Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
at v6.18-rc2 3932 lines 139 kB view raw
1// SPDX-License-Identifier: GPL-2.0 2/* 3 * ext4.h 4 * 5 * Copyright (C) 1992, 1993, 1994, 1995 6 * Remy Card (card@masi.ibp.fr) 7 * Laboratoire MASI - Institut Blaise Pascal 8 * Universite Pierre et Marie Curie (Paris VI) 9 * 10 * from 11 * 12 * linux/include/linux/minix_fs.h 13 * 14 * Copyright (C) 1991, 1992 Linus Torvalds 15 */ 16 17#ifndef _EXT4_H 18#define _EXT4_H 19 20#include <linux/refcount.h> 21#include <linux/types.h> 22#include <linux/blkdev.h> 23#include <linux/magic.h> 24#include <linux/jbd2.h> 25#include <linux/quota.h> 26#include <linux/rwsem.h> 27#include <linux/rbtree.h> 28#include <linux/seqlock.h> 29#include <linux/mutex.h> 30#include <linux/timer.h> 31#include <linux/wait.h> 32#include <linux/sched/signal.h> 33#include <linux/blockgroup_lock.h> 34#include <linux/percpu_counter.h> 35#include <linux/ratelimit.h> 36#include <linux/crc32c.h> 37#include <linux/falloc.h> 38#include <linux/percpu-rwsem.h> 39#include <linux/fiemap.h> 40#ifdef __KERNEL__ 41#include <linux/compat.h> 42#endif 43#include <uapi/linux/ext4.h> 44 45#include <linux/fscrypt.h> 46#include <linux/fsverity.h> 47 48#include <linux/compiler.h> 49 50/* 51 * The fourth extended filesystem constants/structures 52 */ 53 54/* 55 * with AGGRESSIVE_CHECK allocator runs consistency checks over 56 * structures. these checks slow things down a lot 57 */ 58#define AGGRESSIVE_CHECK__ 59 60/* 61 * with DOUBLE_CHECK defined mballoc creates persistent in-core 62 * bitmaps, maintains and uses them to check for double allocations 63 */ 64#define DOUBLE_CHECK__ 65 66/* 67 * Define EXT4FS_DEBUG to produce debug messages 68 */ 69#undef EXT4FS_DEBUG 70 71/* 72 * Debug code 73 */ 74#ifdef EXT4FS_DEBUG 75#define ext4_debug(f, a...) \ 76 do { \ 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \ 78 __FILE__, __LINE__, __func__); \ 79 printk(KERN_DEBUG f, ## a); \ 80 } while (0) 81#else 82#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__) 83#endif 84 85 /* 86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c 87 */ 88#define EXT_DEBUG__ 89 90/* 91 * Dynamic printk for controlled extents debugging. 92 */ 93#ifdef CONFIG_EXT4_DEBUG 94#define ext_debug(ino, fmt, ...) \ 95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \ 96 current->comm, task_pid_nr(current), \ 97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \ 98 __func__, ##__VA_ARGS__) 99#else 100#define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__) 101#endif 102 103#define ASSERT(assert) \ 104do { \ 105 if (unlikely(!(assert))) { \ 106 printk(KERN_EMERG \ 107 "Assertion failure in %s() at %s:%d: '%s'\n", \ 108 __func__, __FILE__, __LINE__, #assert); \ 109 BUG(); \ 110 } \ 111} while (0) 112 113/* data type for block offset of block group */ 114typedef int ext4_grpblk_t; 115 116/* data type for filesystem-wide blocks number */ 117typedef unsigned long long ext4_fsblk_t; 118 119/* data type for file logical block number */ 120typedef __u32 ext4_lblk_t; 121 122/* data type for block group number */ 123typedef unsigned int ext4_group_t; 124 125enum SHIFT_DIRECTION { 126 SHIFT_LEFT = 0, 127 SHIFT_RIGHT, 128}; 129 130/* 131 * For each criteria, mballoc has slightly different way of finding 132 * the required blocks nad usually, higher the criteria the slower the 133 * allocation. We start at lower criterias and keep falling back to 134 * higher ones if we are not able to find any blocks. Lower (earlier) 135 * criteria are faster. 136 */ 137enum criteria { 138 /* 139 * Used when number of blocks needed is a power of 2. This 140 * doesn't trigger any disk IO except prefetch and is the 141 * fastest criteria. 142 */ 143 CR_POWER2_ALIGNED, 144 145 /* 146 * Tries to lookup in-memory data structures to find the most 147 * suitable group that satisfies goal request. No disk IO 148 * except block prefetch. 149 */ 150 CR_GOAL_LEN_FAST, 151 152 /* 153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal 154 * length to the best available length for faster allocation. 155 */ 156 CR_BEST_AVAIL_LEN, 157 158 /* 159 * Reads each block group sequentially, performing disk IO if 160 * necessary, to find suitable block group. Tries to 161 * allocate goal length but might trim the request if nothing 162 * is found after enough tries. 163 */ 164 CR_GOAL_LEN_SLOW, 165 166 /* 167 * Finds the first free set of blocks and allocates 168 * those. This is only used in rare cases when 169 * CR_GOAL_LEN_SLOW also fails to allocate anything. 170 */ 171 CR_ANY_FREE, 172 173 /* 174 * Number of criterias defined. 175 */ 176 EXT4_MB_NUM_CRS 177}; 178 179/* 180 * Flags used in mballoc's allocation_context flags field. 181 * 182 * Also used to show what's going on for debugging purposes when the 183 * flag field is exported via the traceport interface 184 */ 185 186/* prefer goal again. length */ 187#define EXT4_MB_HINT_MERGE 0x0001 188/* first blocks in the file */ 189#define EXT4_MB_HINT_FIRST 0x0008 190/* data is being allocated */ 191#define EXT4_MB_HINT_DATA 0x0020 192/* don't preallocate (for tails) */ 193#define EXT4_MB_HINT_NOPREALLOC 0x0040 194/* allocate for locality group */ 195#define EXT4_MB_HINT_GROUP_ALLOC 0x0080 196/* allocate goal blocks or none */ 197#define EXT4_MB_HINT_GOAL_ONLY 0x0100 198/* goal is meaningful */ 199#define EXT4_MB_HINT_TRY_GOAL 0x0200 200/* blocks already pre-reserved by delayed allocation */ 201#define EXT4_MB_DELALLOC_RESERVED 0x0400 202/* We are doing stream allocation */ 203#define EXT4_MB_STREAM_ALLOC 0x0800 204/* Use reserved root blocks if needed */ 205#define EXT4_MB_USE_ROOT_BLOCKS 0x1000 206/* Use blocks from reserved pool */ 207#define EXT4_MB_USE_RESERVED 0x2000 208/* Do strict check for free blocks while retrying block allocation */ 209#define EXT4_MB_STRICT_CHECK 0x4000 210 211struct ext4_allocation_request { 212 /* target inode for block we're allocating */ 213 struct inode *inode; 214 /* how many blocks we want to allocate */ 215 unsigned int len; 216 /* logical block in target inode */ 217 ext4_lblk_t logical; 218 /* the closest logical allocated block to the left */ 219 ext4_lblk_t lleft; 220 /* the closest logical allocated block to the right */ 221 ext4_lblk_t lright; 222 /* phys. target (a hint) */ 223 ext4_fsblk_t goal; 224 /* phys. block for the closest logical allocated block to the left */ 225 ext4_fsblk_t pleft; 226 /* phys. block for the closest logical allocated block to the right */ 227 ext4_fsblk_t pright; 228 /* flags. see above EXT4_MB_HINT_* */ 229 unsigned int flags; 230}; 231 232/* 233 * Logical to physical block mapping, used by ext4_map_blocks() 234 * 235 * This structure is used to pass requests into ext4_map_blocks() as 236 * well as to store the information returned by ext4_map_blocks(). It 237 * takes less room on the stack than a struct buffer_head. 238 */ 239#define EXT4_MAP_NEW BIT(BH_New) 240#define EXT4_MAP_MAPPED BIT(BH_Mapped) 241#define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten) 242#define EXT4_MAP_BOUNDARY BIT(BH_Boundary) 243#define EXT4_MAP_DELAYED BIT(BH_Delay) 244/* 245 * This is for use in ext4_map_query_blocks() for a special case where we can 246 * have a physically and logically contiguous blocks split across two leaf 247 * nodes instead of a single extent. This is required in case of atomic writes 248 * to know whether the returned extent is last in leaf. If yes, then lookup for 249 * next in leaf block in ext4_map_query_blocks_next_in_leaf(). 250 * - This is never going to be added to any buffer head state. 251 * - We use the next available bit after BH_BITMAP_UPTODATE. 252 */ 253#define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1) 254#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ 255 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ 256 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF) 257 258struct ext4_map_blocks { 259 ext4_fsblk_t m_pblk; 260 ext4_lblk_t m_lblk; 261 unsigned int m_len; 262 unsigned int m_flags; 263}; 264 265/* 266 * Block validity checking, system zone rbtree. 267 */ 268struct ext4_system_blocks { 269 struct rb_root root; 270 struct rcu_head rcu; 271}; 272 273/* 274 * Flags for ext4_io_end->flags 275 */ 276#define EXT4_IO_END_UNWRITTEN 0x0001 277#define EXT4_IO_END_FAILED 0x0002 278 279#define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED) 280 281struct ext4_io_end_vec { 282 struct list_head list; /* list of io_end_vec */ 283 loff_t offset; /* offset in the file */ 284 ssize_t size; /* size of the extent */ 285}; 286 287/* 288 * For converting unwritten extents on a work queue. 'handle' is used for 289 * buffered writeback. 290 */ 291typedef struct ext4_io_end { 292 struct list_head list; /* per-file finished IO list */ 293 handle_t *handle; /* handle reserved for extent 294 * conversion */ 295 struct inode *inode; /* file being written to */ 296 struct bio *bio; /* Linked list of completed 297 * bios covering the extent */ 298 unsigned int flag; /* unwritten or not */ 299 refcount_t count; /* reference counter */ 300 struct list_head list_vec; /* list of ext4_io_end_vec */ 301} ext4_io_end_t; 302 303struct ext4_io_submit { 304 struct writeback_control *io_wbc; 305 struct bio *io_bio; 306 ext4_io_end_t *io_end; 307 sector_t io_next_block; 308}; 309 310/* 311 * Special inodes numbers 312 */ 313#define EXT4_BAD_INO 1 /* Bad blocks inode */ 314#define EXT4_ROOT_INO 2 /* Root inode */ 315#define EXT4_USR_QUOTA_INO 3 /* User quota inode */ 316#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */ 317#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */ 318#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */ 319#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */ 320#define EXT4_JOURNAL_INO 8 /* Journal inode */ 321 322/* First non-reserved inode for old ext4 filesystems */ 323#define EXT4_GOOD_OLD_FIRST_INO 11 324 325/* 326 * Maximal count of links to a file 327 */ 328#define EXT4_LINK_MAX 65000 329 330/* 331 * Macro-instructions used to manage several block sizes 332 */ 333#define EXT4_MIN_BLOCK_SIZE 1024 334#define EXT4_MAX_BLOCK_SIZE 65536 335#define EXT4_MIN_BLOCK_LOG_SIZE 10 336#define EXT4_MAX_BLOCK_LOG_SIZE 16 337#define EXT4_MAX_CLUSTER_LOG_SIZE 30 338#ifdef __KERNEL__ 339# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize) 340#else 341# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size) 342#endif 343#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32)) 344#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \ 345 EXT4_SB(s)->s_cluster_bits) 346#ifdef __KERNEL__ 347# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits) 348# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits) 349#else 350# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10) 351#endif 352#ifdef __KERNEL__ 353#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits) 354#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size) 355#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino) 356#else 357#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 358 EXT4_GOOD_OLD_INODE_SIZE : \ 359 (s)->s_inode_size) 360#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \ 361 EXT4_GOOD_OLD_FIRST_INO : \ 362 (s)->s_first_ino) 363#endif 364#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits))) 365#define EXT4_MAX_BLOCKS(size, offset, blkbits) \ 366 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \ 367 blkbits)) 368#define EXT4_B_TO_LBLK(inode, offset) \ 369 (round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits) 370 371/* Translate a block number to a cluster number */ 372#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits) 373/* Translate a cluster number to a block number */ 374#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits) 375/* Translate # of blks to # of clusters */ 376#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \ 377 (sbi)->s_cluster_bits) 378/* Mask out the low bits to get the starting block of the cluster */ 379#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \ 380 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 381#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \ 382 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 383/* Fill in the low bits to get the last block of the cluster */ 384#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \ 385 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1)) 386/* Get the cluster offset */ 387#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \ 388 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1)) 389#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \ 390 ((ext4_lblk_t) (s)->s_cluster_ratio - 1)) 391 392/* 393 * Structure of a blocks group descriptor 394 */ 395struct ext4_group_desc 396{ 397 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */ 398 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */ 399 __le32 bg_inode_table_lo; /* Inodes table block */ 400 __le16 bg_free_blocks_count_lo;/* Free blocks count */ 401 __le16 bg_free_inodes_count_lo;/* Free inodes count */ 402 __le16 bg_used_dirs_count_lo; /* Directories count */ 403 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */ 404 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */ 405 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */ 406 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */ 407 __le16 bg_itable_unused_lo; /* Unused inodes count */ 408 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */ 409 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */ 410 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */ 411 __le32 bg_inode_table_hi; /* Inodes table block MSB */ 412 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */ 413 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */ 414 __le16 bg_used_dirs_count_hi; /* Directories count MSB */ 415 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */ 416 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */ 417 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */ 418 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */ 419 __u32 bg_reserved; 420}; 421 422#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \ 423 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \ 424 sizeof(__le16)) 425#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \ 426 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \ 427 sizeof(__le16)) 428 429/* 430 * Structure of a flex block group info 431 */ 432 433struct flex_groups { 434 atomic64_t free_clusters; 435 atomic_t free_inodes; 436 atomic_t used_dirs; 437}; 438 439#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */ 440#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */ 441#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */ 442 443/* 444 * Macro-instructions used to manage group descriptors 445 */ 446#define EXT4_MIN_DESC_SIZE 32 447#define EXT4_MIN_DESC_SIZE_64BIT 64 448#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE 449#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size) 450#ifdef __KERNEL__ 451# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group) 452# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group) 453# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block) 454# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group) 455# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits) 456#else 457# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group) 458# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s)) 459# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group) 460#endif 461 462/* 463 * Constants relative to the data blocks 464 */ 465#define EXT4_NDIR_BLOCKS 12 466#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS 467#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1) 468#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1) 469#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1) 470 471/* 472 * Inode flags 473 */ 474#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */ 475#define EXT4_UNRM_FL 0x00000002 /* Undelete */ 476#define EXT4_COMPR_FL 0x00000004 /* Compress file */ 477#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */ 478#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */ 479#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */ 480#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */ 481#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */ 482/* Reserved for compression usage... */ 483#define EXT4_DIRTY_FL 0x00000100 484#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 485#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */ 486 /* nb: was previously EXT2_ECOMPR_FL */ 487#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */ 488/* End compression flags --- maybe not all used */ 489#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */ 490#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */ 491#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 492#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 493#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 494#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 495#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 496#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */ 497#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */ 498#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 499/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */ 500 501#define EXT4_DAX_FL 0x02000000 /* Inode is DAX */ 502 503#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 504#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 505#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */ 506#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 507 508/* User modifiable flags */ 509#define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \ 510 EXT4_UNRM_FL | \ 511 EXT4_COMPR_FL | \ 512 EXT4_SYNC_FL | \ 513 EXT4_IMMUTABLE_FL | \ 514 EXT4_APPEND_FL | \ 515 EXT4_NODUMP_FL | \ 516 EXT4_NOATIME_FL | \ 517 EXT4_JOURNAL_DATA_FL | \ 518 EXT4_NOTAIL_FL | \ 519 EXT4_DIRSYNC_FL | \ 520 EXT4_TOPDIR_FL | \ 521 EXT4_EXTENTS_FL | \ 522 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \ 523 EXT4_DAX_FL | \ 524 EXT4_PROJINHERIT_FL | \ 525 EXT4_CASEFOLD_FL) 526 527/* User visible flags */ 528#define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \ 529 EXT4_DIRTY_FL | \ 530 EXT4_COMPRBLK_FL | \ 531 EXT4_NOCOMPR_FL | \ 532 EXT4_ENCRYPT_FL | \ 533 EXT4_INDEX_FL | \ 534 EXT4_VERITY_FL | \ 535 EXT4_INLINE_DATA_FL) 536 537/* Flags that should be inherited by new inodes from their parent. */ 538#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\ 539 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\ 540 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\ 541 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\ 542 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\ 543 EXT4_DAX_FL) 544 545/* Flags that are appropriate for regular files (all but dir-specific ones). */ 546#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\ 547 EXT4_PROJINHERIT_FL)) 548 549/* Flags that are appropriate for non-directories/regular files. */ 550#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL) 551 552/* The only flags that should be swapped */ 553#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL) 554 555/* Flags which are mutually exclusive to DAX */ 556#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\ 557 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL) 558 559/* Mask out flags that are inappropriate for the given type of inode. */ 560static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags) 561{ 562 if (S_ISDIR(mode)) 563 return flags; 564 else if (S_ISREG(mode)) 565 return flags & EXT4_REG_FLMASK; 566 else 567 return flags & EXT4_OTHER_FLMASK; 568} 569 570/* 571 * Inode flags used for atomic set/get 572 */ 573enum { 574 EXT4_INODE_SECRM = 0, /* Secure deletion */ 575 EXT4_INODE_UNRM = 1, /* Undelete */ 576 EXT4_INODE_COMPR = 2, /* Compress file */ 577 EXT4_INODE_SYNC = 3, /* Synchronous updates */ 578 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */ 579 EXT4_INODE_APPEND = 5, /* writes to file may only append */ 580 EXT4_INODE_NODUMP = 6, /* do not dump file */ 581 EXT4_INODE_NOATIME = 7, /* do not update atime */ 582/* Reserved for compression usage... */ 583 EXT4_INODE_DIRTY = 8, 584 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */ 585 EXT4_INODE_NOCOMPR = 10, /* Don't compress */ 586 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */ 587/* End compression flags --- maybe not all used */ 588 EXT4_INODE_INDEX = 12, /* hash-indexed directory */ 589 EXT4_INODE_IMAGIC = 13, /* AFS directory */ 590 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */ 591 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */ 592 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */ 593 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/ 594 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */ 595 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */ 596 EXT4_INODE_VERITY = 20, /* Verity protected inode */ 597 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */ 598/* 22 was formerly EXT4_INODE_EOFBLOCKS */ 599 EXT4_INODE_DAX = 25, /* Inode is DAX */ 600 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */ 601 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */ 602 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */ 603 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */ 604}; 605 606/* 607 * Since it's pretty easy to mix up bit numbers and hex values, we use a 608 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to 609 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost 610 * any extra space in the compiled kernel image, otherwise, the build will fail. 611 * It's important that these values are the same, since we are using 612 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent 613 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk 614 * values found in ext2, ext3 and ext4 filesystems, and of course the values 615 * defined in e2fsprogs. 616 * 617 * It's not paranoia if the Murphy's Law really *is* out to get you. :-) 618 */ 619#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG)) 620#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG)) 621 622static inline void ext4_check_flag_values(void) 623{ 624 CHECK_FLAG_VALUE(SECRM); 625 CHECK_FLAG_VALUE(UNRM); 626 CHECK_FLAG_VALUE(COMPR); 627 CHECK_FLAG_VALUE(SYNC); 628 CHECK_FLAG_VALUE(IMMUTABLE); 629 CHECK_FLAG_VALUE(APPEND); 630 CHECK_FLAG_VALUE(NODUMP); 631 CHECK_FLAG_VALUE(NOATIME); 632 CHECK_FLAG_VALUE(DIRTY); 633 CHECK_FLAG_VALUE(COMPRBLK); 634 CHECK_FLAG_VALUE(NOCOMPR); 635 CHECK_FLAG_VALUE(ENCRYPT); 636 CHECK_FLAG_VALUE(INDEX); 637 CHECK_FLAG_VALUE(IMAGIC); 638 CHECK_FLAG_VALUE(JOURNAL_DATA); 639 CHECK_FLAG_VALUE(NOTAIL); 640 CHECK_FLAG_VALUE(DIRSYNC); 641 CHECK_FLAG_VALUE(TOPDIR); 642 CHECK_FLAG_VALUE(HUGE_FILE); 643 CHECK_FLAG_VALUE(EXTENTS); 644 CHECK_FLAG_VALUE(VERITY); 645 CHECK_FLAG_VALUE(EA_INODE); 646 CHECK_FLAG_VALUE(INLINE_DATA); 647 CHECK_FLAG_VALUE(PROJINHERIT); 648 CHECK_FLAG_VALUE(CASEFOLD); 649 CHECK_FLAG_VALUE(RESERVED); 650} 651 652#if defined(__KERNEL__) && defined(CONFIG_COMPAT) 653struct compat_ext4_new_group_input { 654 u32 group; 655 compat_u64 block_bitmap; 656 compat_u64 inode_bitmap; 657 compat_u64 inode_table; 658 u32 blocks_count; 659 u16 reserved_blocks; 660 u16 unused; 661}; 662#endif 663 664/* The struct ext4_new_group_input in kernel space, with free_blocks_count */ 665struct ext4_new_group_data { 666 __u32 group; 667 __u64 block_bitmap; 668 __u64 inode_bitmap; 669 __u64 inode_table; 670 __u32 blocks_count; 671 __u16 reserved_blocks; 672 __u16 mdata_blocks; 673 __u32 free_clusters_count; 674}; 675 676/* Indexes used to index group tables in ext4_new_group_data */ 677enum { 678 BLOCK_BITMAP = 0, /* block bitmap */ 679 INODE_BITMAP, /* inode bitmap */ 680 INODE_TABLE, /* inode tables */ 681 GROUP_TABLE_COUNT, 682}; 683 684/* 685 * Flags used by ext4_map_blocks() 686 */ 687 /* Allocate any needed blocks and/or convert an unwritten 688 extent to be an initialized ext4 */ 689#define EXT4_GET_BLOCKS_CREATE 0x0001 690 /* Request the creation of an unwritten extent */ 691#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002 692#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\ 693 EXT4_GET_BLOCKS_CREATE) 694 /* Caller is from the delayed allocation writeout path 695 * finally doing the actual allocation of delayed blocks */ 696#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 697 /* caller is from the direct IO path, request to creation of an 698 unwritten extents if not allocated, split the unwritten 699 extent if blocks has been preallocated already*/ 700#define EXT4_GET_BLOCKS_PRE_IO 0x0008 701#define EXT4_GET_BLOCKS_CONVERT 0x0010 702#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\ 703 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT) 704 /* Eventual metadata allocation (due to growing extent tree) 705 * should not fail, so try to use reserved blocks for that.*/ 706#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020 707 /* Don't normalize allocation size (used for fallocate) */ 708#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 709 /* Convert written extents to unwritten */ 710#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100 711 /* Write zeros to newly created written extents */ 712#define EXT4_GET_BLOCKS_ZERO 0x0200 713#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\ 714 EXT4_GET_BLOCKS_ZERO) 715 /* Caller is in the context of data submission, such as writeback, 716 * fsync, etc. Especially, in the generic writeback path, caller will 717 * submit data before dropping transaction handle. This allows jbd2 718 * to avoid submitting data before commit. */ 719#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400 720 /* Convert extent to initialized after IO complete */ 721#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\ 722 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\ 723 EXT4_GET_BLOCKS_IO_SUBMIT) 724 /* Caller is in the atomic contex, find extent if it has been cached */ 725#define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800 726/* 727 * Atomic write caller needs this to query in the slow path of mixed mapping 728 * case, when a contiguous extent can be split across two adjacent leaf nodes. 729 * Look EXT4_MAP_QUERY_LAST_IN_LEAF. 730 */ 731#define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000 732 733/* 734 * The bit position of these flags must not overlap with any of the 735 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(), 736 * read_extent_tree_block(), ext4_split_extent_at(), 737 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf(). 738 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be 739 * caching the extents when reading from the extent tree while a 740 * truncate or punch hole operation is in progress. 741 */ 742#define EXT4_EX_NOCACHE 0x40000000 743#define EXT4_EX_FORCE_CACHE 0x20000000 744#define EXT4_EX_NOFAIL 0x10000000 745/* 746 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to 747 * pass while lookup/querying of on disk extent tree. 748 */ 749#define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\ 750 EXT4_EX_NOFAIL |\ 751 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF) 752 753/* 754 * Flags used by ext4_free_blocks 755 */ 756#define EXT4_FREE_BLOCKS_METADATA 0x0001 757#define EXT4_FREE_BLOCKS_FORGET 0x0002 758#define EXT4_FREE_BLOCKS_VALIDATED 0x0004 759#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 760#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010 761#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020 762#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040 763 764#if defined(__KERNEL__) && defined(CONFIG_COMPAT) 765/* 766 * ioctl commands in 32 bit emulation 767 */ 768#define EXT4_IOC32_GETVERSION _IOR('f', 3, int) 769#define EXT4_IOC32_SETVERSION _IOW('f', 4, int) 770#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int) 771#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int) 772#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int) 773#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input) 774#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION 775#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION 776#endif 777 778/* Max physical block we can address w/o extents */ 779#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF 780 781/* Max logical block we can support */ 782#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE 783 784/* 785 * Structure of an inode on the disk 786 */ 787struct ext4_inode { 788 __le16 i_mode; /* File mode */ 789 __le16 i_uid; /* Low 16 bits of Owner Uid */ 790 __le32 i_size_lo; /* Size in bytes */ 791 __le32 i_atime; /* Access time */ 792 __le32 i_ctime; /* Inode Change time */ 793 __le32 i_mtime; /* Modification time */ 794 __le32 i_dtime; /* Deletion Time */ 795 __le16 i_gid; /* Low 16 bits of Group Id */ 796 __le16 i_links_count; /* Links count */ 797 __le32 i_blocks_lo; /* Blocks count */ 798 __le32 i_flags; /* File flags */ 799 union { 800 struct { 801 __le32 l_i_version; 802 } linux1; 803 struct { 804 __u32 h_i_translator; 805 } hurd1; 806 struct { 807 __u32 m_i_reserved1; 808 } masix1; 809 } osd1; /* OS dependent 1 */ 810 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */ 811 __le32 i_generation; /* File version (for NFS) */ 812 __le32 i_file_acl_lo; /* File ACL */ 813 __le32 i_size_high; 814 __le32 i_obso_faddr; /* Obsoleted fragment address */ 815 union { 816 struct { 817 __le16 l_i_blocks_high; /* were l_i_reserved1 */ 818 __le16 l_i_file_acl_high; 819 __le16 l_i_uid_high; /* these 2 fields */ 820 __le16 l_i_gid_high; /* were reserved2[0] */ 821 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */ 822 __le16 l_i_reserved; 823 } linux2; 824 struct { 825 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 826 __u16 h_i_mode_high; 827 __u16 h_i_uid_high; 828 __u16 h_i_gid_high; 829 __u32 h_i_author; 830 } hurd2; 831 struct { 832 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */ 833 __le16 m_i_file_acl_high; 834 __u32 m_i_reserved2[2]; 835 } masix2; 836 } osd2; /* OS dependent 2 */ 837 __le16 i_extra_isize; 838 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */ 839 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */ 840 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */ 841 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */ 842 __le32 i_crtime; /* File Creation time */ 843 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */ 844 __le32 i_version_hi; /* high 32 bits for 64-bit version */ 845 __le32 i_projid; /* Project ID */ 846}; 847 848#define EXT4_EPOCH_BITS 2 849#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1) 850#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS) 851 852/* 853 * Extended fields will fit into an inode if the filesystem was formatted 854 * with large inodes (-I 256 or larger) and there are not currently any EAs 855 * consuming all of the available space. For new inodes we always reserve 856 * enough space for the kernel's known extended fields, but for inodes 857 * created with an old kernel this might not have been the case. None of 858 * the extended inode fields is critical for correct filesystem operation. 859 * This macro checks if a certain field fits in the inode. Note that 860 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize 861 */ 862#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \ 863 ((offsetof(typeof(*ext4_inode), field) + \ 864 sizeof((ext4_inode)->field)) \ 865 <= (EXT4_GOOD_OLD_INODE_SIZE + \ 866 (einode)->i_extra_isize)) \ 867 868/* 869 * We use an encoding that preserves the times for extra epoch "00": 870 * 871 * extra msb of adjust for signed 872 * epoch 32-bit 32-bit tv_sec to 873 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range 874 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31 875 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19 876 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07 877 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25 878 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16 879 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04 880 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22 881 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10 882 * 883 * Note that previous versions of the kernel on 64-bit systems would 884 * incorrectly use extra epoch bits 1,1 for dates between 1901 and 885 * 1970. e2fsck will correct this, assuming that it is run on the 886 * affected filesystem before 2242. 887 */ 888 889static inline __le32 ext4_encode_extra_time(struct timespec64 ts) 890{ 891 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK; 892 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS)); 893} 894 895static inline struct timespec64 ext4_decode_extra_time(__le32 base, 896 __le32 extra) 897{ 898 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) }; 899 900 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) 901 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32; 902 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS; 903 return ts; 904} 905 906#define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \ 907do { \ 908 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \ 909 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \ 910 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \ 911 } else \ 912 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \ 913} while (0) 914 915#define EXT4_INODE_SET_ATIME(inode, raw_inode) \ 916 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode)) 917 918#define EXT4_INODE_SET_MTIME(inode, raw_inode) \ 919 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode)) 920 921#define EXT4_INODE_SET_CTIME(inode, raw_inode) \ 922 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode)) 923 924#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \ 925 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 926 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \ 927 raw_inode, (einode)->xtime) 928 929#define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \ 930 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \ 931 ext4_decode_extra_time((raw_inode)->xtime, \ 932 (raw_inode)->xtime ## _extra) : \ 933 (struct timespec64) { \ 934 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \ 935 }) 936 937#define EXT4_INODE_GET_ATIME(inode, raw_inode) \ 938do { \ 939 inode_set_atime_to_ts(inode, \ 940 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \ 941} while (0) 942 943#define EXT4_INODE_GET_MTIME(inode, raw_inode) \ 944do { \ 945 inode_set_mtime_to_ts(inode, \ 946 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \ 947} while (0) 948 949#define EXT4_INODE_GET_CTIME(inode, raw_inode) \ 950do { \ 951 inode_set_ctime_to_ts(inode, \ 952 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \ 953} while (0) 954 955#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \ 956do { \ 957 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \ 958 (einode)->xtime = \ 959 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \ 960 raw_inode); \ 961 else \ 962 (einode)->xtime = (struct timespec64){0, 0}; \ 963} while (0) 964 965#define i_disk_version osd1.linux1.l_i_version 966 967#if defined(__KERNEL__) || defined(__linux__) 968#define i_reserved1 osd1.linux1.l_i_reserved1 969#define i_file_acl_high osd2.linux2.l_i_file_acl_high 970#define i_blocks_high osd2.linux2.l_i_blocks_high 971#define i_uid_low i_uid 972#define i_gid_low i_gid 973#define i_uid_high osd2.linux2.l_i_uid_high 974#define i_gid_high osd2.linux2.l_i_gid_high 975#define i_checksum_lo osd2.linux2.l_i_checksum_lo 976 977#elif defined(__GNU__) 978 979#define i_translator osd1.hurd1.h_i_translator 980#define i_uid_high osd2.hurd2.h_i_uid_high 981#define i_gid_high osd2.hurd2.h_i_gid_high 982#define i_author osd2.hurd2.h_i_author 983 984#elif defined(__masix__) 985 986#define i_reserved1 osd1.masix1.m_i_reserved1 987#define i_file_acl_high osd2.masix2.m_i_file_acl_high 988#define i_reserved2 osd2.masix2.m_i_reserved2 989 990#endif /* defined(__KERNEL__) || defined(__linux__) */ 991 992#include "extents_status.h" 993#include "fast_commit.h" 994 995/* 996 * Lock subclasses for i_data_sem in the ext4_inode_info structure. 997 * 998 * These are needed to avoid lockdep false positives when we need to 999 * allocate blocks to the quota inode during ext4_map_blocks(), while 1000 * holding i_data_sem for a normal (non-quota) inode. Since we don't 1001 * do quota tracking for the quota inode, this avoids deadlock (as 1002 * well as infinite recursion, since it isn't turtles all the way 1003 * down...) 1004 * 1005 * I_DATA_SEM_NORMAL - Used for most inodes 1006 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode 1007 * where the second inode has larger inode number 1008 * than the first 1009 * I_DATA_SEM_QUOTA - Used for quota inodes only 1010 * I_DATA_SEM_EA - Used for ea_inodes only 1011 */ 1012enum { 1013 I_DATA_SEM_NORMAL = 0, 1014 I_DATA_SEM_OTHER, 1015 I_DATA_SEM_QUOTA, 1016 I_DATA_SEM_EA 1017}; 1018 1019 1020/* 1021 * fourth extended file system inode data in memory 1022 */ 1023struct ext4_inode_info { 1024 __le32 i_data[15]; /* unconverted */ 1025 __u32 i_dtime; 1026 ext4_fsblk_t i_file_acl; 1027 1028 /* 1029 * i_block_group is the number of the block group which contains 1030 * this file's inode. Constant across the lifetime of the inode, 1031 * it is used for making block allocation decisions - we try to 1032 * place a file's data blocks near its inode block, and new inodes 1033 * near to their parent directory's inode. 1034 */ 1035 ext4_group_t i_block_group; 1036 ext4_lblk_t i_dir_start_lookup; 1037#if (BITS_PER_LONG < 64) 1038 unsigned long i_state_flags; /* Dynamic state flags */ 1039#endif 1040 unsigned long i_flags; 1041 1042 /* 1043 * Extended attributes can be read independently of the main file 1044 * data. Taking i_rwsem even when reading would cause contention 1045 * between readers of EAs and writers of regular file data, so 1046 * instead we synchronize on xattr_sem when reading or changing 1047 * EAs. 1048 */ 1049 struct rw_semaphore xattr_sem; 1050 1051 /* 1052 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise 1053 * i_orphan is used. 1054 */ 1055 union { 1056 struct list_head i_orphan; /* unlinked but open inodes */ 1057 unsigned int i_orphan_idx; /* Index in orphan file */ 1058 }; 1059 1060 /* Fast commit related info */ 1061 1062 /* For tracking dentry create updates */ 1063 struct list_head i_fc_dilist; 1064 struct list_head i_fc_list; /* 1065 * inodes that need fast commit 1066 * protected by sbi->s_fc_lock. 1067 */ 1068 1069 /* Start of lblk range that needs to be committed in this fast commit */ 1070 ext4_lblk_t i_fc_lblk_start; 1071 1072 /* End of lblk range that needs to be committed in this fast commit */ 1073 ext4_lblk_t i_fc_lblk_len; 1074 1075 spinlock_t i_raw_lock; /* protects updates to the raw inode */ 1076 1077 /* Fast commit wait queue for this inode */ 1078 wait_queue_head_t i_fc_wait; 1079 1080 /* 1081 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len 1082 * and inode's EXT4_FC_STATE_COMMITTING state bit. 1083 */ 1084 spinlock_t i_fc_lock; 1085 1086 /* 1087 * i_disksize keeps track of what the inode size is ON DISK, not 1088 * in memory. During truncate, i_size is set to the new size by 1089 * the VFS prior to calling ext4_truncate(), but the filesystem won't 1090 * set i_disksize to 0 until the truncate is actually under way. 1091 * 1092 * The intent is that i_disksize always represents the blocks which 1093 * are used by this file. This allows recovery to restart truncate 1094 * on orphans if we crash during truncate. We actually write i_disksize 1095 * into the on-disk inode when writing inodes out, instead of i_size. 1096 * 1097 * The only time when i_disksize and i_size may be different is when 1098 * a truncate is in progress. The only things which change i_disksize 1099 * are ext4_get_block (growth) and ext4_truncate (shrinkth). 1100 */ 1101 loff_t i_disksize; 1102 1103 /* 1104 * i_data_sem is for serialising ext4_truncate() against 1105 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's 1106 * data tree are chopped off during truncate. We can't do that in 1107 * ext4 because whenever we perform intermediate commits during 1108 * truncate, the inode and all the metadata blocks *must* be in a 1109 * consistent state which allows truncation of the orphans to restart 1110 * during recovery. Hence we must fix the get_block-vs-truncate race 1111 * by other means, so we have i_data_sem. 1112 */ 1113 struct rw_semaphore i_data_sem; 1114 struct inode vfs_inode; 1115 struct jbd2_inode *jinode; 1116 1117 /* 1118 * File creation time. Its function is same as that of 1119 * struct timespec64 i_{a,c,m}time in the generic inode. 1120 */ 1121 struct timespec64 i_crtime; 1122 1123 /* mballoc */ 1124 atomic_t i_prealloc_active; 1125 1126 /* allocation reservation info for delalloc */ 1127 /* In case of bigalloc, this refer to clusters rather than blocks */ 1128 unsigned int i_reserved_data_blocks; 1129 struct rb_root i_prealloc_node; 1130 rwlock_t i_prealloc_lock; 1131 1132 /* extents status tree */ 1133 struct ext4_es_tree i_es_tree; 1134 rwlock_t i_es_lock; 1135 struct list_head i_es_list; 1136 unsigned int i_es_all_nr; /* protected by i_es_lock */ 1137 unsigned int i_es_shk_nr; /* protected by i_es_lock */ 1138 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for 1139 extents to shrink. Protected by 1140 i_es_lock */ 1141 1142 /* ialloc */ 1143 ext4_group_t i_last_alloc_group; 1144 1145 /* pending cluster reservations for bigalloc file systems */ 1146 struct ext4_pending_tree i_pending_tree; 1147 1148 /* on-disk additional length */ 1149 __u16 i_extra_isize; 1150 1151 /* Indicate the inline data space. */ 1152 u16 i_inline_off; 1153 u16 i_inline_size; 1154 1155#ifdef CONFIG_QUOTA 1156 /* quota space reservation, managed internally by quota code */ 1157 qsize_t i_reserved_quota; 1158#endif 1159 spinlock_t i_block_reservation_lock; 1160 1161 /* Lock protecting lists below */ 1162 spinlock_t i_completed_io_lock; 1163 /* 1164 * Completed IOs that need unwritten extents handling and have 1165 * transaction reserved 1166 */ 1167 struct list_head i_rsv_conversion_list; 1168 struct work_struct i_rsv_conversion_work; 1169 1170 /* 1171 * Transactions that contain inode's metadata needed to complete 1172 * fsync and fdatasync, respectively. 1173 */ 1174 tid_t i_sync_tid; 1175 tid_t i_datasync_tid; 1176 1177#ifdef CONFIG_QUOTA 1178 struct dquot __rcu *i_dquot[MAXQUOTAS]; 1179#endif 1180 1181 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */ 1182 __u32 i_csum_seed; 1183 1184 kprojid_t i_projid; 1185 1186#ifdef CONFIG_FS_ENCRYPTION 1187 struct fscrypt_inode_info *i_crypt_info; 1188#endif 1189 1190#ifdef CONFIG_FS_VERITY 1191 struct fsverity_info *i_verity_info; 1192#endif 1193}; 1194 1195/* 1196 * File system states 1197 */ 1198#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */ 1199#define EXT4_ERROR_FS 0x0002 /* Errors detected */ 1200#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */ 1201#define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */ 1202 1203/* 1204 * Misc. filesystem flags 1205 */ 1206#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */ 1207#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */ 1208#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */ 1209 1210/* 1211 * Mount flags set via mount options or defaults 1212 */ 1213#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */ 1214#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */ 1215#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */ 1216#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */ 1217#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */ 1218#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */ 1219#define EXT4_MOUNT_ERRORS_MASK 0x00070 1220#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */ 1221#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/ 1222#ifdef CONFIG_FS_DAX 1223#define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */ 1224#else 1225#define EXT4_MOUNT_DAX_ALWAYS 0 1226#endif 1227#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */ 1228#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */ 1229#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */ 1230#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */ 1231#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */ 1232#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */ 1233#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */ 1234#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */ 1235#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */ 1236#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */ 1237#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */ 1238#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota, 1239 * enable enforcement for hidden 1240 * quota files */ 1241#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable 1242 * enforcement for hidden quota 1243 * files */ 1244#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota 1245 * enforcement */ 1246#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */ 1247#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */ 1248#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */ 1249#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */ 1250#define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000 1251#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */ 1252#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */ 1253#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */ 1254#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */ 1255#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */ 1256 1257/* 1258 * Mount flags set either automatically (could not be set by mount option) 1259 * based on per file system feature or property or in special cases such as 1260 * distinguishing between explicit mount option definition and default. 1261 */ 1262#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly 1263 specified delalloc */ 1264#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group 1265 size of blocksize * 8 1266 blocks */ 1267#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated 1268 file systems */ 1269#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly 1270 specified journal checksum */ 1271 1272#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */ 1273#define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */ 1274#define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */ 1275#define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group 1276 * scanning in mballoc 1277 */ 1278#define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */ 1279 1280#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \ 1281 ~EXT4_MOUNT_##opt 1282#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \ 1283 EXT4_MOUNT_##opt 1284#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \ 1285 EXT4_MOUNT_##opt) 1286 1287#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \ 1288 ~EXT4_MOUNT2_##opt 1289#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \ 1290 EXT4_MOUNT2_##opt 1291#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \ 1292 EXT4_MOUNT2_##opt) 1293 1294#define ext4_test_and_set_bit __test_and_set_bit_le 1295#define ext4_set_bit __set_bit_le 1296#define ext4_test_and_clear_bit __test_and_clear_bit_le 1297#define ext4_clear_bit __clear_bit_le 1298#define ext4_test_bit test_bit_le 1299#define ext4_find_next_zero_bit find_next_zero_bit_le 1300#define ext4_find_next_bit find_next_bit_le 1301 1302extern void mb_set_bits(void *bm, int cur, int len); 1303 1304/* 1305 * Maximal mount counts between two filesystem checks 1306 */ 1307#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */ 1308#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */ 1309 1310/* 1311 * Behaviour when detecting errors 1312 */ 1313#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */ 1314#define EXT4_ERRORS_RO 2 /* Remount fs read-only */ 1315#define EXT4_ERRORS_PANIC 3 /* Panic */ 1316#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE 1317 1318/* Metadata checksum algorithm codes */ 1319#define EXT4_CRC32C_CHKSUM 1 1320 1321#define EXT4_LABEL_MAX 16 1322 1323/* 1324 * Structure of the super block 1325 */ 1326struct ext4_super_block { 1327/*00*/ __le32 s_inodes_count; /* Inodes count */ 1328 __le32 s_blocks_count_lo; /* Blocks count */ 1329 __le32 s_r_blocks_count_lo; /* Reserved blocks count */ 1330 __le32 s_free_blocks_count_lo; /* Free blocks count */ 1331/*10*/ __le32 s_free_inodes_count; /* Free inodes count */ 1332 __le32 s_first_data_block; /* First Data Block */ 1333 __le32 s_log_block_size; /* Block size */ 1334 __le32 s_log_cluster_size; /* Allocation cluster size */ 1335/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */ 1336 __le32 s_clusters_per_group; /* # Clusters per group */ 1337 __le32 s_inodes_per_group; /* # Inodes per group */ 1338 __le32 s_mtime; /* Mount time */ 1339/*30*/ __le32 s_wtime; /* Write time */ 1340 __le16 s_mnt_count; /* Mount count */ 1341 __le16 s_max_mnt_count; /* Maximal mount count */ 1342 __le16 s_magic; /* Magic signature */ 1343 __le16 s_state; /* File system state */ 1344 __le16 s_errors; /* Behaviour when detecting errors */ 1345 __le16 s_minor_rev_level; /* minor revision level */ 1346/*40*/ __le32 s_lastcheck; /* time of last check */ 1347 __le32 s_checkinterval; /* max. time between checks */ 1348 __le32 s_creator_os; /* OS */ 1349 __le32 s_rev_level; /* Revision level */ 1350/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */ 1351 __le16 s_def_resgid; /* Default gid for reserved blocks */ 1352 /* 1353 * These fields are for EXT4_DYNAMIC_REV superblocks only. 1354 * 1355 * Note: the difference between the compatible feature set and 1356 * the incompatible feature set is that if there is a bit set 1357 * in the incompatible feature set that the kernel doesn't 1358 * know about, it should refuse to mount the filesystem. 1359 * 1360 * e2fsck's requirements are more strict; if it doesn't know 1361 * about a feature in either the compatible or incompatible 1362 * feature set, it must abort and not try to meddle with 1363 * things it doesn't understand... 1364 */ 1365 __le32 s_first_ino; /* First non-reserved inode */ 1366 __le16 s_inode_size; /* size of inode structure */ 1367 __le16 s_block_group_nr; /* block group # of this superblock */ 1368 __le32 s_feature_compat; /* compatible feature set */ 1369/*60*/ __le32 s_feature_incompat; /* incompatible feature set */ 1370 __le32 s_feature_ro_compat; /* readonly-compatible feature set */ 1371/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */ 1372/*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */ 1373/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */ 1374/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */ 1375 /* 1376 * Performance hints. Directory preallocation should only 1377 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on. 1378 */ 1379 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/ 1380 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */ 1381 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */ 1382 /* 1383 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set. 1384 */ 1385/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */ 1386/*E0*/ __le32 s_journal_inum; /* inode number of journal file */ 1387 __le32 s_journal_dev; /* device number of journal file */ 1388 __le32 s_last_orphan; /* start of list of inodes to delete */ 1389 __le32 s_hash_seed[4]; /* HTREE hash seed */ 1390 __u8 s_def_hash_version; /* Default hash version to use */ 1391 __u8 s_jnl_backup_type; 1392 __le16 s_desc_size; /* size of group descriptor */ 1393/*100*/ __le32 s_default_mount_opts; 1394 __le32 s_first_meta_bg; /* First metablock block group */ 1395 __le32 s_mkfs_time; /* When the filesystem was created */ 1396 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */ 1397 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */ 1398/*150*/ __le32 s_blocks_count_hi; /* Blocks count */ 1399 __le32 s_r_blocks_count_hi; /* Reserved blocks count */ 1400 __le32 s_free_blocks_count_hi; /* Free blocks count */ 1401 __le16 s_min_extra_isize; /* All inodes have at least # bytes */ 1402 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */ 1403 __le32 s_flags; /* Miscellaneous flags */ 1404 __le16 s_raid_stride; /* RAID stride */ 1405 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */ 1406 __le64 s_mmp_block; /* Block for multi-mount protection */ 1407 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/ 1408 __u8 s_log_groups_per_flex; /* FLEX_BG group size */ 1409 __u8 s_checksum_type; /* metadata checksum algorithm used */ 1410 __u8 s_encryption_level; /* versioning level for encryption */ 1411 __u8 s_reserved_pad; /* Padding to next 32bits */ 1412 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */ 1413 __le32 s_snapshot_inum; /* Inode number of active snapshot */ 1414 __le32 s_snapshot_id; /* sequential ID of active snapshot */ 1415 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active 1416 snapshot's future use */ 1417 __le32 s_snapshot_list; /* inode number of the head of the 1418 on-disk snapshot list */ 1419#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count) 1420 __le32 s_error_count; /* number of fs errors */ 1421 __le32 s_first_error_time; /* first time an error happened */ 1422 __le32 s_first_error_ino; /* inode involved in first error */ 1423 __le64 s_first_error_block; /* block involved of first error */ 1424 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */ 1425 __le32 s_first_error_line; /* line number where error happened */ 1426 __le32 s_last_error_time; /* most recent time of an error */ 1427 __le32 s_last_error_ino; /* inode involved in last error */ 1428 __le32 s_last_error_line; /* line number where error happened */ 1429 __le64 s_last_error_block; /* block involved of last error */ 1430 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */ 1431#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts) 1432 __u8 s_mount_opts[64]; 1433 __le32 s_usr_quota_inum; /* inode for tracking user quota */ 1434 __le32 s_grp_quota_inum; /* inode for tracking group quota */ 1435 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */ 1436 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */ 1437 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */ 1438 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */ 1439 __le32 s_lpf_ino; /* Location of the lost+found inode */ 1440 __le32 s_prj_quota_inum; /* inode for tracking project quota */ 1441 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */ 1442 __u8 s_wtime_hi; 1443 __u8 s_mtime_hi; 1444 __u8 s_mkfs_time_hi; 1445 __u8 s_lastcheck_hi; 1446 __u8 s_first_error_time_hi; 1447 __u8 s_last_error_time_hi; 1448 __u8 s_first_error_errcode; 1449 __u8 s_last_error_errcode; 1450 __le16 s_encoding; /* Filename charset encoding */ 1451 __le16 s_encoding_flags; /* Filename charset encoding flags */ 1452 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */ 1453 __le16 s_def_resuid_hi; 1454 __le16 s_def_resgid_hi; 1455 __le32 s_reserved[93]; /* Padding to the end of the block */ 1456 __le32 s_checksum; /* crc32c(superblock) */ 1457}; 1458 1459#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START) 1460 1461#ifdef __KERNEL__ 1462 1463/* Number of quota types we support */ 1464#define EXT4_MAXQUOTAS 3 1465 1466#define EXT4_ENC_UTF8_12_1 1 1467 1468/* Types of ext4 journal triggers */ 1469enum ext4_journal_trigger_type { 1470 EXT4_JTR_ORPHAN_FILE, 1471 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */ 1472}; 1473 1474#define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE 1475 1476struct ext4_journal_trigger { 1477 struct jbd2_buffer_trigger_type tr_triggers; 1478 struct super_block *sb; 1479}; 1480 1481static inline struct ext4_journal_trigger *EXT4_TRIGGER( 1482 struct jbd2_buffer_trigger_type *trigger) 1483{ 1484 return container_of(trigger, struct ext4_journal_trigger, tr_triggers); 1485} 1486 1487#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04 1488 1489/* Structure at the tail of orphan block */ 1490struct ext4_orphan_block_tail { 1491 __le32 ob_magic; 1492 __le32 ob_checksum; 1493}; 1494 1495static inline int ext4_inodes_per_orphan_block(struct super_block *sb) 1496{ 1497 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) / 1498 sizeof(u32); 1499} 1500 1501struct ext4_orphan_block { 1502 atomic_t ob_free_entries; /* Number of free orphan entries in block */ 1503 struct buffer_head *ob_bh; /* Buffer for orphan block */ 1504}; 1505 1506/* 1507 * Info about orphan file. 1508 */ 1509struct ext4_orphan_info { 1510 int of_blocks; /* Number of orphan blocks in a file */ 1511 __u32 of_csum_seed; /* Checksum seed for orphan file */ 1512 struct ext4_orphan_block *of_binfo; /* Array with info about orphan 1513 * file blocks */ 1514}; 1515 1516/* 1517 * fourth extended-fs super-block data in memory 1518 */ 1519struct ext4_sb_info { 1520 unsigned long s_desc_size; /* Size of a group descriptor in bytes */ 1521 unsigned long s_inodes_per_block;/* Number of inodes per block */ 1522 unsigned long s_blocks_per_group;/* Number of blocks in a group */ 1523 unsigned long s_clusters_per_group; /* Number of clusters in a group */ 1524 unsigned long s_inodes_per_group;/* Number of inodes in a group */ 1525 unsigned long s_itb_per_group; /* Number of inode table blocks per group */ 1526 unsigned long s_gdb_count; /* Number of group descriptor blocks */ 1527 unsigned long s_desc_per_block; /* Number of group descriptors per block */ 1528 ext4_group_t s_groups_count; /* Number of groups in the fs */ 1529 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ 1530 unsigned long s_overhead; /* # of fs overhead clusters */ 1531 unsigned int s_cluster_ratio; /* Number of blocks per cluster */ 1532 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ 1533 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ 1534 struct buffer_head * s_sbh; /* Buffer containing the super block */ 1535 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */ 1536 /* Array of bh's for the block group descriptors */ 1537 struct buffer_head * __rcu *s_group_desc; 1538 unsigned int s_mount_opt; 1539 unsigned int s_mount_opt2; 1540 unsigned long s_mount_flags; 1541 unsigned int s_def_mount_opt; 1542 unsigned int s_def_mount_opt2; 1543 ext4_fsblk_t s_sb_block; 1544 atomic64_t s_resv_clusters; 1545 kuid_t s_resuid; 1546 kgid_t s_resgid; 1547 unsigned short s_mount_state; 1548 unsigned short s_pad; 1549 int s_addr_per_block_bits; 1550 int s_desc_per_block_bits; 1551 int s_inode_size; 1552 int s_first_ino; 1553 unsigned int s_inode_readahead_blks; 1554 unsigned int s_inode_goal; 1555 u32 s_hash_seed[4]; 1556 int s_def_hash_version; 1557 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */ 1558 struct percpu_counter s_freeclusters_counter; 1559 struct percpu_counter s_freeinodes_counter; 1560 struct percpu_counter s_dirs_counter; 1561 struct percpu_counter s_dirtyclusters_counter; 1562 struct percpu_counter s_sra_exceeded_retry_limit; 1563 struct blockgroup_lock *s_blockgroup_lock; 1564 struct proc_dir_entry *s_proc; 1565 struct kobject s_kobj; 1566 struct completion s_kobj_unregister; 1567 struct super_block *s_sb; 1568 struct buffer_head *s_mmp_bh; 1569 1570 /* Journaling */ 1571 struct journal_s *s_journal; 1572 unsigned long s_ext4_flags; /* Ext4 superblock flags */ 1573 struct mutex s_orphan_lock; /* Protects on disk list changes */ 1574 struct list_head s_orphan; /* List of orphaned inodes in on disk 1575 list */ 1576 struct ext4_orphan_info s_orphan_info; 1577 unsigned long s_commit_interval; 1578 u32 s_max_batch_time; 1579 u32 s_min_batch_time; 1580 struct file *s_journal_bdev_file; 1581#ifdef CONFIG_QUOTA 1582 /* Names of quota files with journalled quota */ 1583 char __rcu *s_qf_names[EXT4_MAXQUOTAS]; 1584 int s_jquota_fmt; /* Format of quota to use */ 1585#endif 1586 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */ 1587 struct ext4_system_blocks __rcu *s_system_blks; 1588 1589#ifdef EXTENTS_STATS 1590 /* ext4 extents stats */ 1591 unsigned long s_ext_min; 1592 unsigned long s_ext_max; 1593 unsigned long s_depth_max; 1594 spinlock_t s_ext_stats_lock; 1595 unsigned long s_ext_blocks; 1596 unsigned long s_ext_extents; 1597#endif 1598 1599 /* for buddy allocator */ 1600 struct ext4_group_info ** __rcu *s_group_info; 1601 struct inode *s_buddy_cache; 1602 spinlock_t s_md_lock; 1603 unsigned short *s_mb_offsets; 1604 unsigned int *s_mb_maxs; 1605 unsigned int s_group_info_size; 1606 atomic_t s_mb_free_pending; 1607 struct list_head s_freed_data_list[2]; /* List of blocks to be freed 1608 after commit completed */ 1609 struct list_head s_discard_list; 1610 struct work_struct s_discard_work; 1611 atomic_t s_retry_alloc_pending; 1612 struct xarray *s_mb_avg_fragment_size; 1613 struct xarray *s_mb_largest_free_orders; 1614 1615 /* tunables */ 1616 unsigned long s_stripe; 1617 unsigned int s_mb_max_linear_groups; 1618 unsigned int s_mb_stream_request; 1619 unsigned int s_mb_max_to_scan; 1620 unsigned int s_mb_min_to_scan; 1621 unsigned int s_mb_stats; 1622 unsigned int s_mb_order2_reqs; 1623 unsigned int s_mb_group_prealloc; 1624 unsigned int s_max_dir_size_kb; 1625 unsigned int s_mb_prefetch; 1626 unsigned int s_mb_prefetch_limit; 1627 unsigned int s_mb_best_avail_max_trim_order; 1628 unsigned int s_sb_update_sec; 1629 unsigned int s_sb_update_kb; 1630 1631 /* where last allocation was done - for stream allocation */ 1632 ext4_group_t *s_mb_last_groups; 1633 unsigned int s_mb_nr_global_goals; 1634 1635 /* stats for buddy allocator */ 1636 atomic_t s_bal_reqs; /* number of reqs with len > 1 */ 1637 atomic_t s_bal_success; /* we found long enough chunks */ 1638 atomic_t s_bal_allocated; /* in blocks */ 1639 atomic_t s_bal_ex_scanned; /* total extents scanned */ 1640 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */ 1641 atomic_t s_bal_groups_scanned; /* number of groups scanned */ 1642 atomic_t s_bal_goals; /* goal hits */ 1643 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */ 1644 atomic_t s_bal_len_goals; /* len goal hits */ 1645 atomic_t s_bal_breaks; /* too long searches */ 1646 atomic_t s_bal_2orders; /* 2^order hits */ 1647 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS]; 1648 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS]; 1649 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */ 1650 atomic_t s_mb_buddies_generated; /* number of buddies generated */ 1651 atomic64_t s_mb_generation_time; 1652 atomic_t s_mb_lost_chunks; 1653 atomic_t s_mb_preallocated; 1654 atomic_t s_mb_discarded; 1655 atomic_t s_lock_busy; 1656 1657 /* locality groups */ 1658 struct ext4_locality_group __percpu *s_locality_groups; 1659 1660 /* for write statistics */ 1661 unsigned long s_sectors_written_start; 1662 u64 s_kbytes_written; 1663 1664 /* the size of zero-out chunk */ 1665 unsigned int s_extent_max_zeroout_kb; 1666 1667 unsigned int s_log_groups_per_flex; 1668 struct flex_groups * __rcu *s_flex_groups; 1669 ext4_group_t s_flex_groups_allocated; 1670 1671 /* workqueue for reserved extent conversions (buffered io) */ 1672 struct workqueue_struct *rsv_conversion_wq; 1673 1674 /* timer for periodic error stats printing */ 1675 struct timer_list s_err_report; 1676 1677 /* Lazy inode table initialization info */ 1678 struct ext4_li_request *s_li_request; 1679 /* Wait multiplier for lazy initialization thread */ 1680 unsigned int s_li_wait_mult; 1681 1682 /* Kernel thread for multiple mount protection */ 1683 struct task_struct *s_mmp_tsk; 1684 1685 /* record the last minlen when FITRIM is called. */ 1686 unsigned long s_last_trim_minblks; 1687 1688 /* Precomputed FS UUID checksum for seeding other checksums */ 1689 __u32 s_csum_seed; 1690 1691 /* Reclaim extents from extent status tree */ 1692 struct shrinker *s_es_shrinker; 1693 struct list_head s_es_list; /* List of inodes with reclaimable extents */ 1694 long s_es_nr_inode; 1695 struct ext4_es_stats s_es_stats; 1696 struct mb_cache *s_ea_block_cache; 1697 struct mb_cache *s_ea_inode_cache; 1698 spinlock_t s_es_lock ____cacheline_aligned_in_smp; 1699 1700 /* Journal triggers for checksum computation */ 1701 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT]; 1702 1703 /* Ratelimit ext4 messages. */ 1704 struct ratelimit_state s_err_ratelimit_state; 1705 struct ratelimit_state s_warning_ratelimit_state; 1706 struct ratelimit_state s_msg_ratelimit_state; 1707 atomic_t s_warning_count; 1708 atomic_t s_msg_count; 1709 1710 /* Encryption policy for '-o test_dummy_encryption' */ 1711 struct fscrypt_dummy_policy s_dummy_enc_policy; 1712 1713 /* 1714 * Barrier between writepages ops and changing any inode's JOURNAL_DATA 1715 * or EXTENTS flag or between writepages ops and changing DELALLOC or 1716 * DIOREAD_NOLOCK mount options on remount. 1717 */ 1718 struct percpu_rw_semaphore s_writepages_rwsem; 1719 struct dax_device *s_daxdev; 1720 u64 s_dax_part_off; 1721#ifdef CONFIG_EXT4_DEBUG 1722 unsigned long s_simulate_fail; 1723#endif 1724 /* Record the errseq of the backing block device */ 1725 errseq_t s_bdev_wb_err; 1726 spinlock_t s_bdev_wb_lock; 1727 1728 /* Information about errors that happened during this mount */ 1729 spinlock_t s_error_lock; 1730 int s_add_error_count; 1731 int s_first_error_code; 1732 __u32 s_first_error_line; 1733 __u32 s_first_error_ino; 1734 __u64 s_first_error_block; 1735 const char *s_first_error_func; 1736 time64_t s_first_error_time; 1737 int s_last_error_code; 1738 __u32 s_last_error_line; 1739 __u32 s_last_error_ino; 1740 __u64 s_last_error_block; 1741 const char *s_last_error_func; 1742 time64_t s_last_error_time; 1743 /* 1744 * If we are in a context where we cannot update the on-disk 1745 * superblock, we queue the work here. This is used to update 1746 * the error information in the superblock, and for periodic 1747 * updates of the superblock called from the commit callback 1748 * function. 1749 */ 1750 struct work_struct s_sb_upd_work; 1751 1752 /* Atomic write unit values in bytes */ 1753 unsigned int s_awu_min; 1754 unsigned int s_awu_max; 1755 1756 /* Ext4 fast commit sub transaction ID */ 1757 atomic_t s_fc_subtid; 1758 1759 /* 1760 * After commit starts, the main queue gets locked, and the further 1761 * updates get added in the staging queue. 1762 */ 1763#define FC_Q_MAIN 0 1764#define FC_Q_STAGING 1 1765 struct list_head s_fc_q[2]; /* Inodes staged for fast commit 1766 * that have data changes in them. 1767 */ 1768 struct list_head s_fc_dentry_q[2]; /* directory entry updates */ 1769 unsigned int s_fc_bytes; 1770 /* 1771 * Main fast commit lock. This lock protects accesses to the 1772 * following fields: 1773 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh. 1774 */ 1775 struct mutex s_fc_lock; 1776 struct buffer_head *s_fc_bh; 1777 struct ext4_fc_stats s_fc_stats; 1778 tid_t s_fc_ineligible_tid; 1779#ifdef CONFIG_EXT4_DEBUG 1780 int s_fc_debug_max_replay; 1781#endif 1782 struct ext4_fc_replay_state s_fc_replay_state; 1783}; 1784 1785static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) 1786{ 1787 return sb->s_fs_info; 1788} 1789static inline struct ext4_inode_info *EXT4_I(struct inode *inode) 1790{ 1791 return container_of(inode, struct ext4_inode_info, vfs_inode); 1792} 1793 1794static inline int ext4_writepages_down_read(struct super_block *sb) 1795{ 1796 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem); 1797 return memalloc_nofs_save(); 1798} 1799 1800static inline void ext4_writepages_up_read(struct super_block *sb, int ctx) 1801{ 1802 memalloc_nofs_restore(ctx); 1803 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem); 1804} 1805 1806static inline int ext4_writepages_down_write(struct super_block *sb) 1807{ 1808 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem); 1809 return memalloc_nofs_save(); 1810} 1811 1812static inline void ext4_writepages_up_write(struct super_block *sb, int ctx) 1813{ 1814 memalloc_nofs_restore(ctx); 1815 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem); 1816} 1817 1818static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) 1819{ 1820 return ino == EXT4_ROOT_INO || 1821 (ino >= EXT4_FIRST_INO(sb) && 1822 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); 1823} 1824 1825static inline int ext4_get_resuid(struct ext4_super_block *es) 1826{ 1827 return le16_to_cpu(es->s_def_resuid) | 1828 le16_to_cpu(es->s_def_resuid_hi) << 16; 1829} 1830 1831static inline int ext4_get_resgid(struct ext4_super_block *es) 1832{ 1833 return le16_to_cpu(es->s_def_resgid) | 1834 le16_to_cpu(es->s_def_resgid_hi) << 16; 1835} 1836 1837/* 1838 * Returns: sbi->field[index] 1839 * Used to access an array element from the following sbi fields which require 1840 * rcu protection to avoid dereferencing an invalid pointer due to reassignment 1841 * - s_group_desc 1842 * - s_group_info 1843 * - s_flex_group 1844 */ 1845#define sbi_array_rcu_deref(sbi, field, index) \ 1846({ \ 1847 typeof(*((sbi)->field)) _v; \ 1848 rcu_read_lock(); \ 1849 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \ 1850 rcu_read_unlock(); \ 1851 _v; \ 1852}) 1853 1854/* 1855 * run-time mount flags 1856 */ 1857enum { 1858 EXT4_MF_MNTDIR_SAMPLED, 1859 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */ 1860 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */ 1861}; 1862 1863static inline void ext4_set_mount_flag(struct super_block *sb, int bit) 1864{ 1865 set_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1866} 1867 1868static inline void ext4_clear_mount_flag(struct super_block *sb, int bit) 1869{ 1870 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1871} 1872 1873static inline int ext4_test_mount_flag(struct super_block *sb, int bit) 1874{ 1875 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags); 1876} 1877 1878 1879/* 1880 * Simulate_fail codes 1881 */ 1882#define EXT4_SIM_BBITMAP_EIO 1 1883#define EXT4_SIM_BBITMAP_CRC 2 1884#define EXT4_SIM_IBITMAP_EIO 3 1885#define EXT4_SIM_IBITMAP_CRC 4 1886#define EXT4_SIM_INODE_EIO 5 1887#define EXT4_SIM_INODE_CRC 6 1888#define EXT4_SIM_DIRBLOCK_EIO 7 1889#define EXT4_SIM_DIRBLOCK_CRC 8 1890 1891static inline bool ext4_simulate_fail(struct super_block *sb, 1892 unsigned long code) 1893{ 1894#ifdef CONFIG_EXT4_DEBUG 1895 struct ext4_sb_info *sbi = EXT4_SB(sb); 1896 1897 if (unlikely(sbi->s_simulate_fail == code)) { 1898 sbi->s_simulate_fail = 0; 1899 return true; 1900 } 1901#endif 1902 return false; 1903} 1904 1905/* 1906 * Error number codes for s_{first,last}_error_errno 1907 * 1908 * Linux errno numbers are architecture specific, so we need to translate 1909 * them into something which is architecture independent. We don't define 1910 * codes for all errno's; just the ones which are most likely to be the cause 1911 * of an ext4_error() call. 1912 */ 1913#define EXT4_ERR_UNKNOWN 1 1914#define EXT4_ERR_EIO 2 1915#define EXT4_ERR_ENOMEM 3 1916#define EXT4_ERR_EFSBADCRC 4 1917#define EXT4_ERR_EFSCORRUPTED 5 1918#define EXT4_ERR_ENOSPC 6 1919#define EXT4_ERR_ENOKEY 7 1920#define EXT4_ERR_EROFS 8 1921#define EXT4_ERR_EFBIG 9 1922#define EXT4_ERR_EEXIST 10 1923#define EXT4_ERR_ERANGE 11 1924#define EXT4_ERR_EOVERFLOW 12 1925#define EXT4_ERR_EBUSY 13 1926#define EXT4_ERR_ENOTDIR 14 1927#define EXT4_ERR_ENOTEMPTY 15 1928#define EXT4_ERR_ESHUTDOWN 16 1929#define EXT4_ERR_EFAULT 17 1930 1931/* 1932 * Inode dynamic state flags 1933 */ 1934enum { 1935 EXT4_STATE_NEW, /* inode is newly created */ 1936 EXT4_STATE_XATTR, /* has in-inode xattrs */ 1937 EXT4_STATE_NO_EXPAND, /* No space for expansion */ 1938 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */ 1939 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */ 1940 EXT4_STATE_NEWENTRY, /* File just added to dir */ 1941 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */ 1942 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */ 1943 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */ 1944 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 1945 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */ 1946 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */ 1947 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */ 1948}; 1949 1950#define EXT4_INODE_BIT_FNS(name, field, offset) \ 1951static inline int ext4_test_inode_##name(struct inode *inode, int bit) \ 1952{ \ 1953 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1954} \ 1955static inline void ext4_set_inode_##name(struct inode *inode, int bit) \ 1956{ \ 1957 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1958} \ 1959static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \ 1960{ \ 1961 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \ 1962} 1963 1964/* Add these declarations here only so that these functions can be 1965 * found by name. Otherwise, they are very hard to locate. */ 1966static inline int ext4_test_inode_flag(struct inode *inode, int bit); 1967static inline void ext4_set_inode_flag(struct inode *inode, int bit); 1968static inline void ext4_clear_inode_flag(struct inode *inode, int bit); 1969EXT4_INODE_BIT_FNS(flag, flags, 0) 1970 1971/* Add these declarations here only so that these functions can be 1972 * found by name. Otherwise, they are very hard to locate. */ 1973static inline int ext4_test_inode_state(struct inode *inode, int bit); 1974static inline void ext4_set_inode_state(struct inode *inode, int bit); 1975static inline void ext4_clear_inode_state(struct inode *inode, int bit); 1976#if (BITS_PER_LONG < 64) 1977EXT4_INODE_BIT_FNS(state, state_flags, 0) 1978 1979static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1980{ 1981 (ei)->i_state_flags = 0; 1982} 1983#else 1984EXT4_INODE_BIT_FNS(state, flags, 32) 1985 1986static inline void ext4_clear_state_flags(struct ext4_inode_info *ei) 1987{ 1988 /* We depend on the fact that callers will set i_flags */ 1989} 1990#endif 1991#else 1992/* Assume that user mode programs are passing in an ext4fs superblock, not 1993 * a kernel struct super_block. This will allow us to call the feature-test 1994 * macros from user land. */ 1995#define EXT4_SB(sb) (sb) 1996#endif 1997 1998static inline bool ext4_verity_in_progress(struct inode *inode) 1999{ 2000 return IS_ENABLED(CONFIG_FS_VERITY) && 2001 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS); 2002} 2003 2004#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime 2005 2006/* 2007 * Check whether the inode is tracked as orphan (either in orphan file or 2008 * orphan list). 2009 */ 2010static inline bool ext4_inode_orphan_tracked(struct inode *inode) 2011{ 2012 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) || 2013 !list_empty(&EXT4_I(inode)->i_orphan); 2014} 2015 2016/* 2017 * Codes for operating systems 2018 */ 2019#define EXT4_OS_LINUX 0 2020#define EXT4_OS_HURD 1 2021#define EXT4_OS_MASIX 2 2022#define EXT4_OS_FREEBSD 3 2023#define EXT4_OS_LITES 4 2024 2025/* 2026 * Revision levels 2027 */ 2028#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */ 2029#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */ 2030 2031#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV 2032 2033#define EXT4_GOOD_OLD_INODE_SIZE 128 2034 2035#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN) 2036#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX 2037#define EXT4_TIMESTAMP_MIN S32_MIN 2038 2039/* 2040 * Feature set definitions 2041 */ 2042 2043#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001 2044#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002 2045#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004 2046#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008 2047#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010 2048#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020 2049#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200 2050/* 2051 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes 2052 * incompatible only if fast commit blocks are present in the FS. Since we 2053 * clear the journal (and thus the fast commit blocks), we don't mark FS as 2054 * incompatible. We also have a JBD2 incompat feature, which gets set when 2055 * there are fast commit blocks present in the journal. 2056 */ 2057#define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400 2058#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800 2059#define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */ 2060 2061#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001 2062#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002 2063#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004 2064#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008 2065#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010 2066#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020 2067#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040 2068#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100 2069#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200 2070/* 2071 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When 2072 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as 2073 * all other data structures' checksums. However, the METADATA_CSUM and 2074 * GDT_CSUM bits are mutually exclusive. 2075 */ 2076#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400 2077#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000 2078#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000 2079#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000 2080#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be 2081 non-empty */ 2082 2083#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001 2084#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002 2085#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */ 2086#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */ 2087#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010 2088#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */ 2089#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080 2090#define EXT4_FEATURE_INCOMPAT_MMP 0x0100 2091#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200 2092#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */ 2093#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */ 2094#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000 2095#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */ 2096#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */ 2097#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000 2098#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000 2099 2100extern void ext4_update_dynamic_rev(struct super_block *sb); 2101 2102#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \ 2103static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2104{ \ 2105 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2106 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \ 2107} \ 2108static inline void ext4_set_feature_##name(struct super_block *sb) \ 2109{ \ 2110 ext4_update_dynamic_rev(sb); \ 2111 EXT4_SB(sb)->s_es->s_feature_compat |= \ 2112 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2113} \ 2114static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2115{ \ 2116 EXT4_SB(sb)->s_es->s_feature_compat &= \ 2117 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \ 2118} 2119 2120#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \ 2121static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2122{ \ 2123 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2124 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \ 2125} \ 2126static inline void ext4_set_feature_##name(struct super_block *sb) \ 2127{ \ 2128 ext4_update_dynamic_rev(sb); \ 2129 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \ 2130 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2131} \ 2132static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2133{ \ 2134 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \ 2135 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \ 2136} 2137 2138#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \ 2139static inline bool ext4_has_feature_##name(struct super_block *sb) \ 2140{ \ 2141 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2142 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \ 2143} \ 2144static inline void ext4_set_feature_##name(struct super_block *sb) \ 2145{ \ 2146 ext4_update_dynamic_rev(sb); \ 2147 EXT4_SB(sb)->s_es->s_feature_incompat |= \ 2148 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2149} \ 2150static inline void ext4_clear_feature_##name(struct super_block *sb) \ 2151{ \ 2152 EXT4_SB(sb)->s_es->s_feature_incompat &= \ 2153 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \ 2154} 2155 2156EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC) 2157EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES) 2158EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL) 2159EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR) 2160EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE) 2161EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX) 2162EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2) 2163EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT) 2164EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES) 2165EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE) 2166 2167EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER) 2168EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE) 2169EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR) 2170EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE) 2171EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM) 2172EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK) 2173EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE) 2174EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA) 2175EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC) 2176EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM) 2177EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY) 2178EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT) 2179EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY) 2180EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT) 2181 2182EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION) 2183EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE) 2184EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER) 2185EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV) 2186EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG) 2187EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS) 2188EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT) 2189EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP) 2190EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG) 2191EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE) 2192EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA) 2193EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED) 2194EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR) 2195EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA) 2196EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT) 2197EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD) 2198 2199#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2200#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2201 EXT4_FEATURE_INCOMPAT_META_BG) 2202#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2203 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2204 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2205 2206#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR 2207#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2208 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2209 EXT4_FEATURE_INCOMPAT_META_BG) 2210#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2211 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2212 EXT4_FEATURE_RO_COMPAT_BTREE_DIR) 2213 2214#define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \ 2215 EXT4_FEATURE_COMPAT_ORPHAN_FILE) 2216#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \ 2217 EXT4_FEATURE_INCOMPAT_RECOVER| \ 2218 EXT4_FEATURE_INCOMPAT_META_BG| \ 2219 EXT4_FEATURE_INCOMPAT_EXTENTS| \ 2220 EXT4_FEATURE_INCOMPAT_64BIT| \ 2221 EXT4_FEATURE_INCOMPAT_FLEX_BG| \ 2222 EXT4_FEATURE_INCOMPAT_EA_INODE| \ 2223 EXT4_FEATURE_INCOMPAT_MMP | \ 2224 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \ 2225 EXT4_FEATURE_INCOMPAT_ENCRYPT | \ 2226 EXT4_FEATURE_INCOMPAT_CASEFOLD | \ 2227 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \ 2228 EXT4_FEATURE_INCOMPAT_LARGEDIR) 2229#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \ 2230 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \ 2231 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \ 2232 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \ 2233 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \ 2234 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\ 2235 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\ 2236 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\ 2237 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\ 2238 EXT4_FEATURE_RO_COMPAT_QUOTA |\ 2239 EXT4_FEATURE_RO_COMPAT_PROJECT |\ 2240 EXT4_FEATURE_RO_COMPAT_VERITY |\ 2241 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT) 2242 2243#define EXTN_FEATURE_FUNCS(ver) \ 2244static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \ 2245{ \ 2246 return ((EXT4_SB(sb)->s_es->s_feature_compat & \ 2247 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \ 2248} \ 2249static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \ 2250{ \ 2251 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \ 2252 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \ 2253} \ 2254static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \ 2255{ \ 2256 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \ 2257 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \ 2258} 2259 2260EXTN_FEATURE_FUNCS(2) 2261EXTN_FEATURE_FUNCS(3) 2262EXTN_FEATURE_FUNCS(4) 2263 2264static inline bool ext4_has_compat_features(struct super_block *sb) 2265{ 2266 return (EXT4_SB(sb)->s_es->s_feature_compat != 0); 2267} 2268static inline bool ext4_has_ro_compat_features(struct super_block *sb) 2269{ 2270 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0); 2271} 2272static inline bool ext4_has_incompat_features(struct super_block *sb) 2273{ 2274 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0); 2275} 2276 2277extern int ext4_feature_set_ok(struct super_block *sb, int readonly); 2278 2279/* 2280 * Superblock flags 2281 */ 2282enum { 2283 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */ 2284 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */ 2285 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */ 2286 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */ 2287}; 2288 2289static inline int ext4_forced_shutdown(struct super_block *sb) 2290{ 2291 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags); 2292} 2293 2294static inline int ext4_emergency_ro(struct super_block *sb) 2295{ 2296 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags); 2297} 2298 2299static inline int ext4_emergency_state(struct super_block *sb) 2300{ 2301 if (unlikely(ext4_forced_shutdown(sb))) 2302 return -EIO; 2303 if (unlikely(ext4_emergency_ro(sb))) 2304 return -EROFS; 2305 return 0; 2306} 2307 2308/* 2309 * Default values for user and/or group using reserved blocks 2310 */ 2311#define EXT4_DEF_RESUID 0 2312#define EXT4_DEF_RESGID 0 2313 2314/* 2315 * Default project ID 2316 */ 2317#define EXT4_DEF_PROJID 0 2318 2319#define EXT4_DEF_INODE_READAHEAD_BLKS 32 2320 2321/* 2322 * Default mount options 2323 */ 2324#define EXT4_DEFM_DEBUG 0x0001 2325#define EXT4_DEFM_BSDGROUPS 0x0002 2326#define EXT4_DEFM_XATTR_USER 0x0004 2327#define EXT4_DEFM_ACL 0x0008 2328#define EXT4_DEFM_UID16 0x0010 2329#define EXT4_DEFM_JMODE 0x0060 2330#define EXT4_DEFM_JMODE_DATA 0x0020 2331#define EXT4_DEFM_JMODE_ORDERED 0x0040 2332#define EXT4_DEFM_JMODE_WBACK 0x0060 2333#define EXT4_DEFM_NOBARRIER 0x0100 2334#define EXT4_DEFM_BLOCK_VALIDITY 0x0200 2335#define EXT4_DEFM_DISCARD 0x0400 2336#define EXT4_DEFM_NODELALLOC 0x0800 2337 2338/* 2339 * Default journal batch times and ioprio. 2340 */ 2341#define EXT4_DEF_MIN_BATCH_TIME 0 2342#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */ 2343#define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) 2344 2345 2346/* 2347 * Default values for superblock update 2348 */ 2349#define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */ 2350#define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */ 2351 2352 2353/* 2354 * Minimum number of groups in a flexgroup before we separate out 2355 * directories into the first block group of a flexgroup 2356 */ 2357#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4 2358 2359/* 2360 * Structure of a directory entry 2361 */ 2362#define EXT4_NAME_LEN 255 2363/* 2364 * Base length of the ext4 directory entry excluding the name length 2365 */ 2366#define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN) 2367 2368struct ext4_dir_entry { 2369 __le32 inode; /* Inode number */ 2370 __le16 rec_len; /* Directory entry length */ 2371 __le16 name_len; /* Name length */ 2372 char name[EXT4_NAME_LEN]; /* File name */ 2373}; 2374 2375 2376/* 2377 * Encrypted Casefolded entries require saving the hash on disk. This structure 2378 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned 2379 * boundary. 2380 */ 2381struct ext4_dir_entry_hash { 2382 __le32 hash; 2383 __le32 minor_hash; 2384}; 2385 2386/* 2387 * The new version of the directory entry. Since EXT4 structures are 2388 * stored in intel byte order, and the name_len field could never be 2389 * bigger than 255 chars, it's safe to reclaim the extra byte for the 2390 * file_type field. 2391 */ 2392struct ext4_dir_entry_2 { 2393 __le32 inode; /* Inode number */ 2394 __le16 rec_len; /* Directory entry length */ 2395 __u8 name_len; /* Name length */ 2396 __u8 file_type; /* See file type macros EXT4_FT_* below */ 2397 char name[EXT4_NAME_LEN]; /* File name */ 2398}; 2399 2400/* 2401 * Access the hashes at the end of ext4_dir_entry_2 2402 */ 2403#define EXT4_DIRENT_HASHES(entry) \ 2404 ((struct ext4_dir_entry_hash *) \ 2405 (((void *)(entry)) + \ 2406 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND))) 2407#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash) 2408#define EXT4_DIRENT_MINOR_HASH(entry) \ 2409 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash) 2410 2411static inline bool ext4_hash_in_dirent(const struct inode *inode) 2412{ 2413 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode); 2414} 2415 2416/* 2417 * This is a bogus directory entry at the end of each leaf block that 2418 * records checksums. 2419 */ 2420struct ext4_dir_entry_tail { 2421 __le32 det_reserved_zero1; /* Pretend to be unused */ 2422 __le16 det_rec_len; /* 12 */ 2423 __u8 det_reserved_zero2; /* Zero name length */ 2424 __u8 det_reserved_ft; /* 0xDE, fake file type */ 2425 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */ 2426}; 2427 2428#define EXT4_DIRENT_TAIL(block, blocksize) \ 2429 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \ 2430 ((blocksize) - \ 2431 sizeof(struct ext4_dir_entry_tail)))) 2432 2433/* 2434 * Ext4 directory file types. Only the low 3 bits are used. The 2435 * other bits are reserved for now. 2436 */ 2437#define EXT4_FT_UNKNOWN 0 2438#define EXT4_FT_REG_FILE 1 2439#define EXT4_FT_DIR 2 2440#define EXT4_FT_CHRDEV 3 2441#define EXT4_FT_BLKDEV 4 2442#define EXT4_FT_FIFO 5 2443#define EXT4_FT_SOCK 6 2444#define EXT4_FT_SYMLINK 7 2445 2446#define EXT4_FT_MAX 8 2447 2448#define EXT4_FT_DIR_CSUM 0xDE 2449 2450/* 2451 * EXT4_DIR_PAD defines the directory entries boundaries 2452 * 2453 * NOTE: It must be a multiple of 4 2454 */ 2455#define EXT4_DIR_PAD 4 2456#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1) 2457#define EXT4_MAX_REC_LEN ((1<<16)-1) 2458 2459/* 2460 * The rec_len is dependent on the type of directory. Directories that are 2461 * casefolded and encrypted need to store the hash as well, so we add room for 2462 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should 2463 * pass NULL for dir, as those entries do not use the extra fields. 2464 */ 2465static inline unsigned int ext4_dir_rec_len(__u8 name_len, 2466 const struct inode *dir) 2467{ 2468 int rec_len = (name_len + 8 + EXT4_DIR_ROUND); 2469 2470 if (dir && ext4_hash_in_dirent(dir)) 2471 rec_len += sizeof(struct ext4_dir_entry_hash); 2472 return (rec_len & ~EXT4_DIR_ROUND); 2473} 2474 2475/* 2476 * If we ever get support for fs block sizes > page_size, we'll need 2477 * to remove the #if statements in the next two functions... 2478 */ 2479static inline unsigned int 2480ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize) 2481{ 2482 unsigned len = le16_to_cpu(dlen); 2483 2484#if (PAGE_SIZE >= 65536) 2485 if (len == EXT4_MAX_REC_LEN || len == 0) 2486 return blocksize; 2487 return (len & 65532) | ((len & 3) << 16); 2488#else 2489 return len; 2490#endif 2491} 2492 2493static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize) 2494{ 2495 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3)); 2496#if (PAGE_SIZE >= 65536) 2497 if (len < 65536) 2498 return cpu_to_le16(len); 2499 if (len == blocksize) { 2500 if (blocksize == 65536) 2501 return cpu_to_le16(EXT4_MAX_REC_LEN); 2502 else 2503 return cpu_to_le16(0); 2504 } 2505 return cpu_to_le16((len & 65532) | ((len >> 16) & 3)); 2506#else 2507 return cpu_to_le16(len); 2508#endif 2509} 2510 2511/* 2512 * Hash Tree Directory indexing 2513 * (c) Daniel Phillips, 2001 2514 */ 2515 2516#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \ 2517 ext4_test_inode_flag((dir), EXT4_INODE_INDEX)) 2518#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \ 2519 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir))) 2520#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1) 2521 2522/* Legal values for the dx_root hash_version field: */ 2523 2524#define DX_HASH_LEGACY 0 2525#define DX_HASH_HALF_MD4 1 2526#define DX_HASH_TEA 2 2527#define DX_HASH_LEGACY_UNSIGNED 3 2528#define DX_HASH_HALF_MD4_UNSIGNED 4 2529#define DX_HASH_TEA_UNSIGNED 5 2530#define DX_HASH_SIPHASH 6 2531#define DX_HASH_LAST DX_HASH_SIPHASH 2532 2533static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length) 2534{ 2535 return crc32c(crc, address, length); 2536} 2537 2538#ifdef __KERNEL__ 2539 2540/* hash info structure used by the directory hash */ 2541struct dx_hash_info 2542{ 2543 u32 hash; 2544 u32 minor_hash; 2545 int hash_version; 2546 u32 *seed; 2547}; 2548 2549 2550/* 32 and 64 bit signed EOF for dx directories */ 2551#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1) 2552#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1) 2553 2554 2555/* 2556 * Control parameters used by ext4_htree_next_block 2557 */ 2558#define HASH_NB_ALWAYS 1 2559 2560struct ext4_filename { 2561 const struct qstr *usr_fname; 2562 struct fscrypt_str disk_name; 2563 struct dx_hash_info hinfo; 2564#ifdef CONFIG_FS_ENCRYPTION 2565 struct fscrypt_str crypto_buf; 2566#endif 2567#if IS_ENABLED(CONFIG_UNICODE) 2568 struct qstr cf_name; 2569#endif 2570}; 2571 2572#define fname_name(p) ((p)->disk_name.name) 2573#define fname_usr_name(p) ((p)->usr_fname->name) 2574#define fname_len(p) ((p)->disk_name.len) 2575 2576/* 2577 * Describe an inode's exact location on disk and in memory 2578 */ 2579struct ext4_iloc 2580{ 2581 struct buffer_head *bh; 2582 unsigned long offset; 2583 ext4_group_t block_group; 2584}; 2585 2586static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc) 2587{ 2588 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset); 2589} 2590 2591static inline bool ext4_is_quota_file(struct inode *inode) 2592{ 2593 return IS_NOQUOTA(inode) && 2594 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL); 2595} 2596 2597/* 2598 * This structure is stuffed into the struct file's private_data field 2599 * for directories. It is where we put information so that we can do 2600 * readdir operations in hash tree order. 2601 */ 2602struct dir_private_info { 2603 struct rb_root root; 2604 struct rb_node *curr_node; 2605 struct fname *extra_fname; 2606 loff_t last_pos; 2607 __u32 curr_hash; 2608 __u32 curr_minor_hash; 2609 __u32 next_hash; 2610 u64 cookie; 2611 bool initialized; 2612}; 2613 2614/* calculate the first block number of the group */ 2615static inline ext4_fsblk_t 2616ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no) 2617{ 2618 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) + 2619 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block); 2620} 2621 2622/* 2623 * Special error return code only used by dx_probe() and its callers. 2624 */ 2625#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1)) 2626 2627/* htree levels for ext4 */ 2628#define EXT4_HTREE_LEVEL_COMPAT 2 2629#define EXT4_HTREE_LEVEL 3 2630 2631static inline int ext4_dir_htree_level(struct super_block *sb) 2632{ 2633 return ext4_has_feature_largedir(sb) ? 2634 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT; 2635} 2636 2637/* 2638 * Timeout and state flag for lazy initialization inode thread. 2639 */ 2640#define EXT4_DEF_LI_WAIT_MULT 10 2641#define EXT4_DEF_LI_MAX_START_DELAY 5 2642#define EXT4_LAZYINIT_QUIT 0x0001 2643#define EXT4_LAZYINIT_RUNNING 0x0002 2644 2645/* 2646 * Lazy inode table initialization info 2647 */ 2648struct ext4_lazy_init { 2649 unsigned long li_state; 2650 struct list_head li_request_list; 2651 struct mutex li_list_mtx; 2652}; 2653 2654enum ext4_li_mode { 2655 EXT4_LI_MODE_PREFETCH_BBITMAP, 2656 EXT4_LI_MODE_ITABLE, 2657}; 2658 2659struct ext4_li_request { 2660 struct super_block *lr_super; 2661 enum ext4_li_mode lr_mode; 2662 ext4_group_t lr_first_not_zeroed; 2663 ext4_group_t lr_next_group; 2664 struct list_head lr_request; 2665 unsigned long lr_next_sched; 2666 unsigned long lr_timeout; 2667}; 2668 2669struct ext4_features { 2670 struct kobject f_kobj; 2671 struct completion f_kobj_unregister; 2672}; 2673 2674/* 2675 * This structure will be used for multiple mount protection. It will be 2676 * written into the block number saved in the s_mmp_block field in the 2677 * superblock. Programs that check MMP should assume that if 2678 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe 2679 * to use the filesystem, regardless of how old the timestamp is. 2680 */ 2681#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */ 2682#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */ 2683#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */ 2684#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */ 2685 2686struct mmp_struct { 2687 __le32 mmp_magic; /* Magic number for MMP */ 2688 __le32 mmp_seq; /* Sequence no. updated periodically */ 2689 2690 /* 2691 * mmp_time, mmp_nodename & mmp_bdevname are only used for information 2692 * purposes and do not affect the correctness of the algorithm 2693 */ 2694 __le64 mmp_time; /* Time last updated */ 2695 char mmp_nodename[64]; /* Node which last updated MMP block */ 2696 char mmp_bdevname[32]; /* Bdev which last updated MMP block */ 2697 2698 /* 2699 * mmp_check_interval is used to verify if the MMP block has been 2700 * updated on the block device. The value is updated based on the 2701 * maximum time to write the MMP block during an update cycle. 2702 */ 2703 __le16 mmp_check_interval; 2704 2705 __le16 mmp_pad1; 2706 __le32 mmp_pad2[226]; 2707 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */ 2708}; 2709 2710/* arguments passed to the mmp thread */ 2711struct mmpd_data { 2712 struct buffer_head *bh; /* bh from initial read_mmp_block() */ 2713 struct super_block *sb; /* super block of the fs */ 2714}; 2715 2716/* 2717 * Check interval multiplier 2718 * The MMP block is written every update interval and initially checked every 2719 * update interval x the multiplier (the value is then adapted based on the 2720 * write latency). The reason is that writes can be delayed under load and we 2721 * don't want readers to incorrectly assume that the filesystem is no longer 2722 * in use. 2723 */ 2724#define EXT4_MMP_CHECK_MULT 2UL 2725 2726/* 2727 * Minimum interval for MMP checking in seconds. 2728 */ 2729#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL 2730 2731/* 2732 * Maximum interval for MMP checking in seconds. 2733 */ 2734#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL 2735 2736/* 2737 * Function prototypes 2738 */ 2739 2740/* 2741 * Ok, these declarations are also in <linux/kernel.h> but none of the 2742 * ext4 source programs needs to include it so they are duplicated here. 2743 */ 2744# define NORET_TYPE /**/ 2745# define ATTRIB_NORET __attribute__((noreturn)) 2746# define NORET_AND noreturn, 2747 2748/* bitmap.c */ 2749extern unsigned int ext4_count_free(char *bitmap, unsigned numchars); 2750void ext4_inode_bitmap_csum_set(struct super_block *sb, 2751 struct ext4_group_desc *gdp, 2752 struct buffer_head *bh); 2753int ext4_inode_bitmap_csum_verify(struct super_block *sb, 2754 struct ext4_group_desc *gdp, 2755 struct buffer_head *bh); 2756void ext4_block_bitmap_csum_set(struct super_block *sb, 2757 struct ext4_group_desc *gdp, 2758 struct buffer_head *bh); 2759int ext4_block_bitmap_csum_verify(struct super_block *sb, 2760 struct ext4_group_desc *gdp, 2761 struct buffer_head *bh); 2762 2763/* balloc.c */ 2764extern void ext4_get_group_no_and_offset(struct super_block *sb, 2765 ext4_fsblk_t blocknr, 2766 ext4_group_t *blockgrpp, 2767 ext4_grpblk_t *offsetp); 2768extern ext4_group_t ext4_get_group_number(struct super_block *sb, 2769 ext4_fsblk_t block); 2770 2771extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group); 2772extern unsigned long ext4_bg_num_gdb(struct super_block *sb, 2773 ext4_group_t group); 2774extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, 2775 ext4_fsblk_t goal, 2776 unsigned int flags, 2777 unsigned long *count, 2778 int *errp); 2779extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, 2780 s64 nclusters, unsigned int flags); 2781extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); 2782extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, 2783 ext4_group_t block_group, 2784 struct buffer_head ** bh); 2785extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb, 2786 ext4_group_t group); 2787extern int ext4_should_retry_alloc(struct super_block *sb, int *retries); 2788 2789extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb, 2790 ext4_group_t block_group, 2791 bool ignore_locked); 2792extern int ext4_wait_block_bitmap(struct super_block *sb, 2793 ext4_group_t block_group, 2794 struct buffer_head *bh); 2795extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb, 2796 ext4_group_t block_group); 2797extern unsigned ext4_free_clusters_after_init(struct super_block *sb, 2798 ext4_group_t block_group, 2799 struct ext4_group_desc *gdp); 2800ext4_fsblk_t ext4_inode_to_goal_block(struct inode *); 2801 2802#if IS_ENABLED(CONFIG_UNICODE) 2803extern int ext4_fname_setup_ci_filename(struct inode *dir, 2804 const struct qstr *iname, 2805 struct ext4_filename *fname); 2806 2807static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2808{ 2809 kfree(fname->cf_name.name); 2810 fname->cf_name.name = NULL; 2811} 2812#else 2813static inline int ext4_fname_setup_ci_filename(struct inode *dir, 2814 const struct qstr *iname, 2815 struct ext4_filename *fname) 2816{ 2817 return 0; 2818} 2819 2820static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname) 2821{ 2822} 2823#endif 2824 2825/* ext4 encryption related stuff goes here crypto.c */ 2826#ifdef CONFIG_FS_ENCRYPTION 2827extern const struct fscrypt_operations ext4_cryptops; 2828 2829int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname, 2830 int lookup, struct ext4_filename *fname); 2831 2832int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry, 2833 struct ext4_filename *fname); 2834 2835void ext4_fname_free_filename(struct ext4_filename *fname); 2836 2837int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg); 2838 2839#else /* !CONFIG_FS_ENCRYPTION */ 2840static inline int ext4_fname_setup_filename(struct inode *dir, 2841 const struct qstr *iname, 2842 int lookup, 2843 struct ext4_filename *fname) 2844{ 2845 fname->usr_fname = iname; 2846 fname->disk_name.name = (unsigned char *) iname->name; 2847 fname->disk_name.len = iname->len; 2848 2849 return ext4_fname_setup_ci_filename(dir, iname, fname); 2850} 2851 2852static inline int ext4_fname_prepare_lookup(struct inode *dir, 2853 struct dentry *dentry, 2854 struct ext4_filename *fname) 2855{ 2856 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname); 2857} 2858 2859static inline void ext4_fname_free_filename(struct ext4_filename *fname) 2860{ 2861 ext4_fname_free_ci_filename(fname); 2862} 2863 2864static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp, 2865 void __user *arg) 2866{ 2867 return -EOPNOTSUPP; 2868} 2869#endif /* !CONFIG_FS_ENCRYPTION */ 2870 2871/* dir.c */ 2872extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *, 2873 struct file *, 2874 struct ext4_dir_entry_2 *, 2875 struct buffer_head *, char *, int, 2876 unsigned int); 2877#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \ 2878 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \ 2879 (de), (bh), (buf), (size), (offset))) 2880extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash, 2881 __u32 minor_hash, 2882 struct ext4_dir_entry_2 *dirent, 2883 struct fscrypt_str *ent_name); 2884extern void ext4_htree_free_dir_info(struct dir_private_info *p); 2885extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh, 2886 void *buf, int buf_size, 2887 struct ext4_filename *fname, 2888 struct ext4_dir_entry_2 **dest_de); 2889void ext4_insert_dentry(struct inode *dir, struct inode *inode, 2890 struct ext4_dir_entry_2 *de, 2891 int buf_size, 2892 struct ext4_filename *fname); 2893static inline void ext4_update_dx_flag(struct inode *inode) 2894{ 2895 if (!ext4_has_feature_dir_index(inode->i_sb) && 2896 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) { 2897 /* ext4_iget() should have caught this... */ 2898 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb)); 2899 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX); 2900 } 2901} 2902static const unsigned char ext4_filetype_table[] = { 2903 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK 2904}; 2905 2906static inline unsigned char get_dtype(struct super_block *sb, int filetype) 2907{ 2908 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX) 2909 return DT_UNKNOWN; 2910 2911 return ext4_filetype_table[filetype]; 2912} 2913extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh, 2914 void *buf, int buf_size); 2915 2916/* fsync.c */ 2917extern int ext4_sync_file(struct file *, loff_t, loff_t, int); 2918 2919/* hash.c */ 2920extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len, 2921 struct dx_hash_info *hinfo); 2922 2923/* ialloc.c */ 2924extern int ext4_mark_inode_used(struct super_block *sb, int ino); 2925extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *, 2926 struct inode *, umode_t, 2927 const struct qstr *qstr, __u32 goal, 2928 uid_t *owner, __u32 i_flags, 2929 int handle_type, unsigned int line_no, 2930 int nblocks); 2931 2932#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \ 2933 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \ 2934 (goal), (owner), i_flags, 0, 0, 0) 2935#define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \ 2936 type, nblocks) \ 2937 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \ 2938 0, (type), __LINE__, (nblocks)) 2939 2940 2941extern void ext4_free_inode(handle_t *, struct inode *); 2942extern struct inode * ext4_orphan_get(struct super_block *, unsigned long); 2943extern unsigned long ext4_count_free_inodes(struct super_block *); 2944extern unsigned long ext4_count_dirs(struct super_block *); 2945extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap); 2946extern int ext4_init_inode_table(struct super_block *sb, 2947 ext4_group_t group, int barrier); 2948extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate); 2949 2950/* fast_commit.c */ 2951int ext4_fc_info_show(struct seq_file *seq, void *v); 2952void ext4_fc_init(struct super_block *sb, journal_t *journal); 2953void ext4_fc_init_inode(struct inode *inode); 2954void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start, 2955 ext4_lblk_t end); 2956void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode, 2957 struct dentry *dentry); 2958void __ext4_fc_track_link(handle_t *handle, struct inode *inode, 2959 struct dentry *dentry); 2960void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry); 2961void ext4_fc_track_link(handle_t *handle, struct dentry *dentry); 2962void __ext4_fc_track_create(handle_t *handle, struct inode *inode, 2963 struct dentry *dentry); 2964void ext4_fc_track_create(handle_t *handle, struct dentry *dentry); 2965void ext4_fc_track_inode(handle_t *handle, struct inode *inode); 2966void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle); 2967void ext4_fc_del(struct inode *inode); 2968bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block); 2969void ext4_fc_replay_cleanup(struct super_block *sb); 2970int ext4_fc_commit(journal_t *journal, tid_t commit_tid); 2971int __init ext4_fc_init_dentry_cache(void); 2972void ext4_fc_destroy_dentry_cache(void); 2973int ext4_fc_record_regions(struct super_block *sb, int ino, 2974 ext4_lblk_t lblk, ext4_fsblk_t pblk, 2975 int len, int replay); 2976 2977/* mballoc.c */ 2978extern const struct seq_operations ext4_mb_seq_groups_ops; 2979extern const struct seq_operations ext4_mb_seq_structs_summary_ops; 2980extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset); 2981extern int ext4_mb_init(struct super_block *); 2982extern void ext4_mb_release(struct super_block *); 2983extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *, 2984 struct ext4_allocation_request *, int *); 2985extern void ext4_discard_preallocations(struct inode *); 2986extern int __init ext4_init_mballoc(void); 2987extern void ext4_exit_mballoc(void); 2988extern ext4_group_t ext4_mb_prefetch(struct super_block *sb, 2989 ext4_group_t group, 2990 unsigned int nr, int *cnt); 2991extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group, 2992 unsigned int nr); 2993 2994extern void ext4_free_blocks(handle_t *handle, struct inode *inode, 2995 struct buffer_head *bh, ext4_fsblk_t block, 2996 unsigned long count, int flags); 2997extern int ext4_mb_alloc_groupinfo(struct super_block *sb, 2998 ext4_group_t ngroups); 2999extern int ext4_mb_add_groupinfo(struct super_block *sb, 3000 ext4_group_t i, struct ext4_group_desc *desc); 3001extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, 3002 ext4_fsblk_t block, unsigned long count); 3003extern int ext4_trim_fs(struct super_block *, struct fstrim_range *); 3004extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid); 3005extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block, 3006 int len, bool state); 3007static inline bool ext4_mb_cr_expensive(enum criteria cr) 3008{ 3009 return cr >= CR_GOAL_LEN_SLOW; 3010} 3011 3012/* inode.c */ 3013void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw, 3014 struct ext4_inode_info *ei); 3015int ext4_inode_is_fast_symlink(struct inode *inode); 3016void ext4_check_map_extents_env(struct inode *inode); 3017struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int); 3018struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int); 3019int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count, 3020 bool wait, struct buffer_head **bhs); 3021int ext4_get_block_unwritten(struct inode *inode, sector_t iblock, 3022 struct buffer_head *bh_result, int create); 3023int ext4_get_block(struct inode *inode, sector_t iblock, 3024 struct buffer_head *bh_result, int create); 3025int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, 3026 struct buffer_head *bh, int create); 3027int ext4_walk_page_buffers(handle_t *handle, 3028 struct inode *inode, 3029 struct buffer_head *head, 3030 unsigned from, 3031 unsigned to, 3032 int *partial, 3033 int (*fn)(handle_t *handle, struct inode *inode, 3034 struct buffer_head *bh)); 3035int do_journal_get_write_access(handle_t *handle, struct inode *inode, 3036 struct buffer_head *bh); 3037void ext4_set_inode_mapping_order(struct inode *inode); 3038#define FALL_BACK_TO_NONDELALLOC 1 3039#define CONVERT_INLINE_DATA 2 3040 3041typedef enum { 3042 EXT4_IGET_NORMAL = 0, 3043 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */ 3044 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */ 3045 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */ 3046 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */ 3047} ext4_iget_flags; 3048 3049extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino, 3050 ext4_iget_flags flags, const char *function, 3051 unsigned int line); 3052 3053#define ext4_iget(sb, ino, flags) \ 3054 __ext4_iget((sb), (ino), (flags), __func__, __LINE__) 3055 3056extern int ext4_write_inode(struct inode *, struct writeback_control *); 3057extern int ext4_setattr(struct mnt_idmap *, struct dentry *, 3058 struct iattr *); 3059extern u32 ext4_dio_alignment(struct inode *inode); 3060extern int ext4_getattr(struct mnt_idmap *, const struct path *, 3061 struct kstat *, u32, unsigned int); 3062extern void ext4_evict_inode(struct inode *); 3063extern void ext4_clear_inode(struct inode *); 3064extern int ext4_file_getattr(struct mnt_idmap *, const struct path *, 3065 struct kstat *, u32, unsigned int); 3066extern void ext4_dirty_inode(struct inode *, int); 3067extern int ext4_change_inode_journal_flag(struct inode *, int); 3068extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *); 3069extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino, 3070 struct ext4_iloc *iloc); 3071extern int ext4_inode_attach_jinode(struct inode *inode); 3072extern int ext4_can_truncate(struct inode *inode); 3073extern int ext4_truncate(struct inode *); 3074extern int ext4_break_layouts(struct inode *); 3075extern int ext4_truncate_page_cache_block_range(struct inode *inode, 3076 loff_t start, loff_t end); 3077extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length); 3078extern void ext4_set_inode_flags(struct inode *, bool init); 3079extern int ext4_alloc_da_blocks(struct inode *inode); 3080extern void ext4_set_aops(struct inode *inode); 3081extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode); 3082extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks); 3083extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks); 3084extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks, 3085 int pextents); 3086extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, 3087 loff_t lstart, loff_t lend); 3088extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); 3089extern qsize_t *ext4_get_reserved_space(struct inode *inode); 3090extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); 3091extern void ext4_da_release_space(struct inode *inode, int to_free); 3092extern void ext4_da_update_reserve_space(struct inode *inode, 3093 int used, int quota_claim); 3094extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, 3095 ext4_fsblk_t pblk, ext4_lblk_t len); 3096 3097static inline bool is_special_ino(struct super_block *sb, unsigned long ino) 3098{ 3099 struct ext4_super_block *es = EXT4_SB(sb)->s_es; 3100 3101 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) || 3102 ino == le32_to_cpu(es->s_usr_quota_inum) || 3103 ino == le32_to_cpu(es->s_grp_quota_inum) || 3104 ino == le32_to_cpu(es->s_prj_quota_inum) || 3105 ino == le32_to_cpu(es->s_orphan_file_inum); 3106} 3107 3108/* indirect.c */ 3109extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, 3110 struct ext4_map_blocks *map, int flags); 3111extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks); 3112extern void ext4_ind_truncate(handle_t *, struct inode *inode); 3113extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode, 3114 ext4_lblk_t start, ext4_lblk_t end); 3115 3116/* ioctl.c */ 3117extern long ext4_ioctl(struct file *, unsigned int, unsigned long); 3118extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); 3119int ext4_fileattr_set(struct mnt_idmap *idmap, 3120 struct dentry *dentry, struct file_kattr *fa); 3121int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3122extern void ext4_reset_inode_seed(struct inode *inode); 3123int ext4_update_overhead(struct super_block *sb, bool force); 3124int ext4_force_shutdown(struct super_block *sb, u32 flags); 3125 3126/* migrate.c */ 3127extern int ext4_ext_migrate(struct inode *); 3128extern int ext4_ind_migrate(struct inode *inode); 3129 3130/* namei.c */ 3131extern int ext4_init_new_dir(handle_t *handle, struct inode *dir, 3132 struct inode *inode); 3133extern int ext4_dirblock_csum_verify(struct inode *inode, 3134 struct buffer_head *bh); 3135extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash, 3136 __u32 start_minor_hash, __u32 *next_hash); 3137extern int ext4_search_dir(struct buffer_head *bh, 3138 char *search_buf, 3139 int buf_size, 3140 struct inode *dir, 3141 struct ext4_filename *fname, 3142 unsigned int offset, 3143 struct ext4_dir_entry_2 **res_dir); 3144extern int ext4_generic_delete_entry(struct inode *dir, 3145 struct ext4_dir_entry_2 *de_del, 3146 struct buffer_head *bh, 3147 void *entry_buf, 3148 int buf_size, 3149 int csum_size); 3150extern bool ext4_empty_dir(struct inode *inode); 3151 3152/* resize.c */ 3153extern void ext4_kvfree_array_rcu(void *to_free); 3154extern int ext4_group_add(struct super_block *sb, 3155 struct ext4_new_group_data *input); 3156extern int ext4_group_extend(struct super_block *sb, 3157 struct ext4_super_block *es, 3158 ext4_fsblk_t n_blocks_count); 3159extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); 3160extern unsigned int ext4_list_backups(struct super_block *sb, 3161 unsigned int *three, unsigned int *five, 3162 unsigned int *seven); 3163 3164/* super.c */ 3165extern struct buffer_head *ext4_sb_bread(struct super_block *sb, 3166 sector_t block, blk_opf_t op_flags); 3167extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb, 3168 sector_t block); 3169extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb, 3170 sector_t block); 3171extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags, 3172 bh_end_io_t *end_io, bool simu_fail); 3173extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags, 3174 bh_end_io_t *end_io, bool simu_fail); 3175extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait); 3176extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block); 3177extern int ext4_seq_options_show(struct seq_file *seq, void *offset); 3178extern int ext4_calculate_overhead(struct super_block *sb); 3179extern __le32 ext4_superblock_csum(struct ext4_super_block *es); 3180extern void ext4_superblock_csum_set(struct super_block *sb); 3181extern int ext4_alloc_flex_bg_array(struct super_block *sb, 3182 ext4_group_t ngroup); 3183extern const char *ext4_decode_error(struct super_block *sb, int errno, 3184 char nbuf[16]); 3185extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb, 3186 ext4_group_t block_group, 3187 unsigned int flags); 3188extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb, 3189 ext4_group_t block_group); 3190 3191extern __printf(7, 8) 3192void __ext4_error(struct super_block *, const char *, unsigned int, bool, 3193 int, __u64, const char *, ...); 3194extern __printf(6, 7) 3195void __ext4_error_inode(struct inode *, const char *, unsigned int, 3196 ext4_fsblk_t, int, const char *, ...); 3197extern __printf(5, 6) 3198void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t, 3199 const char *, ...); 3200extern void __ext4_std_error(struct super_block *, const char *, 3201 unsigned int, int); 3202extern __printf(4, 5) 3203void __ext4_warning(struct super_block *, const char *, unsigned int, 3204 const char *, ...); 3205extern __printf(4, 5) 3206void __ext4_warning_inode(const struct inode *inode, const char *function, 3207 unsigned int line, const char *fmt, ...); 3208extern __printf(3, 4) 3209void __ext4_msg(struct super_block *, const char *, const char *, ...); 3210extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp, 3211 const char *, unsigned int, const char *); 3212extern __printf(7, 8) 3213void __ext4_grp_locked_error(const char *, unsigned int, 3214 struct super_block *, ext4_group_t, 3215 unsigned long, ext4_fsblk_t, 3216 const char *, ...); 3217 3218#define EXT4_ERROR_INODE(inode, fmt, a...) \ 3219 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a) 3220 3221#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \ 3222 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a) 3223 3224#define ext4_error_inode_block(inode, block, err, fmt, a...) \ 3225 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \ 3226 (fmt), ## a) 3227 3228#define EXT4_ERROR_FILE(file, block, fmt, a...) \ 3229 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a) 3230 3231#define ext4_abort(sb, err, fmt, a...) \ 3232 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a) 3233 3234#ifdef CONFIG_PRINTK 3235 3236#define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3237 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__) 3238#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3239 __ext4_error_inode((inode), (func), (line), (block), \ 3240 (err), (fmt), ##__VA_ARGS__) 3241#define ext4_error_file(file, func, line, block, fmt, ...) \ 3242 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__) 3243#define ext4_error(sb, fmt, ...) \ 3244 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \ 3245 ##__VA_ARGS__) 3246#define ext4_error_err(sb, err, fmt, ...) \ 3247 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \ 3248 ##__VA_ARGS__) 3249#define ext4_warning(sb, fmt, ...) \ 3250 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__) 3251#define ext4_warning_inode(inode, fmt, ...) \ 3252 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__) 3253#define ext4_msg(sb, level, fmt, ...) \ 3254 __ext4_msg(sb, level, fmt, ##__VA_ARGS__) 3255#define dump_mmp_msg(sb, mmp, msg) \ 3256 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg) 3257#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3258 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \ 3259 fmt, ##__VA_ARGS__) 3260 3261#else 3262 3263#define ext4_error_inode(inode, func, line, block, fmt, ...) \ 3264do { \ 3265 no_printk(fmt, ##__VA_ARGS__); \ 3266 __ext4_error_inode(inode, "", 0, block, 0, " "); \ 3267} while (0) 3268#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \ 3269do { \ 3270 no_printk(fmt, ##__VA_ARGS__); \ 3271 __ext4_error_inode(inode, "", 0, block, err, " "); \ 3272} while (0) 3273#define ext4_error_file(file, func, line, block, fmt, ...) \ 3274do { \ 3275 no_printk(fmt, ##__VA_ARGS__); \ 3276 __ext4_error_file(file, "", 0, block, " "); \ 3277} while (0) 3278#define ext4_error(sb, fmt, ...) \ 3279do { \ 3280 no_printk(fmt, ##__VA_ARGS__); \ 3281 __ext4_error(sb, "", 0, false, 0, 0, " "); \ 3282} while (0) 3283#define ext4_error_err(sb, err, fmt, ...) \ 3284do { \ 3285 no_printk(fmt, ##__VA_ARGS__); \ 3286 __ext4_error(sb, "", 0, false, err, 0, " "); \ 3287} while (0) 3288#define ext4_warning(sb, fmt, ...) \ 3289do { \ 3290 no_printk(fmt, ##__VA_ARGS__); \ 3291 __ext4_warning(sb, "", 0, " "); \ 3292} while (0) 3293#define ext4_warning_inode(inode, fmt, ...) \ 3294do { \ 3295 no_printk(fmt, ##__VA_ARGS__); \ 3296 __ext4_warning_inode(inode, "", 0, " "); \ 3297} while (0) 3298#define ext4_msg(sb, level, fmt, ...) \ 3299do { \ 3300 no_printk(fmt, ##__VA_ARGS__); \ 3301 __ext4_msg(sb, "", " "); \ 3302} while (0) 3303#define dump_mmp_msg(sb, mmp, msg) \ 3304 __dump_mmp_msg(sb, mmp, "", 0, "") 3305#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \ 3306do { \ 3307 no_printk(fmt, ##__VA_ARGS__); \ 3308 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \ 3309} while (0) 3310 3311#endif 3312 3313extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, 3314 struct ext4_group_desc *bg); 3315extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, 3316 struct ext4_group_desc *bg); 3317extern ext4_fsblk_t ext4_inode_table(struct super_block *sb, 3318 struct ext4_group_desc *bg); 3319extern __u32 ext4_free_group_clusters(struct super_block *sb, 3320 struct ext4_group_desc *bg); 3321extern __u32 ext4_free_inodes_count(struct super_block *sb, 3322 struct ext4_group_desc *bg); 3323extern __u32 ext4_used_dirs_count(struct super_block *sb, 3324 struct ext4_group_desc *bg); 3325extern __u32 ext4_itable_unused_count(struct super_block *sb, 3326 struct ext4_group_desc *bg); 3327extern void ext4_block_bitmap_set(struct super_block *sb, 3328 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3329extern void ext4_inode_bitmap_set(struct super_block *sb, 3330 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3331extern void ext4_inode_table_set(struct super_block *sb, 3332 struct ext4_group_desc *bg, ext4_fsblk_t blk); 3333extern void ext4_free_group_clusters_set(struct super_block *sb, 3334 struct ext4_group_desc *bg, 3335 __u32 count); 3336extern void ext4_free_inodes_set(struct super_block *sb, 3337 struct ext4_group_desc *bg, __u32 count); 3338extern void ext4_used_dirs_set(struct super_block *sb, 3339 struct ext4_group_desc *bg, __u32 count); 3340extern void ext4_itable_unused_set(struct super_block *sb, 3341 struct ext4_group_desc *bg, __u32 count); 3342extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group, 3343 struct ext4_group_desc *gdp); 3344extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group, 3345 struct ext4_group_desc *gdp); 3346extern int ext4_register_li_request(struct super_block *sb, 3347 ext4_group_t first_not_zeroed); 3348 3349static inline int ext4_has_group_desc_csum(struct super_block *sb) 3350{ 3351 return ext4_has_feature_gdt_csum(sb) || 3352 ext4_has_feature_metadata_csum(sb); 3353} 3354 3355#define ext4_read_incompat_64bit_val(es, name) \ 3356 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \ 3357 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \ 3358 le32_to_cpu(es->name##_lo)) 3359 3360static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es) 3361{ 3362 return ext4_read_incompat_64bit_val(es, s_blocks_count); 3363} 3364 3365static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es) 3366{ 3367 return ext4_read_incompat_64bit_val(es, s_r_blocks_count); 3368} 3369 3370static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es) 3371{ 3372 return ext4_read_incompat_64bit_val(es, s_free_blocks_count); 3373} 3374 3375static inline void ext4_blocks_count_set(struct ext4_super_block *es, 3376 ext4_fsblk_t blk) 3377{ 3378 es->s_blocks_count_lo = cpu_to_le32((u32)blk); 3379 es->s_blocks_count_hi = cpu_to_le32(blk >> 32); 3380} 3381 3382static inline void ext4_free_blocks_count_set(struct ext4_super_block *es, 3383 ext4_fsblk_t blk) 3384{ 3385 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk); 3386 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32); 3387} 3388 3389static inline void ext4_r_blocks_count_set(struct ext4_super_block *es, 3390 ext4_fsblk_t blk) 3391{ 3392 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk); 3393 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32); 3394} 3395 3396static inline loff_t ext4_isize(struct super_block *sb, 3397 struct ext4_inode *raw_inode) 3398{ 3399 if (ext4_has_feature_largedir(sb) || 3400 S_ISREG(le16_to_cpu(raw_inode->i_mode))) 3401 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) | 3402 le32_to_cpu(raw_inode->i_size_lo); 3403 3404 return (loff_t) le32_to_cpu(raw_inode->i_size_lo); 3405} 3406 3407static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size) 3408{ 3409 raw_inode->i_size_lo = cpu_to_le32(i_size); 3410 raw_inode->i_size_high = cpu_to_le32(i_size >> 32); 3411} 3412 3413/* 3414 * Reading s_groups_count requires using smp_rmb() afterwards. See 3415 * the locking protocol documented in the comments of ext4_group_add() 3416 * in resize.c 3417 */ 3418static inline ext4_group_t ext4_get_groups_count(struct super_block *sb) 3419{ 3420 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count; 3421 3422 smp_rmb(); 3423 return ngroups; 3424} 3425 3426static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi, 3427 ext4_group_t block_group) 3428{ 3429 return block_group >> sbi->s_log_groups_per_flex; 3430} 3431 3432static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi) 3433{ 3434 return 1 << sbi->s_log_groups_per_flex; 3435} 3436 3437static inline loff_t ext4_get_maxbytes(struct inode *inode) 3438{ 3439 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) 3440 return inode->i_sb->s_maxbytes; 3441 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; 3442} 3443 3444#define ext4_std_error(sb, errno) \ 3445do { \ 3446 if ((errno)) \ 3447 __ext4_std_error((sb), __func__, __LINE__, (errno)); \ 3448} while (0) 3449 3450#ifdef CONFIG_SMP 3451/* Each CPU can accumulate percpu_counter_batch clusters in their local 3452 * counters. So we need to make sure we have free clusters more 3453 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. 3454 */ 3455#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) 3456#else 3457#define EXT4_FREECLUSTERS_WATERMARK 0 3458#endif 3459 3460/* Update i_disksize. Requires i_rwsem to avoid races with truncate */ 3461static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) 3462{ 3463 WARN_ON_ONCE(S_ISREG(inode->i_mode) && 3464 !inode_is_locked(inode)); 3465 down_write(&EXT4_I(inode)->i_data_sem); 3466 if (newsize > EXT4_I(inode)->i_disksize) 3467 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize); 3468 up_write(&EXT4_I(inode)->i_data_sem); 3469} 3470 3471/* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */ 3472static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize) 3473{ 3474 int changed = 0; 3475 3476 if (newsize > inode->i_size) { 3477 i_size_write(inode, newsize); 3478 changed = 1; 3479 } 3480 if (newsize > EXT4_I(inode)->i_disksize) { 3481 ext4_update_i_disksize(inode, newsize); 3482 changed |= 2; 3483 } 3484 return changed; 3485} 3486 3487int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset, 3488 loff_t len); 3489 3490struct ext4_group_info { 3491 unsigned long bb_state; 3492#ifdef AGGRESSIVE_CHECK 3493 unsigned long bb_check_counter; 3494#endif 3495 struct rb_root bb_free_root; 3496 ext4_grpblk_t bb_first_free; /* first free block */ 3497 ext4_grpblk_t bb_free; /* total free blocks */ 3498 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */ 3499 int bb_avg_fragment_size_order; /* order of average 3500 fragment in BG */ 3501 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */ 3502 ext4_group_t bb_group; /* Group number */ 3503 struct list_head bb_prealloc_list; 3504#ifdef DOUBLE_CHECK 3505 void *bb_bitmap; 3506#endif 3507 struct rw_semaphore alloc_sem; 3508 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block 3509 * regions, index is order. 3510 * bb_counters[3] = 5 means 3511 * 5 free 8-block regions. */ 3512}; 3513 3514#define EXT4_GROUP_INFO_NEED_INIT_BIT 0 3515#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1 3516#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2 3517#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3 3518#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \ 3519 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT) 3520#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \ 3521 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT) 3522#define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4 3523 3524#define EXT4_MB_GRP_NEED_INIT(grp) \ 3525 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state))) 3526#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \ 3527 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3528#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \ 3529 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state))) 3530 3531#define EXT4_MB_GRP_WAS_TRIMMED(grp) \ 3532 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3533#define EXT4_MB_GRP_SET_TRIMMED(grp) \ 3534 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3535#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \ 3536 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state))) 3537#define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \ 3538 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state))) 3539 3540#define EXT4_MAX_CONTENTION 8 3541#define EXT4_CONTENTION_THRESHOLD 2 3542 3543static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb, 3544 ext4_group_t group) 3545{ 3546 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group); 3547} 3548 3549/* 3550 * Returns true if the filesystem is busy enough that attempts to 3551 * access the block group locks has run into contention. 3552 */ 3553static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi) 3554{ 3555 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD); 3556} 3557 3558static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group) 3559{ 3560 if (!spin_trylock(ext4_group_lock_ptr(sb, group))) 3561 return false; 3562 /* 3563 * We're able to grab the lock right away, so drop the lock 3564 * contention counter. 3565 */ 3566 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0); 3567 return true; 3568} 3569 3570static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group) 3571{ 3572 if (!ext4_try_lock_group(sb, group)) { 3573 /* 3574 * The lock is busy, so bump the contention counter, 3575 * and then wait on the spin lock. 3576 */ 3577 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1, 3578 EXT4_MAX_CONTENTION); 3579 spin_lock(ext4_group_lock_ptr(sb, group)); 3580 } 3581} 3582 3583static inline void ext4_unlock_group(struct super_block *sb, 3584 ext4_group_t group) 3585{ 3586 spin_unlock(ext4_group_lock_ptr(sb, group)); 3587} 3588 3589#ifdef CONFIG_QUOTA 3590static inline bool ext4_quota_capable(struct super_block *sb) 3591{ 3592 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb)); 3593} 3594 3595static inline bool ext4_is_quota_journalled(struct super_block *sb) 3596{ 3597 struct ext4_sb_info *sbi = EXT4_SB(sb); 3598 3599 return (ext4_has_feature_quota(sb) || 3600 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]); 3601} 3602int ext4_enable_quotas(struct super_block *sb); 3603#endif 3604 3605/* 3606 * Block validity checking 3607 */ 3608#define ext4_check_indirect_blockref(inode, bh) \ 3609 ext4_check_blockref(__func__, __LINE__, inode, \ 3610 (__le32 *)(bh)->b_data, \ 3611 EXT4_ADDR_PER_BLOCK((inode)->i_sb)) 3612 3613#define ext4_ind_check_inode(inode) \ 3614 ext4_check_blockref(__func__, __LINE__, inode, \ 3615 EXT4_I(inode)->i_data, \ 3616 EXT4_NDIR_BLOCKS) 3617 3618/* 3619 * Inodes and files operations 3620 */ 3621 3622/* dir.c */ 3623extern const struct file_operations ext4_dir_operations; 3624 3625/* file.c */ 3626extern const struct inode_operations ext4_file_inode_operations; 3627extern const struct file_operations ext4_file_operations; 3628extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin); 3629 3630/* inline.c */ 3631extern int ext4_get_max_inline_size(struct inode *inode); 3632extern int ext4_find_inline_data_nolock(struct inode *inode); 3633extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode); 3634extern void ext4_update_final_de(void *de_buf, int old_size, int new_size); 3635 3636int ext4_readpage_inline(struct inode *inode, struct folio *folio); 3637extern int ext4_try_to_write_inline_data(struct address_space *mapping, 3638 struct inode *inode, 3639 loff_t pos, unsigned len, 3640 struct folio **foliop); 3641int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len, 3642 unsigned copied, struct folio *folio); 3643extern int ext4_generic_write_inline_data(struct address_space *mapping, 3644 struct inode *inode, 3645 loff_t pos, unsigned len, 3646 struct folio **foliop, 3647 void **fsdata, bool da); 3648extern int ext4_try_add_inline_entry(handle_t *handle, 3649 struct ext4_filename *fname, 3650 struct inode *dir, struct inode *inode); 3651extern int ext4_try_create_inline_dir(handle_t *handle, 3652 struct inode *parent, 3653 struct inode *inode); 3654extern int ext4_read_inline_dir(struct file *filp, 3655 struct dir_context *ctx, 3656 int *has_inline_data); 3657extern int ext4_inlinedir_to_tree(struct file *dir_file, 3658 struct inode *dir, ext4_lblk_t block, 3659 struct dx_hash_info *hinfo, 3660 __u32 start_hash, __u32 start_minor_hash, 3661 int *has_inline_data); 3662extern struct buffer_head *ext4_find_inline_entry(struct inode *dir, 3663 struct ext4_filename *fname, 3664 struct ext4_dir_entry_2 **res_dir, 3665 int *has_inline_data); 3666extern int ext4_delete_inline_entry(handle_t *handle, 3667 struct inode *dir, 3668 struct ext4_dir_entry_2 *de_del, 3669 struct buffer_head *bh, 3670 int *has_inline_data); 3671extern bool empty_inline_dir(struct inode *dir, int *has_inline_data); 3672extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, 3673 struct ext4_dir_entry_2 **parent_de, 3674 int *retval); 3675extern void *ext4_read_inline_link(struct inode *inode); 3676 3677struct iomap; 3678extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap); 3679 3680extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline); 3681 3682extern int ext4_convert_inline_data(struct inode *inode); 3683 3684static inline int ext4_has_inline_data(struct inode *inode) 3685{ 3686 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) && 3687 EXT4_I(inode)->i_inline_off; 3688} 3689 3690/* namei.c */ 3691extern const struct inode_operations ext4_dir_inode_operations; 3692extern const struct inode_operations ext4_special_inode_operations; 3693extern struct dentry *ext4_get_parent(struct dentry *child); 3694extern int ext4_init_dirblock(handle_t *handle, struct inode *inode, 3695 struct buffer_head *dir_block, 3696 unsigned int parent_ino, void *inline_buf, 3697 int inline_size); 3698extern void ext4_initialize_dirent_tail(struct buffer_head *bh, 3699 unsigned int blocksize); 3700extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode, 3701 struct buffer_head *bh); 3702extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name, 3703 struct inode *inode, struct dentry *dentry); 3704extern int __ext4_link(struct inode *dir, struct inode *inode, 3705 struct dentry *dentry); 3706 3707#define S_SHIFT 12 3708static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = { 3709 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE, 3710 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR, 3711 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV, 3712 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV, 3713 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO, 3714 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK, 3715 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK, 3716}; 3717 3718static inline void ext4_set_de_type(struct super_block *sb, 3719 struct ext4_dir_entry_2 *de, 3720 umode_t mode) { 3721 if (ext4_has_feature_filetype(sb)) 3722 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT]; 3723} 3724 3725/* readpages.c */ 3726extern int ext4_mpage_readpages(struct inode *inode, 3727 struct readahead_control *rac, struct folio *folio); 3728extern int __init ext4_init_post_read_processing(void); 3729extern void ext4_exit_post_read_processing(void); 3730 3731/* symlink.c */ 3732extern const struct inode_operations ext4_encrypted_symlink_inode_operations; 3733extern const struct inode_operations ext4_symlink_inode_operations; 3734extern const struct inode_operations ext4_fast_symlink_inode_operations; 3735 3736/* sysfs.c */ 3737extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi); 3738extern int ext4_register_sysfs(struct super_block *sb); 3739extern void ext4_unregister_sysfs(struct super_block *sb); 3740extern int __init ext4_init_sysfs(void); 3741extern void ext4_exit_sysfs(void); 3742 3743/* block_validity */ 3744extern void ext4_release_system_zone(struct super_block *sb); 3745extern int ext4_setup_system_zone(struct super_block *sb); 3746extern int __init ext4_init_system_zone(void); 3747extern void ext4_exit_system_zone(void); 3748extern int ext4_inode_block_valid(struct inode *inode, 3749 ext4_fsblk_t start_blk, 3750 unsigned int count); 3751extern int ext4_check_blockref(const char *, unsigned int, 3752 struct inode *, __le32 *, unsigned int); 3753extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode, 3754 ext4_fsblk_t start_blk, unsigned int count); 3755 3756 3757/* extents.c */ 3758struct ext4_ext_path; 3759struct ext4_extent; 3760 3761/* 3762 * Maximum number of logical blocks in a file; ext4_extent's ee_block is 3763 * __le32. 3764 */ 3765#define EXT_MAX_BLOCKS 0xffffffff 3766 3767extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode); 3768extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents); 3769extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, 3770 struct ext4_map_blocks *map, int flags); 3771extern int ext4_ext_truncate(handle_t *, struct inode *); 3772extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start, 3773 ext4_lblk_t end); 3774extern void ext4_ext_init(struct super_block *); 3775extern void ext4_ext_release(struct super_block *); 3776extern long ext4_fallocate(struct file *file, int mode, loff_t offset, 3777 loff_t len); 3778extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode, 3779 loff_t offset, ssize_t len); 3780extern int ext4_convert_unwritten_extents_atomic(handle_t *handle, 3781 struct inode *inode, loff_t offset, ssize_t len); 3782extern int ext4_convert_unwritten_io_end_vec(handle_t *handle, 3783 ext4_io_end_t *io_end); 3784extern int ext4_map_blocks(handle_t *handle, struct inode *inode, 3785 struct ext4_map_blocks *map, int flags); 3786extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode, 3787 int num, 3788 struct ext4_ext_path *path); 3789extern struct ext4_ext_path *ext4_ext_insert_extent( 3790 handle_t *handle, struct inode *inode, 3791 struct ext4_ext_path *path, 3792 struct ext4_extent *newext, int gb_flags); 3793extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t, 3794 struct ext4_ext_path *, 3795 int flags); 3796extern void ext4_free_ext_path(struct ext4_ext_path *); 3797extern int ext4_ext_check_inode(struct inode *inode); 3798extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path); 3799extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3800 __u64 start, __u64 len); 3801extern int ext4_get_es_cache(struct inode *inode, 3802 struct fiemap_extent_info *fieinfo, 3803 __u64 start, __u64 len); 3804extern int ext4_ext_precache(struct inode *inode); 3805extern int ext4_swap_extents(handle_t *handle, struct inode *inode1, 3806 struct inode *inode2, ext4_lblk_t lblk1, 3807 ext4_lblk_t lblk2, ext4_lblk_t count, 3808 int mark_unwritten,int *err); 3809extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu); 3810extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode, 3811 int check_cred, int restart_cred, 3812 int revoke_cred); 3813extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end); 3814extern int ext4_ext_replay_set_iblocks(struct inode *inode); 3815extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start, 3816 int len, int unwritten, ext4_fsblk_t pblk); 3817extern int ext4_ext_clear_bb(struct inode *inode); 3818 3819 3820/* move_extent.c */ 3821extern void ext4_double_down_write_data_sem(struct inode *first, 3822 struct inode *second); 3823extern void ext4_double_up_write_data_sem(struct inode *orig_inode, 3824 struct inode *donor_inode); 3825extern int ext4_move_extents(struct file *o_filp, struct file *d_filp, 3826 __u64 start_orig, __u64 start_donor, 3827 __u64 len, __u64 *moved_len); 3828 3829/* page-io.c */ 3830extern int __init ext4_init_pageio(void); 3831extern void ext4_exit_pageio(void); 3832extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags); 3833extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end); 3834extern int ext4_put_io_end(ext4_io_end_t *io_end); 3835extern void ext4_put_io_end_defer(ext4_io_end_t *io_end); 3836extern void ext4_io_submit_init(struct ext4_io_submit *io, 3837 struct writeback_control *wbc); 3838extern void ext4_end_io_rsv_work(struct work_struct *work); 3839extern void ext4_io_submit(struct ext4_io_submit *io); 3840int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page, 3841 size_t len); 3842extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end); 3843extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end); 3844 3845/* mmp.c */ 3846extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); 3847 3848/* mmp.c */ 3849extern void ext4_stop_mmpd(struct ext4_sb_info *sbi); 3850 3851/* verity.c */ 3852extern const struct fsverity_operations ext4_verityops; 3853 3854/* orphan.c */ 3855extern int ext4_orphan_add(handle_t *, struct inode *); 3856extern int ext4_orphan_del(handle_t *, struct inode *); 3857extern void ext4_orphan_cleanup(struct super_block *sb, 3858 struct ext4_super_block *es); 3859extern void ext4_release_orphan_info(struct super_block *sb); 3860extern int ext4_init_orphan_info(struct super_block *sb); 3861extern int ext4_orphan_file_empty(struct super_block *sb); 3862extern void ext4_orphan_file_block_trigger( 3863 struct jbd2_buffer_trigger_type *triggers, 3864 struct buffer_head *bh, 3865 void *data, size_t size); 3866 3867/* 3868 * Add new method to test whether block and inode bitmaps are properly 3869 * initialized. With uninit_bg reading the block from disk is not enough 3870 * to mark the bitmap uptodate. We need to also zero-out the bitmap 3871 */ 3872#define BH_BITMAP_UPTODATE BH_JBDPrivateStart 3873 3874static inline int bitmap_uptodate(struct buffer_head *bh) 3875{ 3876 return (buffer_uptodate(bh) && 3877 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state)); 3878} 3879static inline void set_bitmap_uptodate(struct buffer_head *bh) 3880{ 3881 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state); 3882} 3883 3884extern int ext4_resize_begin(struct super_block *sb); 3885extern int ext4_resize_end(struct super_block *sb, bool update_backups); 3886 3887static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end) 3888{ 3889 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) 3890 io_end->flag |= EXT4_IO_END_UNWRITTEN; 3891} 3892 3893static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end) 3894{ 3895 if (io_end->flag & EXT4_IO_END_UNWRITTEN) 3896 io_end->flag &= ~EXT4_IO_END_UNWRITTEN; 3897} 3898 3899extern const struct iomap_ops ext4_iomap_ops; 3900extern const struct iomap_ops ext4_iomap_overwrite_ops; 3901extern const struct iomap_ops ext4_iomap_report_ops; 3902 3903static inline int ext4_buffer_uptodate(struct buffer_head *bh) 3904{ 3905 /* 3906 * If the buffer has the write error flag, we have failed 3907 * to write out data in the block. In this case, we don't 3908 * have to read the block because we may read the old data 3909 * successfully. 3910 */ 3911 if (buffer_write_io_error(bh)) 3912 set_buffer_uptodate(bh); 3913 return buffer_uptodate(bh); 3914} 3915 3916static inline bool ext4_inode_can_atomic_write(struct inode *inode) 3917{ 3918 3919 return S_ISREG(inode->i_mode) && 3920 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) && 3921 EXT4_SB(inode->i_sb)->s_awu_min > 0; 3922} 3923 3924extern int ext4_block_write_begin(handle_t *handle, struct folio *folio, 3925 loff_t pos, unsigned len, 3926 get_block_t *get_block); 3927#endif /* __KERNEL__ */ 3928 3929#define EFSBADCRC EBADMSG /* Bad CRC detected */ 3930#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 3931 3932#endif /* _EXT4_H */