 2199extern void ext4_superblock_csum_set(struct super_block *sb); 2200extern void *ext4_kvmalloc(size_t size, gfp_t flags); 2201extern void *ext4_kvzalloc(size_t size, gfp_t flags); 2202extern void ext4_kvfree(void *ptr); 2203extern int ext4_alloc_flex_bg_array(struct super_block *sb, 2204 ext4_group_t ngroup); 2205extern const char *ext4_decode_error(struct super_block *sb, int errno, 2206 char nbuf[16]); 2207 2208extern __printf(4, 5) 2209void __ext4_error(struct super_block *, const char *, unsigned int, 2210 const char *, ...); 2211extern __printf(5, 6) 2212void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t, 2213 const char *, ...); 2214extern __printf(5, 6) 2215void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 2216 const char *, ...); 2217extern void __ext4_std_error(struct super_block *, const char *, 2218 unsigned int, int); 2219extern __printf(4, 5) 2220void __ext4_abort(struct super_block *, const char *, unsigned int, 2221 const char *, ...); 2222extern __printf(4, 5) 2223void __ext4_warning(struct super_block *, const char *, unsigned int, 2224 const char *, ...); 2225extern __printf(3, 4) 2226void __ext4_msg(struct super_block *, const char *, const char *, ...); 2227extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 2228 const char *, unsigned int, const char *); 2229extern __printf(7, 8) 2230void __ext4_grp_locked_error(const char *, unsigned int, 2231 struct super_block *, ext4_group_t, 2232 unsigned long, ext4_fsblk_t, 2233 const char *, ...); 2234 2235#ifdef CONFIG_PRINTK 2236 2237#define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2238 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__) 2239#define ext4_error_file(file, func, line, block, fmt, ...) \ 2240 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 2241#define ext4_error(sb, fmt, ...) \ 2242 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2243#define ext4_abort(sb, fmt, ...) \ 2244 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2245#define ext4_warning(sb, fmt, ...) \ 2246 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 2247#define ext4_msg(sb, level, fmt, ...) \ 2248 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 2249#define dump_mmp_msg(sb, mmp, msg) \ 2250 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 2251#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2252 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 2253 fmt, ##__VA_ARGS__) 2254 2255#else 2256 2257#define ext4_error_inode(inode, func, line, block, fmt, ...) \ 2258do { \ 2259 no_printk(fmt, ##__VA_ARGS__); \ 2260 __ext4_error_inode(inode, "", 0, block, " "); \ 2261} while (0) 2262#define ext4_error_file(file, func, line, block, fmt, ...) \ 2263do { \ 2264 no_printk(fmt, ##__VA_ARGS__); \ 2265 __ext4_error_file(file, "", 0, block, " "); \ 2266} while (0) 2267#define ext4_error(sb, fmt, ...) \ 2268do { \ 2269 no_printk(fmt, ##__VA_ARGS__); \ 2270 __ext4_error(sb, "", 0, " "); \ 2271} while (0) 2272#define ext4_abort(sb, fmt, ...) \ 2273do { \ 2274 no_printk(fmt, ##__VA_ARGS__); \ 2275 __ext4_abort(sb, "", 0, " "); \ 2276} while (0) 2277#define ext4_warning(sb, fmt, ...) \ 2278do { \ 2279 no_printk(fmt, ##__VA_ARGS__); \ 2280 __ext4_warning(sb, "", 0, " "); \ 2281} while (0) 2282#define ext4_msg(sb, level, fmt, ...) \ 2283do { \ 2284 no_printk(fmt, ##__VA_ARGS__); \ 2285 __ext4_msg(sb, "", " "); \ 2286} while (0) 2287#define dump_mmp_msg(sb, mmp, msg) \ 2288 __dump_mmp_msg(sb, mmp, "", 0, "") 2289#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 2290do { \ 2291 no_printk(fmt, ##__VA_ARGS__); \ 2292 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 2293} while (0) 2294 2295#endif 2296 2297extern void ext4_update_dynamic_rev(struct super_block *sb); 2298extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb, 2299 __u32 compat); 2300extern int ext4_update_rocompat_feature(handle_t *handle, 2301 struct super_block *sb, __u32 rocompat); 2302extern int ext4_update_incompat_feature(handle_t *handle, 2303 struct super_block *sb, __u32 incompat); 2304extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 2305 struct ext4_group_desc *bg); 2306extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 2307 struct ext4_group_desc *bg); 2308extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 2309 struct ext4_group_desc *bg); 2310extern __u32 ext4_free_group_clusters(struct super_block *sb, 2311 struct ext4_group_desc *bg); 2312extern __u32 ext4_free_inodes_count(struct super_block *sb, 2313 struct ext4_group_desc *bg); 2314extern __u32 ext4_used_dirs_count(struct super_block *sb, 2315 struct ext4_group_desc *bg); 2316extern __u32 ext4_itable_unused_count(struct super_block *sb, 2317 struct ext4_group_desc *bg); 2318extern void ext4_block_bitmap_set(struct super_block *sb, 2319 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2320extern void ext4_inode_bitmap_set(struct super_block *sb, 2321 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2322extern void ext4_inode_table_set(struct super_block *sb, 2323 struct ext4_group_desc *bg, ext4_fsblk_t blk); 2324extern void ext4_free_group_clusters_set(struct super_block *sb, 2325 struct ext4_group_desc *bg, 2326 __u32 count); 2327extern void ext4_free_inodes_set(struct super_block *sb, 2328 struct ext4_group_desc *bg, __u32 count); 2329extern void ext4_used_dirs_set(struct super_block *sb, 2330 struct ext4_group_desc *bg, __u32 count); 2331extern void ext4_itable_unused_set(struct super_block *sb, 2332 struct ext4_group_desc *bg, __u32 count); 2333extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 2334 struct ext4_group_desc *gdp); 2335extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 2336 struct ext4_group_desc *gdp); 2337extern int ext4_register_li_request(struct super_block *sb, 2338 ext4_group_t first_not_zeroed); 2339 2340static inline int ext4_has_group_desc_csum(struct super_block *sb) 2341{ 2342 return EXT4_HAS_RO_COMPAT_FEATURE(sb, 2343 EXT4_FEATURE_RO_COMPAT_GDT_CSUM | 2344 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM); 2345} 2346 2347static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 2348{ 2349 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) | 2350 le32_to_cpu(es->s_blocks_count_lo); 2351} 2352 2353static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 2354{ 2355 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) | 2356 le32_to_cpu(es->s_r_blocks_count_lo); 2357} 2358 2359static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 2360{ 2361 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) | 2362 le32_to_cpu(es->s_free_blocks_count_lo); 2363} 2364 2365static inline void ext4_blocks_count_set(struct ext4_super_block *es, 2366 ext4_fsblk_t blk) 2367{ 2368 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 2369 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 2370} 2371 2372static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 2373 ext4_fsblk_t blk) 2374{ 2375 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 2376 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 2377} 2378 2379static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 2380 ext4_fsblk_t blk) 2381{ 2382 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 2383 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 2384} 2385 2386static inline loff_t ext4_isize(struct ext4_inode *raw_inode) 2387{ 2388 if (S_ISREG(le16_to_cpu(raw_inode->i_mode))) 2389 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 2390 le32_to_cpu(raw_inode->i_size_lo); 2391 else 2392 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 2393} 2394 2395static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 2396{ 2397 raw_inode->i_size_lo = cpu_to_le32(i_size); 2398 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 2399} 2400 2401static inline 2402struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2403 ext4_group_t group) 2404{ 2405 struct ext4_group_info ***grp_info; 2406 long indexv, indexh; 2407 BUG_ON(group >= EXT4_SB(sb)->s_groups_count); 2408 grp_info = EXT4_SB(sb)->s_group_info; 2409 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb)); 2410 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1); 2411 return grp_info[indexv][indexh]; 2412} 2413 2414/* 2415 * Reading s_groups_count requires using smp_rmb() afterwards. See 2416 * the locking protocol documented in the comments of ext4_group_add() 2417 * in resize.c 2418 */ 2419static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 2420{ 2421 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 2422 2423 smp_rmb(); 2424 return ngroups; 2425} 2426 2427static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 2428 ext4_group_t block_group) 2429{ 2430 return block_group >> sbi->s_log_groups_per_flex; 2431} 2432 2433static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 2434{ 2435 return 1 << sbi->s_log_groups_per_flex; 2436} 2437 2438#define ext4_std_error(sb, errno) \ 2439do { \ 2440 if ((errno)) \ 2441 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 2442} while (0) 2443 2444#ifdef CONFIG_SMP 2445/* Each CPU can accumulate percpu_counter_batch clusters in their local 2446 * counters. So we need to make sure we have free clusters more 2447 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 2448 */ 2449#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 2450#else 2451#define EXT4_FREECLUSTERS_WATERMARK 0 2452#endif 2453 2454/* Update i_disksize. Requires i_mutex to avoid races with truncate */ 2455static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 2456{ 2457 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 2458 !mutex_is_locked(&inode->i_mutex)); 2459 down_write(&EXT4_I(inode)->i_data_sem); 2460 if (newsize > EXT4_I(inode)->i_disksize) 2461 EXT4_I(inode)->i_disksize = newsize; 2462 up_write(&EXT4_I(inode)->i_data_sem); 2463} 2464 2465/* 2466 * Update i_disksize after writeback has been started. Races with truncate 2467 * are avoided by checking i_size under i_data_sem. 2468 */ 2469static inline void ext4_wb_update_i_disksize(struct inode *inode, loff_t newsize) 2470{ 2471 loff_t i_size; 2472 2473 down_write(&EXT4_I(inode)->i_data_sem); 2474 i_size = i_size_read(inode); 2475 if (newsize > i_size) 2476 newsize = i_size; 2477 if (newsize > EXT4_I(inode)->i_disksize) 2478 EXT4_I(inode)->i_disksize = newsize; 2479 up_write(&EXT4_I(inode)->i_data_sem); 2480} 2481 2482struct ext4_group_info { 2483 unsigned long bb_state; 2484 struct rb_root bb_free_root; 2485 ext4_grpblk_t bb_first_free; /* first free block */ 2486 ext4_grpblk_t bb_free; /* total free blocks */ 2487 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 2488 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 2489 struct list_head bb_prealloc_list; 2490#ifdef DOUBLE_CHECK 2491 void *bb_bitmap; 2492#endif 2493 struct rw_semaphore alloc_sem; 2494 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 2495 * regions, index is order. 2496 * bb_counters[3] = 5 means 2497 * 5 free 8-block regions. */ 2498}; 2499 2500#define EXT4_GROUP_INFO_NEED_INIT_BIT 0 2501#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 2502#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 2503#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 2504 2505#define EXT4_MB_GRP_NEED_INIT(grp) \ 2506 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 2507#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 2508 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2509#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 2510 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 2511 2512#define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 2513 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2514#define EXT4_MB_GRP_SET_TRIMMED(grp) \ 2515 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2516#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 2517 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 2518 2519#define EXT4_MAX_CONTENTION 8 2520#define EXT4_CONTENTION_THRESHOLD 2 2521 2522static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 2523 ext4_group_t group) 2524{ 2525 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 2526} 2527 2528/* 2529 * Returns true if the filesystem is busy enough that attempts to 2530 * access the block group locks has run into contention. 2531 */ 2532static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 2533{ 2534 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 2535} 2536 2537static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 2538{ 2539 spinlock_t *lock = ext4_group_lock_ptr(sb, group); 2540 if (spin_trylock(lock)) 2541 /* 2542 * We're able to grab the lock right away, so drop the 2543 * lock contention counter. 2544 */ 2545 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 2546 else { 2547 /* 2548 * The lock is busy, so bump the contention counter, 2549 * and then wait on the spin lock. 2550 */ 2551 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 2552 EXT4_MAX_CONTENTION); 2553 spin_lock(lock); 2554 } 2555} 2556 2557static inline void ext4_unlock_group(struct super_block *sb, 2558 ext4_group_t group) 2559{ 2560 spin_unlock(ext4_group_lock_ptr(sb, group)); 2561} 2562 2563/* 2564 * Block validity checking 2565 */ 2566#define ext4_check_indirect_blockref(inode, bh) \ 2567 ext4_check_blockref(__func__, __LINE__, inode, \ 2568 (__le32 *)(bh)->b_data, \ 2569 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 2570 2571#define ext4_ind_check_inode(inode) \ 2572 ext4_check_blockref(__func__, __LINE__, inode, \ 2573 EXT4_I(inode)->i_data, \ 2574 EXT4_NDIR_BLOCKS) 2575 2576/* 2577 * Inodes and files operations 2578 */ 2579 2580/* dir.c */ 2581extern const struct file_operations ext4_dir_operations; 2582 2583/* file.c */ 2584extern const struct inode_operations ext4_file_inode_operations; 2585extern const struct file_operations ext4_file_operations; 2586extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 2587extern void ext4_unwritten_wait(struct inode *inode); 2588 2589/* inline.c */ 2590extern int ext4_has_inline_data(struct inode *inode); 2591extern int ext4_get_inline_size(struct inode *inode); 2592extern int ext4_get_max_inline_size(struct inode *inode); 2593extern int ext4_find_inline_data_nolock(struct inode *inode); 2594extern void ext4_write_inline_data(struct inode *inode, 2595 struct ext4_iloc *iloc, 2596 void *buffer, loff_t pos, 2597 unsigned int len); 2598extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode, 2599 unsigned int len); 2600extern int ext4_init_inline_data(handle_t *handle, struct inode *inode, 2601 unsigned int len); 2602extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 2603 2604extern int ext4_readpage_inline(struct inode *inode, struct page *page); 2605extern int ext4_try_to_write_inline_data(struct address_space *mapping, 2606 struct inode *inode, 2607 loff_t pos, unsigned len, 2608 unsigned flags, 2609 struct page **pagep); 2610extern int ext4_write_inline_data_end(struct inode *inode, 2611 loff_t pos, unsigned len, 2612 unsigned copied, 2613 struct page *page); 2614extern struct buffer_head * 2615ext4_journalled_write_inline_data(struct inode *inode, 2616 unsigned len, 2617 struct page *page); 2618extern int ext4_da_write_inline_data_begin(struct address_space *mapping, 2619 struct inode *inode, 2620 loff_t pos, unsigned len, 2621 unsigned flags, 2622 struct page **pagep, 2623 void **fsdata); 2624extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos, 2625 unsigned len, unsigned copied, 2626 struct page *page); 2627extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry, 2628 struct inode *inode); 2629extern int ext4_try_create_inline_dir(handle_t *handle, 2630 struct inode *parent, 2631 struct inode *inode); 2632extern int ext4_read_inline_dir(struct file *filp, 2633 struct dir_context *ctx, 2634 int *has_inline_data); 2635extern int htree_inlinedir_to_tree(struct file *dir_file, 2636 struct inode *dir, ext4_lblk_t block, 2637 struct dx_hash_info *hinfo, 2638 __u32 start_hash, __u32 start_minor_hash, 2639 int *has_inline_data); 2640extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 2641 const struct qstr *d_name, 2642 struct ext4_dir_entry_2 **res_dir, 2643 int *has_inline_data); 2644extern int ext4_delete_inline_entry(handle_t *handle, 2645 struct inode *dir, 2646 struct ext4_dir_entry_2 *de_del, 2647 struct buffer_head *bh, 2648 int *has_inline_data); 2649extern int empty_inline_dir(struct inode *dir, int *has_inline_data); 2650extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 2651 struct ext4_dir_entry_2 **parent_de, 2652 int *retval); 2653extern int ext4_inline_data_fiemap(struct inode *inode, 2654 struct fiemap_extent_info *fieinfo, 2655 int *has_inline); 2656extern int ext4_try_to_evict_inline_data(handle_t *handle, 2657 struct inode *inode, 2658 int needed); 2659extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); 2660 2661extern int ext4_convert_inline_data(struct inode *inode); 2662 2663/* namei.c */ 2664extern const struct inode_operations ext4_dir_inode_operations; 2665extern const struct inode_operations ext4_special_inode_operations; 2666extern struct dentry *ext4_get_parent(struct dentry *child); 2667extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode, 2668 struct ext4_dir_entry_2 *de, 2669 int blocksize, int csum_size, 2670 unsigned int parent_ino, int dotdot_real_len); 2671extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t, 2672 unsigned int blocksize); 2673extern int ext4_handle_dirty_dirent_node(handle_t *handle, 2674 struct inode *inode, 2675 struct buffer_head *bh); 2676#define S_SHIFT 12 2677static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = { 2678 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 2679 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 2680 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 2681 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 2682 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 2683 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 2684 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 2685}; 2686 2687static inline void ext4_set_de_type(struct super_block *sb, 2688 struct ext4_dir_entry_2 *de, 2689 umode_t mode) { 2690 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE)) 2691 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 2692} 2693 2694 2695/* symlink.c */ 2696extern const struct inode_operations ext4_symlink_inode_operations; 2697extern const struct inode_operations ext4_fast_symlink_inode_operations; 2698 2699/* block_validity */ 2700extern void ext4_release_system_zone(struct super_block *sb); 2701extern int ext4_setup_system_zone(struct super_block *sb); 2702extern int __init ext4_init_system_zone(void); 2703extern void ext4_exit_system_zone(void); 2704extern int ext4_data_block_valid(struct ext4_sb_info *sbi, 2705 ext4_fsblk_t start_blk, 2706 unsigned int count); 2707extern int ext4_check_blockref(const char *, unsigned int, 2708 struct inode *, __le32 *, unsigned int); 2709 2710/* extents.c */ 2711struct ext4_ext_path; 2712struct ext4_extent; 2713 2714/* 2715 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 2716 * __le32. 2717 */ 2718#define EXT_MAX_BLOCKS 0xffffffff 2719 2720extern int ext4_ext_tree_init(handle_t *handle, struct inode *); 2721extern int ext4_ext_writepage_trans_blocks(struct inode *, int); 2722extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 2723extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 2724 struct ext4_map_blocks *map, int flags); 2725extern void ext4_ext_truncate(handle_t *, struct inode *); 2726extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 2727 ext4_lblk_t end); 2728extern void ext4_ext_init(struct super_block *); 2729extern void ext4_ext_release(struct super_block *); 2730extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 2731 loff_t len); 2732extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 2733 loff_t offset, ssize_t len); 2734extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 2735 struct ext4_map_blocks *map, int flags); 2736extern int ext4_ext_calc_metadata_amount(struct inode *inode, 2737 ext4_lblk_t lblocks); 2738extern int ext4_extent_tree_init(handle_t *, struct inode *); 2739extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 2740 int num, 2741 struct ext4_ext_path *path); 2742extern int ext4_can_extents_be_merged(struct inode *inode, 2743 struct ext4_extent *ex1, 2744 struct ext4_extent *ex2); 2745extern int ext4_ext_insert_extent(handle_t *, struct inode *, 2746 struct ext4_ext_path *, 2747 struct ext4_extent *, int); 2748extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t, 2749 struct ext4_ext_path *, 2750 int flags); 2751extern void ext4_ext_drop_refs(struct ext4_ext_path *); 2752extern int ext4_ext_check_inode(struct inode *inode); 2753extern int ext4_find_delalloc_range(struct inode *inode, 2754 ext4_lblk_t lblk_start, 2755 ext4_lblk_t lblk_end); 2756extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk); 2757extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 2758 __u64 start, __u64 len); 2759extern int ext4_ext_precache(struct inode *inode); 2760 2761/* move_extent.c */ 2762extern void ext4_double_down_write_data_sem(struct inode *first, 2763 struct inode *second); 2764extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 2765 struct inode *donor_inode); 2766extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 2767 __u64 start_orig, __u64 start_donor, 2768 __u64 len, __u64 *moved_len); 2769 2770/* page-io.c */ 2771extern int __init ext4_init_pageio(void); 2772extern void ext4_exit_pageio(void); 2773extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 2774extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 2775extern int ext4_put_io_end(ext4_io_end_t *io_end); 2776extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 2777extern void ext4_io_submit_init(struct ext4_io_submit *io, 2778 struct writeback_control *wbc); 2779extern void ext4_end_io_rsv_work(struct work_struct *work); 2780extern void ext4_io_submit(struct ext4_io_submit *io); 2781extern int ext4_bio_write_page(struct ext4_io_submit *io, 2782 struct page *page, 2783 int len, 2784 struct writeback_control *wbc); 2785 2786/* mmp.c */ 2787extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 2788extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp); 2789extern int ext4_mmp_csum_verify(struct super_block *sb, 2790 struct mmp_struct *mmp); 2791 2792/* 2793 * Note that these flags will never ever appear in a buffer_head's state flag. 2794 * See EXT4_MAP_... to see where this is used. 2795 */ 2796enum ext4_state_bits { 2797 BH_Uninit /* blocks are allocated but uninitialized on disk */ 2798 = BH_JBDPrivateStart, 2799 BH_AllocFromCluster, /* allocated blocks were part of already 2800 * allocated cluster. */ 2801}; 2802 2803/* 2804 * Add new method to test whether block and inode bitmaps are properly 2805 * initialized. With uninit_bg reading the block from disk is not enough 2806 * to mark the bitmap uptodate. We need to also zero-out the bitmap 2807 */ 2808#define BH_BITMAP_UPTODATE BH_JBDPrivateStart 2809 2810static inline int bitmap_uptodate(struct buffer_head *bh) 2811{ 2812 return (buffer_uptodate(bh) && 2813 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 2814} 2815static inline void set_bitmap_uptodate(struct buffer_head *bh) 2816{ 2817 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 2818} 2819 2820/* 2821 * Disable DIO read nolock optimization, so new dioreaders will be forced 2822 * to grab i_mutex 2823 */ 2824static inline void ext4_inode_block_unlocked_dio(struct inode *inode) 2825{ 2826 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2827 smp_mb(); 2828} 2829static inline void ext4_inode_resume_unlocked_dio(struct inode *inode) 2830{ 2831 smp_mb(); 2832 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK); 2833} 2834 2835#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1) 2836 2837/* For ioend & aio unwritten conversion wait queues */ 2838#define EXT4_WQ_HASH_SZ 37 2839#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\ 2840 EXT4_WQ_HASH_SZ]) 2841#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\ 2842 EXT4_WQ_HASH_SZ]) 2843extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ]; 2844extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ]; 2845 2846#define EXT4_RESIZING 0 2847extern int ext4_resize_begin(struct super_block *sb); 2848extern void ext4_resize_end(struct super_block *sb); 2849 2850#endif /* __KERNEL__ */ 2851 2852#endif /* _EXT4_H */