fs/ext4/ext4.h

   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/types.h>
  21 #include <linux/blkdev.h>
  22 #include <linux/magic.h>
  23 #include <linux/jbd2.h>
  24 #include <linux/quota.h>
  25 #include <linux/rwsem.h>
  26 #include <linux/rbtree.h>
  27 #include <linux/seqlock.h>
  28 #include <linux/mutex.h>
  29 #include <linux/timer.h>
  30 #include <linux/version.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 <crypto/hash.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
  44 #include <linux/fscrypt.h>
  45 #include <linux/fsverity.h>
  46
  47 #include <linux/compiler.h>
  48
  49 /*
  50  * The fourth extended filesystem constants/structures
  51  */
  52
  53 /*
  54  * with AGGRESSIVE_CHECK allocator runs consistency checks over
  55  * structures. these checks slow things down a lot
  56  */
  57 #define AGGRESSIVE_CHECK__
  58
  59 /*
  60  * with DOUBLE_CHECK defined mballoc creates persistent in-core
  61  * bitmaps, maintains and uses them to check for double allocations
  62  */
  63 #define DOUBLE_CHECK__
  64
  65 /*
  66  * Define EXT4FS_DEBUG to produce debug messages
  67  */
  68 #undef EXT4FS_DEBUG
  69
  70 /*
  71  * Debug code
  72  */
  73 #ifdef EXT4FS_DEBUG
  74 #define ext4_debug(f, a...)                                             \
  75         do {                                                            \
  76                 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
  77                         __FILE__, __LINE__, __func__);                  \
  78                 printk(KERN_DEBUG f, ## a);                             \
  79         } while (0)
  80 #else
  81 #define ext4_debug(fmt, ...)    no_printk(fmt, ##__VA_ARGS__)
  82 #endif
  83
  84  /*
  85   * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
  86   */
  87 #define EXT_DEBUG__
  88
  89 /*
  90  * Dynamic printk for controlled extents debugging.
  91  */
  92 #ifdef CONFIG_EXT4_DEBUG
  93 #define ext_debug(ino, fmt, ...)                                        \
  94         pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt,    \
  95                  current->comm, task_pid_nr(current),                   \
  96                  ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__,       \
  97                  __func__, ##__VA_ARGS__)
  98 #else
  99 #define ext_debug(ino, fmt, ...)        no_printk(fmt, ##__VA_ARGS__)
 100 #endif
 101
 102 /* data type for block offset of block group */
 103 typedef int ext4_grpblk_t;
 104
 105 /* data type for filesystem-wide blocks number */
 106 typedef unsigned long long ext4_fsblk_t;
 107
 108 /* data type for file logical block number */
 109 typedef __u32 ext4_lblk_t;
 110
 111 /* data type for block group number */
 112 typedef unsigned int ext4_group_t;
 113
 114 enum SHIFT_DIRECTION {
 115         SHIFT_LEFT = 0,
 116         SHIFT_RIGHT,
 117 };
 118
 119 /*
 120  * Flags used in mballoc's allocation_context flags field.
 121  *
 122  * Also used to show what's going on for debugging purposes when the
 123  * flag field is exported via the traceport interface
 124  */
 125
 126 /* prefer goal again. length */
 127 #define EXT4_MB_HINT_MERGE              0x0001
 128 /* blocks already reserved */
 129 #define EXT4_MB_HINT_RESERVED           0x0002
 130 /* metadata is being allocated */
 131 #define EXT4_MB_HINT_METADATA           0x0004
 132 /* first blocks in the file */
 133 #define EXT4_MB_HINT_FIRST              0x0008
 134 /* search for the best chunk */
 135 #define EXT4_MB_HINT_BEST               0x0010
 136 /* data is being allocated */
 137 #define EXT4_MB_HINT_DATA               0x0020
 138 /* don't preallocate (for tails) */
 139 #define EXT4_MB_HINT_NOPREALLOC         0x0040
 140 /* allocate for locality group */
 141 #define EXT4_MB_HINT_GROUP_ALLOC        0x0080
 142 /* allocate goal blocks or none */
 143 #define EXT4_MB_HINT_GOAL_ONLY          0x0100
 144 /* goal is meaningful */
 145 #define EXT4_MB_HINT_TRY_GOAL           0x0200
 146 /* blocks already pre-reserved by delayed allocation */
 147 #define EXT4_MB_DELALLOC_RESERVED       0x0400
 148 /* We are doing stream allocation */
 149 #define EXT4_MB_STREAM_ALLOC            0x0800
 150 /* Use reserved root blocks if needed */
 151 #define EXT4_MB_USE_ROOT_BLOCKS         0x1000
 152 /* Use blocks from reserved pool */
 153 #define EXT4_MB_USE_RESERVED            0x2000
 154 /* Do strict check for free blocks while retrying block allocation */
 155 #define EXT4_MB_STRICT_CHECK            0x4000
 156
 157 struct ext4_allocation_request {
 158         /* target inode for block we're allocating */
 159         struct inode *inode;
 160         /* how many blocks we want to allocate */
 161         unsigned int len;
 162         /* logical block in target inode */
 163         ext4_lblk_t logical;
 164         /* the closest logical allocated block to the left */
 165         ext4_lblk_t lleft;
 166         /* the closest logical allocated block to the right */
 167         ext4_lblk_t lright;
 168         /* phys. target (a hint) */
 169         ext4_fsblk_t goal;
 170         /* phys. block for the closest logical allocated block to the left */
 171         ext4_fsblk_t pleft;
 172         /* phys. block for the closest logical allocated block to the right */
 173         ext4_fsblk_t pright;
 174         /* flags. see above EXT4_MB_HINT_* */
 175         unsigned int flags;
 176 };
 177
 178 /*
 179  * Logical to physical block mapping, used by ext4_map_blocks()
 180  *
 181  * This structure is used to pass requests into ext4_map_blocks() as
 182  * well as to store the information returned by ext4_map_blocks().  It
 183  * takes less room on the stack than a struct buffer_head.
 184  */
 185 #define EXT4_MAP_NEW            BIT(BH_New)
 186 #define EXT4_MAP_MAPPED         BIT(BH_Mapped)
 187 #define EXT4_MAP_UNWRITTEN      BIT(BH_Unwritten)
 188 #define EXT4_MAP_BOUNDARY       BIT(BH_Boundary)
 189 #define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
 190                                  EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
 191
 192 struct ext4_map_blocks {
 193         ext4_fsblk_t m_pblk;
 194         ext4_lblk_t m_lblk;
 195         unsigned int m_len;
 196         unsigned int m_flags;
 197 };
 198
 199 /*
 200  * Block validity checking, system zone rbtree.
 201  */
 202 struct ext4_system_blocks {
 203         struct rb_root root;
 204         struct rcu_head rcu;
 205 };
 206
 207 /*
 208  * Flags for ext4_io_end->flags
 209  */
 210 #define EXT4_IO_END_UNWRITTEN   0x0001
 211
 212 struct ext4_io_end_vec {
 213         struct list_head list;          /* list of io_end_vec */
 214         loff_t offset;                  /* offset in the file */
 215         ssize_t size;                   /* size of the extent */
 216 };
 217
 218 /*
 219  * For converting unwritten extents on a work queue. 'handle' is used for
 220  * buffered writeback.
 221  */
 222 typedef struct ext4_io_end {
 223         struct list_head        list;           /* per-file finished IO list */
 224         handle_t                *handle;        /* handle reserved for extent
 225                                                  * conversion */
 226         struct inode            *inode;         /* file being written to */
 227         struct bio              *bio;           /* Linked list of completed
 228                                                  * bios covering the extent */
 229         unsigned int            flag;           /* unwritten or not */
 230         atomic_t                count;          /* reference counter */
 231         struct list_head        list_vec;       /* list of ext4_io_end_vec */
 232 } ext4_io_end_t;
 233
 234 struct ext4_io_submit {
 235         struct writeback_control *io_wbc;
 236         struct bio              *io_bio;
 237         ext4_io_end_t           *io_end;
 238         sector_t                io_next_block;
 239 };
 240
 241 /*
 242  * Special inodes numbers
 243  */
 244 #define EXT4_BAD_INO             1      /* Bad blocks inode */
 245 #define EXT4_ROOT_INO            2      /* Root inode */
 246 #define EXT4_USR_QUOTA_INO       3      /* User quota inode */
 247 #define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
 248 #define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
 249 #define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
 250 #define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
 251 #define EXT4_JOURNAL_INO         8      /* Journal inode */
 252
 253 /* First non-reserved inode for old ext4 filesystems */
 254 #define EXT4_GOOD_OLD_FIRST_INO 11
 255
 256 /*
 257  * Maximal count of links to a file
 258  */
 259 #define EXT4_LINK_MAX           65000
 260
 261 /*
 262  * Macro-instructions used to manage several block sizes
 263  */
 264 #define EXT4_MIN_BLOCK_SIZE             1024
 265 #define EXT4_MAX_BLOCK_SIZE             65536
 266 #define EXT4_MIN_BLOCK_LOG_SIZE         10
 267 #define EXT4_MAX_BLOCK_LOG_SIZE         16
 268 #define EXT4_MAX_CLUSTER_LOG_SIZE       30
 269 #ifdef __KERNEL__
 270 # define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
 271 #else
 272 # define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
 273 #endif
 274 #define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
 275 #define EXT4_CLUSTER_SIZE(s)            (EXT4_BLOCK_SIZE(s) << \
 276                                          EXT4_SB(s)->s_cluster_bits)
 277 #ifdef __KERNEL__
 278 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
 279 # define EXT4_CLUSTER_BITS(s)           (EXT4_SB(s)->s_cluster_bits)
 280 #else
 281 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
 282 #endif
 283 #ifdef __KERNEL__
 284 #define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
 285 #define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
 286 #define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
 287 #else
 288 #define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 289                                  EXT4_GOOD_OLD_INODE_SIZE : \
 290                                  (s)->s_inode_size)
 291 #define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
 292                                  EXT4_GOOD_OLD_FIRST_INO : \
 293                                  (s)->s_first_ino)
 294 #endif
 295 #define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
 296 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
 297         ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
 298                                                                   blkbits))
 299
 300 /* Translate a block number to a cluster number */
 301 #define EXT4_B2C(sbi, blk)      ((blk) >> (sbi)->s_cluster_bits)
 302 /* Translate a cluster number to a block number */
 303 #define EXT4_C2B(sbi, cluster)  ((cluster) << (sbi)->s_cluster_bits)
 304 /* Translate # of blks to # of clusters */
 305 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
 306                                  (sbi)->s_cluster_bits)
 307 /* Mask out the low bits to get the starting block of the cluster */
 308 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &                              \
 309                                   ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
 310 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &                              \
 311                                   ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
 312 /* Fill in the low bits to get the last block of the cluster */
 313 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |                            \
 314                                     ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
 315 /* Get the cluster offset */
 316 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &                               \
 317                                  ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
 318 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &                               \
 319                                  ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
 320
 321 /*
 322  * Structure of a blocks group descriptor
 323  */
 324 struct ext4_group_desc
 325 {
 326         __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
 327         __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
 328         __le32  bg_inode_table_lo;      /* Inodes table block */
 329         __le16  bg_free_blocks_count_lo;/* Free blocks count */
 330         __le16  bg_free_inodes_count_lo;/* Free inodes count */
 331         __le16  bg_used_dirs_count_lo;  /* Directories count */
 332         __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
 333         __le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
 334         __le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
 335         __le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
 336         __le16  bg_itable_unused_lo;    /* Unused inodes count */
 337         __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
 338         __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
 339         __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
 340         __le32  bg_inode_table_hi;      /* Inodes table block MSB */
 341         __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
 342         __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
 343         __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
 344         __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
 345         __le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
 346         __le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
 347         __le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
 348         __u32   bg_reserved;
 349 };
 350
 351 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END        \
 352         (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
 353          sizeof(__le16))
 354 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END        \
 355         (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
 356          sizeof(__le16))
 357
 358 /*
 359  * Structure of a flex block group info
 360  */
 361
 362 struct flex_groups {
 363         atomic64_t      free_clusters;
 364         atomic_t        free_inodes;
 365         atomic_t        used_dirs;
 366 };
 367
 368 #define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
 369 #define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
 370 #define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
 371
 372 /*
 373  * Macro-instructions used to manage group descriptors
 374  */
 375 #define EXT4_MIN_DESC_SIZE              32
 376 #define EXT4_MIN_DESC_SIZE_64BIT        64
 377 #define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
 378 #define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
 379 #ifdef __KERNEL__
 380 # define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
 381 # define EXT4_CLUSTERS_PER_GROUP(s)     (EXT4_SB(s)->s_clusters_per_group)
 382 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
 383 # define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
 384 # define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
 385 #else
 386 # define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
 387 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
 388 # define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
 389 #endif
 390
 391 /*
 392  * Constants relative to the data blocks
 393  */
 394 #define EXT4_NDIR_BLOCKS                12
 395 #define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
 396 #define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
 397 #define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
 398 #define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
 399
 400 /*
 401  * Inode flags
 402  */
 403 #define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
 404 #define EXT4_UNRM_FL                    0x00000002 /* Undelete */
 405 #define EXT4_COMPR_FL                   0x00000004 /* Compress file */
 406 #define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
 407 #define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
 408 #define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
 409 #define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
 410 #define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
 411 /* Reserved for compression usage... */
 412 #define EXT4_DIRTY_FL                   0x00000100
 413 #define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
 414 #define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
 415         /* nb: was previously EXT2_ECOMPR_FL */
 416 #define EXT4_ENCRYPT_FL                 0x00000800 /* encrypted file */
 417 /* End compression flags --- maybe not all used */
 418 #define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
 419 #define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
 420 #define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
 421 #define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
 422 #define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
 423 #define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
 424 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
 425 #define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
 426 #define EXT4_VERITY_FL                  0x00100000 /* Verity protected inode */
 427 #define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
 428 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
 429
 430 #define EXT4_DAX_FL                     0x02000000 /* Inode is DAX */
 431
 432 #define EXT4_INLINE_DATA_FL             0x10000000 /* Inode has inline data. */
 433 #define EXT4_PROJINHERIT_FL             0x20000000 /* Create with parents projid */
 434 #define EXT4_CASEFOLD_FL                0x40000000 /* Casefolded directory */
 435 #define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
 436
 437 #define EXT4_FL_USER_VISIBLE            0x725BDFFF /* User visible flags */
 438 #define EXT4_FL_USER_MODIFIABLE         0x624BC0FF /* User modifiable flags */
 439
 440 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
 441 #define EXT4_FL_XFLAG_VISIBLE           (EXT4_SYNC_FL | \
 442                                          EXT4_IMMUTABLE_FL | \
 443                                          EXT4_APPEND_FL | \
 444                                          EXT4_NODUMP_FL | \
 445                                          EXT4_NOATIME_FL | \
 446                                          EXT4_PROJINHERIT_FL | \
 447                                          EXT4_DAX_FL)
 448
 449 /* Flags that should be inherited by new inodes from their parent. */
 450 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 451                            EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 452                            EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 453                            EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
 454                            EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
 455                            EXT4_DAX_FL)
 456
 457 /* Flags that are appropriate for regular files (all but dir-specific ones). */
 458 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
 459                            EXT4_PROJINHERIT_FL))
 460
 461 /* Flags that are appropriate for non-directories/regular files. */
 462 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 463
 464 /* The only flags that should be swapped */
 465 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
 466
 467 /* Flags which are mutually exclusive to DAX */
 468 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
 469                            EXT4_JOURNAL_DATA_FL)
 470
 471 /* Mask out flags that are inappropriate for the given type of inode. */
 472 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 473 {
 474         if (S_ISDIR(mode))
 475                 return flags;
 476         else if (S_ISREG(mode))
 477                 return flags & EXT4_REG_FLMASK;
 478         else
 479                 return flags & EXT4_OTHER_FLMASK;
 480 }
 481
 482 /*
 483  * Inode flags used for atomic set/get
 484  */
 485 enum {
 486         EXT4_INODE_SECRM        = 0,    /* Secure deletion */
 487         EXT4_INODE_UNRM         = 1,    /* Undelete */
 488         EXT4_INODE_COMPR        = 2,    /* Compress file */
 489         EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
 490         EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
 491         EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
 492         EXT4_INODE_NODUMP       = 6,    /* do not dump file */
 493         EXT4_INODE_NOATIME      = 7,    /* do not update atime */
 494 /* Reserved for compression usage... */
 495         EXT4_INODE_DIRTY        = 8,
 496         EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
 497         EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
 498         EXT4_INODE_ENCRYPT      = 11,   /* Encrypted file */
 499 /* End compression flags --- maybe not all used */
 500         EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
 501         EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
 502         EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
 503         EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
 504         EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
 505         EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
 506         EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
 507         EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
 508         EXT4_INODE_VERITY       = 20,   /* Verity protected inode */
 509         EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
 510 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
 511         EXT4_INODE_DAX          = 25,   /* Inode is DAX */
 512         EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
 513         EXT4_INODE_PROJINHERIT  = 29,   /* Create with parents projid */
 514         EXT4_INODE_CASEFOLD     = 30,   /* Casefolded directory */
 515         EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
 516 };
 517
 518 /*
 519  * Since it's pretty easy to mix up bit numbers and hex values, we use a
 520  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
 521  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
 522  * any extra space in the compiled kernel image, otherwise, the build will fail.
 523  * It's important that these values are the same, since we are using
 524  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
 525  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
 526  * values found in ext2, ext3 and ext4 filesystems, and of course the values
 527  * defined in e2fsprogs.
 528  *
 529  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
 530  */
 531 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
 532 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
 533
 534 static inline void ext4_check_flag_values(void)
 535 {
 536         CHECK_FLAG_VALUE(SECRM);
 537         CHECK_FLAG_VALUE(UNRM);
 538         CHECK_FLAG_VALUE(COMPR);
 539         CHECK_FLAG_VALUE(SYNC);
 540         CHECK_FLAG_VALUE(IMMUTABLE);
 541         CHECK_FLAG_VALUE(APPEND);
 542         CHECK_FLAG_VALUE(NODUMP);
 543         CHECK_FLAG_VALUE(NOATIME);
 544         CHECK_FLAG_VALUE(DIRTY);
 545         CHECK_FLAG_VALUE(COMPRBLK);
 546         CHECK_FLAG_VALUE(NOCOMPR);
 547         CHECK_FLAG_VALUE(ENCRYPT);
 548         CHECK_FLAG_VALUE(INDEX);
 549         CHECK_FLAG_VALUE(IMAGIC);
 550         CHECK_FLAG_VALUE(JOURNAL_DATA);
 551         CHECK_FLAG_VALUE(NOTAIL);
 552         CHECK_FLAG_VALUE(DIRSYNC);
 553         CHECK_FLAG_VALUE(TOPDIR);
 554         CHECK_FLAG_VALUE(HUGE_FILE);
 555         CHECK_FLAG_VALUE(EXTENTS);
 556         CHECK_FLAG_VALUE(VERITY);
 557         CHECK_FLAG_VALUE(EA_INODE);
 558         CHECK_FLAG_VALUE(INLINE_DATA);
 559         CHECK_FLAG_VALUE(PROJINHERIT);
 560         CHECK_FLAG_VALUE(CASEFOLD);
 561         CHECK_FLAG_VALUE(RESERVED);
 562 }
 563
 564 /* Used to pass group descriptor data when online resize is done */
 565 struct ext4_new_group_input {
 566         __u32 group;            /* Group number for this data */
 567         __u64 block_bitmap;     /* Absolute block number of block bitmap */
 568         __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
 569         __u64 inode_table;      /* Absolute block number of inode table start */
 570         __u32 blocks_count;     /* Total number of blocks in this group */
 571         __u16 reserved_blocks;  /* Number of reserved blocks in this group */
 572         __u16 unused;
 573 };
 574
 575 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 576 struct compat_ext4_new_group_input {
 577         u32 group;
 578         compat_u64 block_bitmap;
 579         compat_u64 inode_bitmap;
 580         compat_u64 inode_table;
 581         u32 blocks_count;
 582         u16 reserved_blocks;
 583         u16 unused;
 584 };
 585 #endif
 586
 587 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 588 struct ext4_new_group_data {
 589         __u32 group;
 590         __u64 block_bitmap;
 591         __u64 inode_bitmap;
 592         __u64 inode_table;
 593         __u32 blocks_count;
 594         __u16 reserved_blocks;
 595         __u16 mdata_blocks;
 596         __u32 free_clusters_count;
 597 };
 598
 599 /* Indexes used to index group tables in ext4_new_group_data */
 600 enum {
 601         BLOCK_BITMAP = 0,       /* block bitmap */
 602         INODE_BITMAP,           /* inode bitmap */
 603         INODE_TABLE,            /* inode tables */
 604         GROUP_TABLE_COUNT,
 605 };
 606
 607 /*
 608  * Flags used by ext4_map_blocks()
 609  */
 610         /* Allocate any needed blocks and/or convert an unwritten
 611            extent to be an initialized ext4 */
 612 #define EXT4_GET_BLOCKS_CREATE                  0x0001
 613         /* Request the creation of an unwritten extent */
 614 #define EXT4_GET_BLOCKS_UNWRIT_EXT              0x0002
 615 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT       (EXT4_GET_BLOCKS_UNWRIT_EXT|\
 616                                                  EXT4_GET_BLOCKS_CREATE)
 617         /* Caller is from the delayed allocation writeout path
 618          * finally doing the actual allocation of delayed blocks */
 619 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
 620         /* caller is from the direct IO path, request to creation of an
 621         unwritten extents if not allocated, split the unwritten
 622         extent if blocks has been preallocated already*/
 623 #define EXT4_GET_BLOCKS_PRE_IO                  0x0008
 624 #define EXT4_GET_BLOCKS_CONVERT                 0x0010
 625 #define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
 626                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
 627         /* Convert extent to initialized after IO complete */
 628 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
 629                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
 630         /* Eventual metadata allocation (due to growing extent tree)
 631          * should not fail, so try to use reserved blocks for that.*/
 632 #define EXT4_GET_BLOCKS_METADATA_NOFAIL         0x0020
 633         /* Don't normalize allocation size (used for fallocate) */
 634 #define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
 635         /* Convert written extents to unwritten */
 636 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN       0x0100
 637         /* Write zeros to newly created written extents */
 638 #define EXT4_GET_BLOCKS_ZERO                    0x0200
 639 #define EXT4_GET_BLOCKS_CREATE_ZERO             (EXT4_GET_BLOCKS_CREATE |\
 640                                         EXT4_GET_BLOCKS_ZERO)
 641         /* Caller will submit data before dropping transaction handle. This
 642          * allows jbd2 to avoid submitting data before commit. */
 643 #define EXT4_GET_BLOCKS_IO_SUBMIT               0x0400
 644
 645 /*
 646  * The bit position of these flags must not overlap with any of the
 647  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
 648  * read_extent_tree_block(), ext4_split_extent_at(),
 649  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
 650  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
 651  * caching the extents when reading from the extent tree while a
 652  * truncate or punch hole operation is in progress.
 653  */
 654 #define EXT4_EX_NOCACHE                         0x40000000
 655 #define EXT4_EX_FORCE_CACHE                     0x20000000
 656 #define EXT4_EX_NOFAIL                          0x10000000
 657
 658 /*
 659  * Flags used by ext4_free_blocks
 660  */
 661 #define EXT4_FREE_BLOCKS_METADATA               0x0001
 662 #define EXT4_FREE_BLOCKS_FORGET                 0x0002
 663 #define EXT4_FREE_BLOCKS_VALIDATED              0x0004
 664 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE         0x0008
 665 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
 666 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
 667 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
 668
 669 /*
 670  * ioctl commands
 671  */
 672 #define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
 673 #define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
 674 #define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
 675 #define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
 676 #define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
 677 #define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
 678 #define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
 679 #define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
 680 #define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
 681 #define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
 682 #define EXT4_IOC_MIGRATE                _IO('f', 9)
 683  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 684  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 685 #define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
 686 #define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
 687 #define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
 688 #define EXT4_IOC_SWAP_BOOT              _IO('f', 17)
 689 #define EXT4_IOC_PRECACHE_EXTENTS       _IO('f', 18)
 690 #define EXT4_IOC_SET_ENCRYPTION_POLICY  FS_IOC_SET_ENCRYPTION_POLICY
 691 #define EXT4_IOC_GET_ENCRYPTION_PWSALT  FS_IOC_GET_ENCRYPTION_PWSALT
 692 #define EXT4_IOC_GET_ENCRYPTION_POLICY  FS_IOC_GET_ENCRYPTION_POLICY
 693 /* ioctl codes 19--39 are reserved for fscrypt */
 694 #define EXT4_IOC_CLEAR_ES_CACHE         _IO('f', 40)
 695 #define EXT4_IOC_GETSTATE               _IOW('f', 41, __u32)
 696 #define EXT4_IOC_GET_ES_CACHE           _IOWR('f', 42, struct fiemap)
 697
 698 #define EXT4_IOC_FSGETXATTR             FS_IOC_FSGETXATTR
 699 #define EXT4_IOC_FSSETXATTR             FS_IOC_FSSETXATTR
 700
 701 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
 702
 703 /*
 704  * Flags for going down operation
 705  */
 706 #define EXT4_GOING_FLAGS_DEFAULT                0x0     /* going down */
 707 #define EXT4_GOING_FLAGS_LOGFLUSH               0x1     /* flush log but not data */
 708 #define EXT4_GOING_FLAGS_NOLOGFLUSH             0x2     /* don't flush log nor data */
 709
 710 /*
 711  * Flags returned by EXT4_IOC_GETSTATE
 712  *
 713  * We only expose to userspace a subset of the state flags in
 714  * i_state_flags
 715  */
 716 #define EXT4_STATE_FLAG_EXT_PRECACHED   0x00000001
 717 #define EXT4_STATE_FLAG_NEW             0x00000002
 718 #define EXT4_STATE_FLAG_NEWENTRY        0x00000004
 719 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE  0x00000008
 720
 721 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 722 /*
 723  * ioctl commands in 32 bit emulation
 724  */
 725 #define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
 726 #define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
 727 #define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
 728 #define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
 729 #define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
 730 #define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
 731 #define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
 732 #define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
 733 #define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
 734 #define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
 735 #endif
 736
 737 /*
 738  * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
 739  * It indicates that the entry in extent status cache is for a hole.
 740  */
 741 #define EXT4_FIEMAP_EXTENT_HOLE         0x08000000
 742
 743 /* Max physical block we can address w/o extents */
 744 #define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
 745
 746 /* Max logical block we can support */
 747 #define EXT4_MAX_LOGICAL_BLOCK          0xFFFFFFFE
 748
 749 /*
 750  * Structure of an inode on the disk
 751  */
 752 struct ext4_inode {
 753         __le16  i_mode;         /* File mode */
 754         __le16  i_uid;          /* Low 16 bits of Owner Uid */
 755         __le32  i_size_lo;      /* Size in bytes */
 756         __le32  i_atime;        /* Access time */
 757         __le32  i_ctime;        /* Inode Change time */
 758         __le32  i_mtime;        /* Modification time */
 759         __le32  i_dtime;        /* Deletion Time */
 760         __le16  i_gid;          /* Low 16 bits of Group Id */
 761         __le16  i_links_count;  /* Links count */
 762         __le32  i_blocks_lo;    /* Blocks count */
 763         __le32  i_flags;        /* File flags */
 764         union {
 765                 struct {
 766                         __le32  l_i_version;
 767                 } linux1;
 768                 struct {
 769                         __u32  h_i_translator;
 770                 } hurd1;
 771                 struct {
 772                         __u32  m_i_reserved1;
 773                 } masix1;
 774         } osd1;                         /* OS dependent 1 */
 775         __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 776         __le32  i_generation;   /* File version (for NFS) */
 777         __le32  i_file_acl_lo;  /* File ACL */
 778         __le32  i_size_high;
 779         __le32  i_obso_faddr;   /* Obsoleted fragment address */
 780         union {
 781                 struct {
 782                         __le16  l_i_blocks_high; /* were l_i_reserved1 */
 783                         __le16  l_i_file_acl_high;
 784                         __le16  l_i_uid_high;   /* these 2 fields */
 785                         __le16  l_i_gid_high;   /* were reserved2[0] */
 786                         __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
 787                         __le16  l_i_reserved;
 788                 } linux2;
 789                 struct {
 790                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 791                         __u16   h_i_mode_high;
 792                         __u16   h_i_uid_high;
 793                         __u16   h_i_gid_high;
 794                         __u32   h_i_author;
 795                 } hurd2;
 796                 struct {
 797                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 798                         __le16  m_i_file_acl_high;
 799                         __u32   m_i_reserved2[2];
 800                 } masix2;
 801         } osd2;                         /* OS dependent 2 */
 802         __le16  i_extra_isize;
 803         __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
 804         __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 805         __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 806         __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 807         __le32  i_crtime;       /* File Creation time */
 808         __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 809         __le32  i_version_hi;   /* high 32 bits for 64-bit version */
 810         __le32  i_projid;       /* Project ID */
 811 };
 812
 813 struct move_extent {
 814         __u32 reserved;         /* should be zero */
 815         __u32 donor_fd;         /* donor file descriptor */
 816         __u64 orig_start;       /* logical start offset in block for orig */
 817         __u64 donor_start;      /* logical start offset in block for donor */
 818         __u64 len;              /* block length to be moved */
 819         __u64 moved_len;        /* moved block length */
 820 };
 821
 822 #define EXT4_EPOCH_BITS 2
 823 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 824 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 825
 826 /*
 827  * Extended fields will fit into an inode if the filesystem was formatted
 828  * with large inodes (-I 256 or larger) and there are not currently any EAs
 829  * consuming all of the available space. For new inodes we always reserve
 830  * enough space for the kernel's known extended fields, but for inodes
 831  * created with an old kernel this might not have been the case. None of
 832  * the extended inode fields is critical for correct filesystem operation.
 833  * This macro checks if a certain field fits in the inode. Note that
 834  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 835  */
 836 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
 837         ((offsetof(typeof(*ext4_inode), field) +        \
 838           sizeof((ext4_inode)->field))                  \
 839         <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
 840             (einode)->i_extra_isize))                   \
 841
 842 /*
 843  * We use an encoding that preserves the times for extra epoch "00":
 844  *
 845  * extra  msb of                         adjust for signed
 846  * epoch  32-bit                         32-bit tv_sec to
 847  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
 848  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
 849  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
 850  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
 851  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
 852  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
 853  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
 854  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
 855  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
 856  *
 857  * Note that previous versions of the kernel on 64-bit systems would
 858  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
 859  * 1970.  e2fsck will correct this, assuming that it is run on the
 860  * affected filesystem before 2242.
 861  */
 862
 863 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
 864 {
 865         u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
 866         return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
 867 }
 868
 869 static inline void ext4_decode_extra_time(struct timespec64 *time,
 870                                           __le32 extra)
 871 {
 872         if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
 873                 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
 874         time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 875 }
 876
 877 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                           \
 878 do {                                                                            \
 879         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
 880                 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);        \
 881                 (raw_inode)->xtime ## _extra =                                  \
 882                                 ext4_encode_extra_time(&(inode)->xtime);        \
 883                 }                                                               \
 884         else    \
 885                 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));    \
 886 } while (0)
 887
 888 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
 889 do {                                                                           \
 890         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 891                 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 892         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 893                 (raw_inode)->xtime ## _extra =                                 \
 894                                 ext4_encode_extra_time(&(einode)->xtime);      \
 895 } while (0)
 896
 897 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                           \
 898 do {                                                                            \
 899         (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);        \
 900         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {    \
 901                 ext4_decode_extra_time(&(inode)->xtime,                         \
 902                                        raw_inode->xtime ## _extra);             \
 903                 }                                                               \
 904         else                                                                    \
 905                 (inode)->xtime.tv_nsec = 0;                                     \
 906 } while (0)
 907
 908
 909 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
 910 do {                                                                           \
 911         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 912                 (einode)->xtime.tv_sec =                                       \
 913                         (signed)le32_to_cpu((raw_inode)->xtime);               \
 914         else                                                                   \
 915                 (einode)->xtime.tv_sec = 0;                                    \
 916         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 917                 ext4_decode_extra_time(&(einode)->xtime,                       \
 918                                        raw_inode->xtime ## _extra);            \
 919         else                                                                   \
 920                 (einode)->xtime.tv_nsec = 0;                                   \
 921 } while (0)
 922
 923 #define i_disk_version osd1.linux1.l_i_version
 924
 925 #if defined(__KERNEL__) || defined(__linux__)
 926 #define i_reserved1     osd1.linux1.l_i_reserved1
 927 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
 928 #define i_blocks_high   osd2.linux2.l_i_blocks_high
 929 #define i_uid_low       i_uid
 930 #define i_gid_low       i_gid
 931 #define i_uid_high      osd2.linux2.l_i_uid_high
 932 #define i_gid_high      osd2.linux2.l_i_gid_high
 933 #define i_checksum_lo   osd2.linux2.l_i_checksum_lo
 934
 935 #elif defined(__GNU__)
 936
 937 #define i_translator    osd1.hurd1.h_i_translator
 938 #define i_uid_high      osd2.hurd2.h_i_uid_high
 939 #define i_gid_high      osd2.hurd2.h_i_gid_high
 940 #define i_author        osd2.hurd2.h_i_author
 941
 942 #elif defined(__masix__)
 943
 944 #define i_reserved1     osd1.masix1.m_i_reserved1
 945 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
 946 #define i_reserved2     osd2.masix2.m_i_reserved2
 947
 948 #endif /* defined(__KERNEL__) || defined(__linux__) */
 949
 950 #include "extents_status.h"
 951
 952 /*
 953  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
 954  *
 955  * These are needed to avoid lockdep false positives when we need to
 956  * allocate blocks to the quota inode during ext4_map_blocks(), while
 957  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
 958  * do quota tracking for the quota inode, this avoids deadlock (as
 959  * well as infinite recursion, since it isn't turtles all the way
 960  * down...)
 961  *
 962  *  I_DATA_SEM_NORMAL - Used for most inodes
 963  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
 964  *                        where the second inode has larger inode number
 965  *                        than the first
 966  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
 967  */
 968 enum {
 969         I_DATA_SEM_NORMAL = 0,
 970         I_DATA_SEM_OTHER,
 971         I_DATA_SEM_QUOTA,
 972 };
 973
 974
 975 /*
 976  * fourth extended file system inode data in memory
 977  */
 978 struct ext4_inode_info {
 979         __le32  i_data[15];     /* unconverted */
 980         __u32   i_dtime;
 981         ext4_fsblk_t    i_file_acl;
 982
 983         /*
 984          * i_block_group is the number of the block group which contains
 985          * this file's inode.  Constant across the lifetime of the inode,
 986          * it is used for making block allocation decisions - we try to
 987          * place a file's data blocks near its inode block, and new inodes
 988          * near to their parent directory's inode.
 989          */
 990         ext4_group_t    i_block_group;
 991         ext4_lblk_t     i_dir_start_lookup;
 992 #if (BITS_PER_LONG < 64)
 993         unsigned long   i_state_flags;          /* Dynamic state flags */
 994 #endif
 995         unsigned long   i_flags;
 996
 997         /*
 998          * Extended attributes can be read independently of the main file
 999          * data. Taking i_mutex even when reading would cause contention
1000          * between readers of EAs and writers of regular file data, so
1001          * instead we synchronize on xattr_sem when reading or changing
1002          * EAs.
1003          */
1004         struct rw_semaphore xattr_sem;
1005
1006         struct list_head i_orphan;      /* unlinked but open inodes */
1007
1008         /*
1009          * i_disksize keeps track of what the inode size is ON DISK, not
1010          * in memory.  During truncate, i_size is set to the new size by
1011          * the VFS prior to calling ext4_truncate(), but the filesystem won't
1012          * set i_disksize to 0 until the truncate is actually under way.
1013          *
1014          * The intent is that i_disksize always represents the blocks which
1015          * are used by this file.  This allows recovery to restart truncate
1016          * on orphans if we crash during truncate.  We actually write i_disksize
1017          * into the on-disk inode when writing inodes out, instead of i_size.
1018          *
1019          * The only time when i_disksize and i_size may be different is when
1020          * a truncate is in progress.  The only things which change i_disksize
1021          * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1022          */
1023         loff_t  i_disksize;
1024
1025         /*
1026          * i_data_sem is for serialising ext4_truncate() against
1027          * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1028          * data tree are chopped off during truncate. We can't do that in
1029          * ext4 because whenever we perform intermediate commits during
1030          * truncate, the inode and all the metadata blocks *must* be in a
1031          * consistent state which allows truncation of the orphans to restart
1032          * during recovery.  Hence we must fix the get_block-vs-truncate race
1033          * by other means, so we have i_data_sem.
1034          */
1035         struct rw_semaphore i_data_sem;
1036         /*
1037          * i_mmap_sem is for serializing page faults with truncate / punch hole
1038          * operations. We have to make sure that new page cannot be faulted in
1039          * a section of the inode that is being punched. We cannot easily use
1040          * i_data_sem for this since we need protection for the whole punch
1041          * operation and i_data_sem ranks below transaction start so we have
1042          * to occasionally drop it.
1043          */
1044         struct rw_semaphore i_mmap_sem;
1045         struct inode vfs_inode;
1046         struct jbd2_inode *jinode;
1047
1048         spinlock_t i_raw_lock;  /* protects updates to the raw inode */
1049
1050         /*
1051          * File creation time. Its function is same as that of
1052          * struct timespec64 i_{a,c,m}time in the generic inode.
1053          */
1054         struct timespec64 i_crtime;
1055
1056         /* mballoc */
1057         struct list_head i_prealloc_list;
1058         spinlock_t i_prealloc_lock;
1059
1060         /* extents status tree */
1061         struct ext4_es_tree i_es_tree;
1062         rwlock_t i_es_lock;
1063         struct list_head i_es_list;
1064         unsigned int i_es_all_nr;       /* protected by i_es_lock */
1065         unsigned int i_es_shk_nr;       /* protected by i_es_lock */
1066         ext4_lblk_t i_es_shrink_lblk;   /* Offset where we start searching for
1067                                            extents to shrink. Protected by
1068                                            i_es_lock  */
1069
1070         /* ialloc */
1071         ext4_group_t    i_last_alloc_group;
1072
1073         /* allocation reservation info for delalloc */
1074         /* In case of bigalloc, this refer to clusters rather than blocks */
1075         unsigned int i_reserved_data_blocks;
1076
1077         /* pending cluster reservations for bigalloc file systems */
1078         struct ext4_pending_tree i_pending_tree;
1079
1080         /* on-disk additional length */
1081         __u16 i_extra_isize;
1082
1083         /* Indicate the inline data space. */
1084         u16 i_inline_off;
1085         u16 i_inline_size;
1086
1087 #ifdef CONFIG_QUOTA
1088         /* quota space reservation, managed internally by quota code */
1089         qsize_t i_reserved_quota;
1090 #endif
1091
1092         /* Lock protecting lists below */
1093         spinlock_t i_completed_io_lock;
1094         /*
1095          * Completed IOs that need unwritten extents handling and have
1096          * transaction reserved
1097          */
1098         struct list_head i_rsv_conversion_list;
1099         struct work_struct i_rsv_conversion_work;
1100         atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1101
1102         spinlock_t i_block_reservation_lock;
1103
1104         /*
1105          * Transactions that contain inode's metadata needed to complete
1106          * fsync and fdatasync, respectively.
1107          */
1108         tid_t i_sync_tid;
1109         tid_t i_datasync_tid;
1110
1111 #ifdef CONFIG_QUOTA
1112         struct dquot *i_dquot[MAXQUOTAS];
1113 #endif
1114
1115         /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1116         __u32 i_csum_seed;
1117
1118         kprojid_t i_projid;
1119 };
1120
1121 /*
1122  * File system states
1123  */
1124 #define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
1125 #define EXT4_ERROR_FS                   0x0002  /* Errors detected */
1126 #define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
1127
1128 /*
1129  * Misc. filesystem flags
1130  */
1131 #define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
1132 #define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
1133 #define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
1134
1135 /*
1136  * Mount flags set via mount options or defaults
1137  */
1138 #define EXT4_MOUNT_NO_MBCACHE           0x00001 /* Do not use mbcache */
1139 #define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
1140 #define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
1141 #define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
1142 #define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
1143 #define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
1144 #define EXT4_MOUNT_ERRORS_MASK          0x00070
1145 #define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
1146 #define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
1147 #ifdef CONFIG_FS_DAX
1148 #define EXT4_MOUNT_DAX_ALWAYS           0x00200 /* Direct Access */
1149 #else
1150 #define EXT4_MOUNT_DAX_ALWAYS           0
1151 #endif
1152 #define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
1153 #define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
1154 #define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
1155 #define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
1156 #define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
1157 #define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
1158 #define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
1159 #define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
1160 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
1161 #define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
1162 #define EXT4_MOUNT_QUOTA                0x40000 /* Some quota option set */
1163 #define EXT4_MOUNT_USRQUOTA             0x80000 /* "old" user quota,
1164                                                  * enable enforcement for hidden
1165                                                  * quota files */
1166 #define EXT4_MOUNT_GRPQUOTA             0x100000 /* "old" group quota, enable
1167                                                   * enforcement for hidden quota
1168                                                   * files */
1169 #define EXT4_MOUNT_PRJQUOTA             0x200000 /* Enable project quota
1170                                                   * enforcement */
1171 #define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
1172 #define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
1173 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1174 #define EXT4_MOUNT_WARN_ON_ERROR        0x2000000 /* Trigger WARN_ON on error */
1175 #define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
1176 #define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
1177 #define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
1178 #define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
1179 #define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
1180
1181 /*
1182  * Mount flags set either automatically (could not be set by mount option)
1183  * based on per file system feature or property or in special cases such as
1184  * distinguishing between explicit mount option definition and default.
1185  */
1186 #define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
1187                                                       specified delalloc */
1188 #define EXT4_MOUNT2_STD_GROUP_SIZE      0x00000002 /* We have standard group
1189                                                       size of blocksize * 8
1190                                                       blocks */
1191 #define EXT4_MOUNT2_HURD_COMPAT         0x00000004 /* Support HURD-castrated
1192                                                       file systems */
1193 #define EXT4_MOUNT2_DAX_NEVER           0x00000008 /* Do not allow Direct Access */
1194 #define EXT4_MOUNT2_DAX_INODE           0x00000010 /* For printing options only */
1195
1196 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM   0x00000008 /* User explicitly
1197                                                 specified journal checksum */
1198
1199 #define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
1200                                                 ~EXT4_MOUNT_##opt
1201 #define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1202                                                 EXT4_MOUNT_##opt
1203 #define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1204                                          EXT4_MOUNT_##opt)
1205
1206 #define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1207                                                 ~EXT4_MOUNT2_##opt
1208 #define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1209                                                 EXT4_MOUNT2_##opt
1210 #define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1211                                          EXT4_MOUNT2_##opt)
1212
1213 #define ext4_test_and_set_bit           __test_and_set_bit_le
1214 #define ext4_set_bit                    __set_bit_le
1215 #define ext4_set_bit_atomic             ext2_set_bit_atomic
1216 #define ext4_test_and_clear_bit         __test_and_clear_bit_le
1217 #define ext4_clear_bit                  __clear_bit_le
1218 #define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1219 #define ext4_test_bit                   test_bit_le
1220 #define ext4_find_next_zero_bit         find_next_zero_bit_le
1221 #define ext4_find_next_bit              find_next_bit_le
1222
1223 extern void ext4_set_bits(void *bm, int cur, int len);
1224
1225 /*
1226  * Maximal mount counts between two filesystem checks
1227  */
1228 #define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1229 #define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1230
1231 /*
1232  * Behaviour when detecting errors
1233  */
1234 #define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1235 #define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1236 #define EXT4_ERRORS_PANIC               3       /* Panic */
1237 #define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1238
1239 /* Metadata checksum algorithm codes */
1240 #define EXT4_CRC32C_CHKSUM              1
1241
1242 /*
1243  * Structure of the super block
1244  */
1245 struct ext4_super_block {
1246 /*00*/  __le32  s_inodes_count;         /* Inodes count */
1247         __le32  s_blocks_count_lo;      /* Blocks count */
1248         __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1249         __le32  s_free_blocks_count_lo; /* Free blocks count */
1250 /*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1251         __le32  s_first_data_block;     /* First Data Block */
1252         __le32  s_log_block_size;       /* Block size */
1253         __le32  s_log_cluster_size;     /* Allocation cluster size */
1254 /*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1255         __le32  s_clusters_per_group;   /* # Clusters per group */
1256         __le32  s_inodes_per_group;     /* # Inodes per group */
1257         __le32  s_mtime;                /* Mount time */
1258 /*30*/  __le32  s_wtime;                /* Write time */
1259         __le16  s_mnt_count;            /* Mount count */
1260         __le16  s_max_mnt_count;        /* Maximal mount count */
1261         __le16  s_magic;                /* Magic signature */
1262         __le16  s_state;                /* File system state */
1263         __le16  s_errors;               /* Behaviour when detecting errors */
1264         __le16  s_minor_rev_level;      /* minor revision level */
1265 /*40*/  __le32  s_lastcheck;            /* time of last check */
1266         __le32  s_checkinterval;        /* max. time between checks */
1267         __le32  s_creator_os;           /* OS */
1268         __le32  s_rev_level;            /* Revision level */
1269 /*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1270         __le16  s_def_resgid;           /* Default gid for reserved blocks */
1271         /*
1272          * These fields are for EXT4_DYNAMIC_REV superblocks only.
1273          *
1274          * Note: the difference between the compatible feature set and
1275          * the incompatible feature set is that if there is a bit set
1276          * in the incompatible feature set that the kernel doesn't
1277          * know about, it should refuse to mount the filesystem.
1278          *
1279          * e2fsck's requirements are more strict; if it doesn't know
1280          * about a feature in either the compatible or incompatible
1281          * feature set, it must abort and not try to meddle with
1282          * things it doesn't understand...
1283          */
1284         __le32  s_first_ino;            /* First non-reserved inode */
1285         __le16  s_inode_size;           /* size of inode structure */
1286         __le16  s_block_group_nr;       /* block group # of this superblock */
1287         __le32  s_feature_compat;       /* compatible feature set */
1288 /*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1289         __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1290 /*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1291 /*78*/  char    s_volume_name[16];      /* volume name */
1292 /*88*/  char    s_last_mounted[64] __nonstring; /* directory where last mounted */
1293 /*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1294         /*
1295          * Performance hints.  Directory preallocation should only
1296          * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1297          */
1298         __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1299         __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1300         __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1301         /*
1302          * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1303          */
1304 /*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1305 /*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1306         __le32  s_journal_dev;          /* device number of journal file */
1307         __le32  s_last_orphan;          /* start of list of inodes to delete */
1308         __le32  s_hash_seed[4];         /* HTREE hash seed */
1309         __u8    s_def_hash_version;     /* Default hash version to use */
1310         __u8    s_jnl_backup_type;
1311         __le16  s_desc_size;            /* size of group descriptor */
1312 /*100*/ __le32  s_default_mount_opts;
1313         __le32  s_first_meta_bg;        /* First metablock block group */
1314         __le32  s_mkfs_time;            /* When the filesystem was created */
1315         __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1316         /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1317 /*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1318         __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1319         __le32  s_free_blocks_count_hi; /* Free blocks count */
1320         __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1321         __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1322         __le32  s_flags;                /* Miscellaneous flags */
1323         __le16  s_raid_stride;          /* RAID stride */
1324         __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1325         __le64  s_mmp_block;            /* Block for multi-mount protection */
1326         __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1327         __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1328         __u8    s_checksum_type;        /* metadata checksum algorithm used */
1329         __u8    s_encryption_level;     /* versioning level for encryption */
1330         __u8    s_reserved_pad;         /* Padding to next 32bits */
1331         __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1332         __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1333         __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1334         __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1335                                               snapshot's future use */
1336         __le32  s_snapshot_list;        /* inode number of the head of the
1337                                            on-disk snapshot list */
1338 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1339         __le32  s_error_count;          /* number of fs errors */
1340         __le32  s_first_error_time;     /* first time an error happened */
1341         __le32  s_first_error_ino;      /* inode involved in first error */
1342         __le64  s_first_error_block;    /* block involved of first error */
1343         __u8    s_first_error_func[32] __nonstring;     /* function where the error happened */
1344         __le32  s_first_error_line;     /* line number where error happened */
1345         __le32  s_last_error_time;      /* most recent time of an error */
1346         __le32  s_last_error_ino;       /* inode involved in last error */
1347         __le32  s_last_error_line;      /* line number where error happened */
1348         __le64  s_last_error_block;     /* block involved of last error */
1349         __u8    s_last_error_func[32] __nonstring;      /* function where the error happened */
1350 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1351         __u8    s_mount_opts[64];
1352         __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1353         __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1354         __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1355         __le32  s_backup_bgs[2];        /* groups with sparse_super2 SBs */
1356         __u8    s_encrypt_algos[4];     /* Encryption algorithms in use  */
1357         __u8    s_encrypt_pw_salt[16];  /* Salt used for string2key algorithm */
1358         __le32  s_lpf_ino;              /* Location of the lost+found inode */
1359         __le32  s_prj_quota_inum;       /* inode for tracking project quota */
1360         __le32  s_checksum_seed;        /* crc32c(uuid) if csum_seed set */
1361         __u8    s_wtime_hi;
1362         __u8    s_mtime_hi;
1363         __u8    s_mkfs_time_hi;
1364         __u8    s_lastcheck_hi;
1365         __u8    s_first_error_time_hi;
1366         __u8    s_last_error_time_hi;
1367         __u8    s_first_error_errcode;
1368         __u8    s_last_error_errcode;
1369         __le16  s_encoding;             /* Filename charset encoding */
1370         __le16  s_encoding_flags;       /* Filename charset encoding flags */
1371         __le32  s_reserved[95];         /* Padding to the end of the block */
1372         __le32  s_checksum;             /* crc32c(superblock) */
1373 };
1374
1375 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1376
1377 #ifdef __KERNEL__
1378
1379 /*
1380  * run-time mount flags
1381  */
1382 #define EXT4_MF_MNTDIR_SAMPLED          0x0001
1383 #define EXT4_MF_FS_ABORTED              0x0002  /* Fatal error detected */
1384
1385 #ifdef CONFIG_FS_ENCRYPTION
1386 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL)
1387 #else
1388 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1389 #endif
1390
1391 /* Number of quota types we support */
1392 #define EXT4_MAXQUOTAS 3
1393
1394 #define EXT4_ENC_UTF8_12_1      1
1395
1396 /*
1397  * Flags for ext4_sb_info.s_encoding_flags.
1398  */
1399 #define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1400
1401 #define ext4_has_strict_mode(sbi) \
1402         (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1403
1404 /*
1405  * fourth extended-fs super-block data in memory
1406  */
1407 struct ext4_sb_info {
1408         unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1409         unsigned long s_inodes_per_block;/* Number of inodes per block */
1410         unsigned long s_blocks_per_group;/* Number of blocks in a group */
1411         unsigned long s_clusters_per_group; /* Number of clusters in a group */
1412         unsigned long s_inodes_per_group;/* Number of inodes in a group */
1413         unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1414         unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1415         unsigned long s_desc_per_block; /* Number of group descriptors per block */
1416         ext4_group_t s_groups_count;    /* Number of groups in the fs */
1417         ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1418         unsigned long s_overhead;  /* # of fs overhead clusters */
1419         unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1420         unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1421         loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1422         struct buffer_head * s_sbh;     /* Buffer containing the super block */
1423         struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1424         struct buffer_head * __rcu *s_group_desc;
1425         unsigned int s_mount_opt;
1426         unsigned int s_mount_opt2;
1427         unsigned int s_mount_flags;
1428         unsigned int s_def_mount_opt;
1429         ext4_fsblk_t s_sb_block;
1430         atomic64_t s_resv_clusters;
1431         kuid_t s_resuid;
1432         kgid_t s_resgid;
1433         unsigned short s_mount_state;
1434         unsigned short s_pad;
1435         int s_addr_per_block_bits;
1436         int s_desc_per_block_bits;
1437         int s_inode_size;
1438         int s_first_ino;
1439         unsigned int s_inode_readahead_blks;
1440         unsigned int s_inode_goal;
1441         u32 s_hash_seed[4];
1442         int s_def_hash_version;
1443         int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1444         struct percpu_counter s_freeclusters_counter;
1445         struct percpu_counter s_freeinodes_counter;
1446         struct percpu_counter s_dirs_counter;
1447         struct percpu_counter s_dirtyclusters_counter;
1448         struct blockgroup_lock *s_blockgroup_lock;
1449         struct proc_dir_entry *s_proc;
1450         struct kobject s_kobj;
1451         struct completion s_kobj_unregister;
1452         struct super_block *s_sb;
1453 #ifdef CONFIG_UNICODE
1454         struct unicode_map *s_encoding;
1455         __u16 s_encoding_flags;
1456 #endif
1457
1458         /* Journaling */
1459         struct journal_s *s_journal;
1460         struct list_head s_orphan;
1461         struct mutex s_orphan_lock;
1462         unsigned long s_ext4_flags;             /* Ext4 superblock flags */
1463         unsigned long s_commit_interval;
1464         u32 s_max_batch_time;
1465         u32 s_min_batch_time;
1466         struct block_device *journal_bdev;
1467 #ifdef CONFIG_QUOTA
1468         /* Names of quota files with journalled quota */
1469         char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1470         int s_jquota_fmt;                       /* Format of quota to use */
1471 #endif
1472         unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1473         struct ext4_system_blocks __rcu *system_blks;
1474
1475 #ifdef EXTENTS_STATS
1476         /* ext4 extents stats */
1477         unsigned long s_ext_min;
1478         unsigned long s_ext_max;
1479         unsigned long s_depth_max;
1480         spinlock_t s_ext_stats_lock;
1481         unsigned long s_ext_blocks;
1482         unsigned long s_ext_extents;
1483 #endif
1484
1485         /* for buddy allocator */
1486         struct ext4_group_info ** __rcu *s_group_info;
1487         struct inode *s_buddy_cache;
1488         spinlock_t s_md_lock;
1489         unsigned short *s_mb_offsets;
1490         unsigned int *s_mb_maxs;
1491         unsigned int s_group_info_size;
1492         unsigned int s_mb_free_pending;
1493         struct list_head s_freed_data_list;     /* List of blocks to be freed
1494                                                    after commit completed */
1495
1496         /* tunables */
1497         unsigned long s_stripe;
1498         unsigned int s_mb_stream_request;
1499         unsigned int s_mb_max_to_scan;
1500         unsigned int s_mb_min_to_scan;
1501         unsigned int s_mb_stats;
1502         unsigned int s_mb_order2_reqs;
1503         unsigned int s_mb_group_prealloc;
1504         unsigned int s_max_dir_size_kb;
1505         /* where last allocation was done - for stream allocation */
1506         unsigned long s_mb_last_group;
1507         unsigned long s_mb_last_start;
1508
1509         /* stats for buddy allocator */
1510         atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1511         atomic_t s_bal_success; /* we found long enough chunks */
1512         atomic_t s_bal_allocated;       /* in blocks */
1513         atomic_t s_bal_ex_scanned;      /* total extents scanned */
1514         atomic_t s_bal_goals;   /* goal hits */
1515         atomic_t s_bal_breaks;  /* too long searches */
1516         atomic_t s_bal_2orders; /* 2^order hits */
1517         spinlock_t s_bal_lock;
1518         unsigned long s_mb_buddies_generated;
1519         unsigned long long s_mb_generation_time;
1520         atomic_t s_mb_lost_chunks;
1521         atomic_t s_mb_preallocated;
1522         atomic_t s_mb_discarded;
1523         atomic_t s_lock_busy;
1524
1525         /* locality groups */
1526         struct ext4_locality_group __percpu *s_locality_groups;
1527
1528         /* for write statistics */
1529         unsigned long s_sectors_written_start;
1530         u64 s_kbytes_written;
1531
1532         /* the size of zero-out chunk */
1533         unsigned int s_extent_max_zeroout_kb;
1534
1535         unsigned int s_log_groups_per_flex;
1536         struct flex_groups * __rcu *s_flex_groups;
1537         ext4_group_t s_flex_groups_allocated;
1538
1539         /* workqueue for reserved extent conversions (buffered io) */
1540         struct workqueue_struct *rsv_conversion_wq;
1541
1542         /* timer for periodic error stats printing */
1543         struct timer_list s_err_report;
1544
1545         /* Lazy inode table initialization info */
1546         struct ext4_li_request *s_li_request;
1547         /* Wait multiplier for lazy initialization thread */
1548         unsigned int s_li_wait_mult;
1549
1550         /* Kernel thread for multiple mount protection */
1551         struct task_struct *s_mmp_tsk;
1552
1553         /* record the last minlen when FITRIM is called. */
1554         atomic_t s_last_trim_minblks;
1555
1556         /* Reference to checksum algorithm driver via cryptoapi */
1557         struct crypto_shash *s_chksum_driver;
1558
1559         /* Precomputed FS UUID checksum for seeding other checksums */
1560         __u32 s_csum_seed;
1561
1562         /* Reclaim extents from extent status tree */
1563         struct shrinker s_es_shrinker;
1564         struct list_head s_es_list;     /* List of inodes with reclaimable extents */
1565         long s_es_nr_inode;
1566         struct ext4_es_stats s_es_stats;
1567         struct mb_cache *s_ea_block_cache;
1568         struct mb_cache *s_ea_inode_cache;
1569         spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1570
1571         /* Ratelimit ext4 messages. */
1572         struct ratelimit_state s_err_ratelimit_state;
1573         struct ratelimit_state s_warning_ratelimit_state;
1574         struct ratelimit_state s_msg_ratelimit_state;
1575
1576         /* Encryption context for '-o test_dummy_encryption' */
1577         struct fscrypt_dummy_context s_dummy_enc_ctx;
1578
1579         /*
1580          * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1581          * or EXTENTS flag.
1582          */
1583         struct percpu_rw_semaphore s_writepages_rwsem;
1584         struct dax_device *s_daxdev;
1585 #ifdef CONFIG_EXT4_DEBUG
1586         unsigned long s_simulate_fail;
1587 #endif
1588 };
1589
1590 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1591 {
1592         return sb->s_fs_info;
1593 }
1594 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1595 {
1596         return container_of(inode, struct ext4_inode_info, vfs_inode);
1597 }
1598
1599 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1600 {
1601         return ino == EXT4_ROOT_INO ||
1602                 (ino >= EXT4_FIRST_INO(sb) &&
1603                  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1604 }
1605
1606 /*
1607  * Returns: sbi->field[index]
1608  * Used to access an array element from the following sbi fields which require
1609  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1610  * - s_group_desc
1611  * - s_group_info
1612  * - s_flex_group
1613  */
1614 #define sbi_array_rcu_deref(sbi, field, index)                             \
1615 ({                                                                         \
1616         typeof(*((sbi)->field)) _v;                                        \
1617         rcu_read_lock();                                                   \
1618         _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];          \
1619         rcu_read_unlock();                                                 \
1620         _v;                                                                \
1621 })
1622
1623 /*
1624  * Simulate_fail codes
1625  */
1626 #define EXT4_SIM_BBITMAP_EIO    1
1627 #define EXT4_SIM_BBITMAP_CRC    2
1628 #define EXT4_SIM_IBITMAP_EIO    3
1629 #define EXT4_SIM_IBITMAP_CRC    4
1630 #define EXT4_SIM_INODE_EIO      5
1631 #define EXT4_SIM_INODE_CRC      6
1632 #define EXT4_SIM_DIRBLOCK_EIO   7
1633 #define EXT4_SIM_DIRBLOCK_CRC   8
1634
1635 static inline bool ext4_simulate_fail(struct super_block *sb,
1636                                      unsigned long code)
1637 {
1638 #ifdef CONFIG_EXT4_DEBUG
1639         struct ext4_sb_info *sbi = EXT4_SB(sb);
1640
1641         if (unlikely(sbi->s_simulate_fail == code)) {
1642                 sbi->s_simulate_fail = 0;
1643                 return true;
1644         }
1645 #endif
1646         return false;
1647 }
1648
1649 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1650                                          struct buffer_head *bh,
1651                                          unsigned long code)
1652 {
1653         if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1654                 clear_buffer_uptodate(bh);
1655 }
1656
1657 /*
1658  * Error number codes for s_{first,last}_error_errno
1659  *
1660  * Linux errno numbers are architecture specific, so we need to translate
1661  * them into something which is architecture independent.   We don't define
1662  * codes for all errno's; just the ones which are most likely to be the cause
1663  * of an ext4_error() call.
1664  */
1665 #define EXT4_ERR_UNKNOWN         1
1666 #define EXT4_ERR_EIO             2
1667 #define EXT4_ERR_ENOMEM          3
1668 #define EXT4_ERR_EFSBADCRC       4
1669 #define EXT4_ERR_EFSCORRUPTED    5
1670 #define EXT4_ERR_ENOSPC          6
1671 #define EXT4_ERR_ENOKEY          7
1672 #define EXT4_ERR_EROFS           8
1673 #define EXT4_ERR_EFBIG           9
1674 #define EXT4_ERR_EEXIST         10
1675 #define EXT4_ERR_ERANGE         11
1676 #define EXT4_ERR_EOVERFLOW      12
1677 #define EXT4_ERR_EBUSY          13
1678 #define EXT4_ERR_ENOTDIR        14
1679 #define EXT4_ERR_ENOTEMPTY      15
1680 #define EXT4_ERR_ESHUTDOWN      16
1681 #define EXT4_ERR_EFAULT         17
1682
1683 /*
1684  * Inode dynamic state flags
1685  */
1686 enum {
1687         EXT4_STATE_JDATA,               /* journaled data exists */
1688         EXT4_STATE_NEW,                 /* inode is newly created */
1689         EXT4_STATE_XATTR,               /* has in-inode xattrs */
1690         EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1691         EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1692         EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1693         EXT4_STATE_NEWENTRY,            /* File just added to dir */
1694         EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1695         EXT4_STATE_EXT_PRECACHED,       /* extents have been precached */
1696         EXT4_STATE_LUSTRE_EA_INODE,     /* Lustre-style ea_inode */
1697         EXT4_STATE_VERITY_IN_PROGRESS,  /* building fs-verity Merkle tree */
1698 };
1699
1700 #define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1701 static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1702 {                                                                       \
1703         return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1704 }                                                                       \
1705 static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1706 {                                                                       \
1707         set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1708 }                                                                       \
1709 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1710 {                                                                       \
1711         clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1712 }
1713
1714 /* Add these declarations here only so that these functions can be
1715  * found by name.  Otherwise, they are very hard to locate. */
1716 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1717 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1718 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1719 EXT4_INODE_BIT_FNS(flag, flags, 0)
1720
1721 /* Add these declarations here only so that these functions can be
1722  * found by name.  Otherwise, they are very hard to locate. */
1723 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1724 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1725 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1726 #if (BITS_PER_LONG < 64)
1727 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1728
1729 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1730 {
1731         (ei)->i_state_flags = 0;
1732 }
1733 #else
1734 EXT4_INODE_BIT_FNS(state, flags, 32)
1735
1736 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1737 {
1738         /* We depend on the fact that callers will set i_flags */
1739 }
1740 #endif
1741 #else
1742 /* Assume that user mode programs are passing in an ext4fs superblock, not
1743  * a kernel struct super_block.  This will allow us to call the feature-test
1744  * macros from user land. */
1745 #define EXT4_SB(sb)     (sb)
1746 #endif
1747
1748 static inline bool ext4_verity_in_progress(struct inode *inode)
1749 {
1750         return IS_ENABLED(CONFIG_FS_VERITY) &&
1751                ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1752 }
1753
1754 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1755
1756 /*
1757  * Codes for operating systems
1758  */
1759 #define EXT4_OS_LINUX           0
1760 #define EXT4_OS_HURD            1
1761 #define EXT4_OS_MASIX           2
1762 #define EXT4_OS_FREEBSD         3
1763 #define EXT4_OS_LITES           4
1764
1765 /*
1766  * Revision levels
1767  */
1768 #define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1769 #define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1770
1771 #define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1772 #define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1773
1774 #define EXT4_GOOD_OLD_INODE_SIZE 128
1775
1776 #define EXT4_EXTRA_TIMESTAMP_MAX        (((s64)1 << 34) - 1  + S32_MIN)
1777 #define EXT4_NON_EXTRA_TIMESTAMP_MAX    S32_MAX
1778 #define EXT4_TIMESTAMP_MIN              S32_MIN
1779
1780 /*
1781  * Feature set definitions
1782  */
1783
1784 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1785 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1786 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1787 #define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1788 #define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1789 #define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1790 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2       0x0200
1791 #define EXT4_FEATURE_COMPAT_STABLE_INODES       0x0800
1792
1793 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1794 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1795 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1796 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1797 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1798 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1799 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1800 #define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1801 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1802 /*
1803  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1804  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1805  * all other data structures' checksums.  However, the METADATA_CSUM and
1806  * GDT_CSUM bits are mutually exclusive.
1807  */
1808 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1809 #define EXT4_FEATURE_RO_COMPAT_READONLY         0x1000
1810 #define EXT4_FEATURE_RO_COMPAT_PROJECT          0x2000
1811 #define EXT4_FEATURE_RO_COMPAT_VERITY           0x8000
1812
1813 #define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1814 #define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1815 #define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1816 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1817 #define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1818 #define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1819 #define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1820 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1821 #define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1822 #define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1823 #define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1824 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED         0x2000
1825 #define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1826 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1827 #define EXT4_FEATURE_INCOMPAT_ENCRYPT           0x10000
1828 #define EXT4_FEATURE_INCOMPAT_CASEFOLD          0x20000
1829
1830 extern void ext4_update_dynamic_rev(struct super_block *sb);
1831
1832 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1833 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1834 { \
1835         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1836                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1837 } \
1838 static inline void ext4_set_feature_##name(struct super_block *sb) \
1839 { \
1840         ext4_update_dynamic_rev(sb); \
1841         EXT4_SB(sb)->s_es->s_feature_compat |= \
1842                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1843 } \
1844 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1845 { \
1846         EXT4_SB(sb)->s_es->s_feature_compat &= \
1847                 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1848 }
1849
1850 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1851 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1852 { \
1853         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1854                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1855 } \
1856 static inline void ext4_set_feature_##name(struct super_block *sb) \
1857 { \
1858         ext4_update_dynamic_rev(sb); \
1859         EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1860                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1861 } \
1862 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1863 { \
1864         EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1865                 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1866 }
1867
1868 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1869 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1870 { \
1871         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1872                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1873 } \
1874 static inline void ext4_set_feature_##name(struct super_block *sb) \
1875 { \
1876         ext4_update_dynamic_rev(sb); \
1877         EXT4_SB(sb)->s_es->s_feature_incompat |= \
1878                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1879 } \
1880 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1881 { \
1882         EXT4_SB(sb)->s_es->s_feature_incompat &= \
1883                 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1884 }
1885
1886 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,         DIR_PREALLOC)
1887 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,        IMAGIC_INODES)
1888 EXT4_FEATURE_COMPAT_FUNCS(journal,              HAS_JOURNAL)
1889 EXT4_FEATURE_COMPAT_FUNCS(xattr,                EXT_ATTR)
1890 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,         RESIZE_INODE)
1891 EXT4_FEATURE_COMPAT_FUNCS(dir_index,            DIR_INDEX)
1892 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,        SPARSE_SUPER2)
1893 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,        STABLE_INODES)
1894
1895 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,      SPARSE_SUPER)
1896 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,        LARGE_FILE)
1897 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,         BTREE_DIR)
1898 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,         HUGE_FILE)
1899 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,          GDT_CSUM)
1900 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,         DIR_NLINK)
1901 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,       EXTRA_ISIZE)
1902 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,             QUOTA)
1903 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,          BIGALLOC)
1904 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,     METADATA_CSUM)
1905 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,          READONLY)
1906 EXT4_FEATURE_RO_COMPAT_FUNCS(project,           PROJECT)
1907 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,            VERITY)
1908
1909 EXT4_FEATURE_INCOMPAT_FUNCS(compression,        COMPRESSION)
1910 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,           FILETYPE)
1911 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,     RECOVER)
1912 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,        JOURNAL_DEV)
1913 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,            META_BG)
1914 EXT4_FEATURE_INCOMPAT_FUNCS(extents,            EXTENTS)
1915 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,              64BIT)
1916 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,                MMP)
1917 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,            FLEX_BG)
1918 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,           EA_INODE)
1919 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,            DIRDATA)
1920 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,          CSUM_SEED)
1921 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,           LARGEDIR)
1922 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,        INLINE_DATA)
1923 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,            ENCRYPT)
1924 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,           CASEFOLD)
1925
1926 #define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1927 #define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1928                                          EXT4_FEATURE_INCOMPAT_META_BG)
1929 #define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1930                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1931                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1932
1933 #define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1934 #define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1935                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1936                                          EXT4_FEATURE_INCOMPAT_META_BG)
1937 #define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1938                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1939                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1940
1941 #define EXT4_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1942 #define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1943                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1944                                          EXT4_FEATURE_INCOMPAT_META_BG| \
1945                                          EXT4_FEATURE_INCOMPAT_EXTENTS| \
1946                                          EXT4_FEATURE_INCOMPAT_64BIT| \
1947                                          EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1948                                          EXT4_FEATURE_INCOMPAT_EA_INODE| \
1949                                          EXT4_FEATURE_INCOMPAT_MMP | \
1950                                          EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1951                                          EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1952                                          EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1953                                          EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1954                                          EXT4_FEATURE_INCOMPAT_LARGEDIR)
1955 #define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1956                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1957                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1958                                          EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1959                                          EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1960                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1961                                          EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1962                                          EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1963                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1964                                          EXT4_FEATURE_RO_COMPAT_QUOTA |\
1965                                          EXT4_FEATURE_RO_COMPAT_PROJECT |\
1966                                          EXT4_FEATURE_RO_COMPAT_VERITY)
1967
1968 #define EXTN_FEATURE_FUNCS(ver) \
1969 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1970 { \
1971         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1972                 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1973 } \
1974 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1975 { \
1976         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1977                 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1978 } \
1979 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1980 { \
1981         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1982                 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1983 }
1984
1985 EXTN_FEATURE_FUNCS(2)
1986 EXTN_FEATURE_FUNCS(3)
1987 EXTN_FEATURE_FUNCS(4)
1988
1989 static inline bool ext4_has_compat_features(struct super_block *sb)
1990 {
1991         return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1992 }
1993 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1994 {
1995         return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1996 }
1997 static inline bool ext4_has_incompat_features(struct super_block *sb)
1998 {
1999         return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2000 }
2001
2002 /*
2003  * Superblock flags
2004  */
2005 #define EXT4_FLAGS_RESIZING     0
2006 #define EXT4_FLAGS_SHUTDOWN     1
2007 #define EXT4_FLAGS_BDEV_IS_DAX  2
2008
2009 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2010 {
2011         return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2012 }
2013
2014
2015 /*
2016  * Default values for user and/or group using reserved blocks
2017  */
2018 #define EXT4_DEF_RESUID         0
2019 #define EXT4_DEF_RESGID         0
2020
2021 /*
2022  * Default project ID
2023  */
2024 #define EXT4_DEF_PROJID         0
2025
2026 #define EXT4_DEF_INODE_READAHEAD_BLKS   32
2027
2028 /*
2029  * Default mount options
2030  */
2031 #define EXT4_DEFM_DEBUG         0x0001
2032 #define EXT4_DEFM_BSDGROUPS     0x0002
2033 #define EXT4_DEFM_XATTR_USER    0x0004
2034 #define EXT4_DEFM_ACL           0x0008
2035 #define EXT4_DEFM_UID16         0x0010
2036 #define EXT4_DEFM_JMODE         0x0060
2037 #define EXT4_DEFM_JMODE_DATA    0x0020
2038 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2039 #define EXT4_DEFM_JMODE_WBACK   0x0060
2040 #define EXT4_DEFM_NOBARRIER     0x0100
2041 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2042 #define EXT4_DEFM_DISCARD       0x0400
2043 #define EXT4_DEFM_NODELALLOC    0x0800
2044
2045 /*
2046  * Default journal batch times
2047  */
2048 #define EXT4_DEF_MIN_BATCH_TIME 0
2049 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2050
2051 /*
2052  * Minimum number of groups in a flexgroup before we separate out
2053  * directories into the first block group of a flexgroup
2054  */
2055 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2056
2057 /*
2058  * Structure of a directory entry
2059  */
2060 #define EXT4_NAME_LEN 255
2061
2062 struct ext4_dir_entry {
2063         __le32  inode;                  /* Inode number */
2064         __le16  rec_len;                /* Directory entry length */
2065         __le16  name_len;               /* Name length */
2066         char    name[EXT4_NAME_LEN];    /* File name */
2067 };
2068
2069 /*
2070  * The new version of the directory entry.  Since EXT4 structures are
2071  * stored in intel byte order, and the name_len field could never be
2072  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2073  * file_type field.
2074  */
2075 struct ext4_dir_entry_2 {
2076         __le32  inode;                  /* Inode number */
2077         __le16  rec_len;                /* Directory entry length */
2078         __u8    name_len;               /* Name length */
2079         __u8    file_type;              /* See file type macros EXT4_FT_* below */
2080         char    name[EXT4_NAME_LEN];    /* File name */
2081 };
2082
2083 /*
2084  * This is a bogus directory entry at the end of each leaf block that
2085  * records checksums.
2086  */
2087 struct ext4_dir_entry_tail {
2088         __le32  det_reserved_zero1;     /* Pretend to be unused */
2089         __le16  det_rec_len;            /* 12 */
2090         __u8    det_reserved_zero2;     /* Zero name length */
2091         __u8    det_reserved_ft;        /* 0xDE, fake file type */
2092         __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
2093 };
2094
2095 #define EXT4_DIRENT_TAIL(block, blocksize) \
2096         ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2097                                         ((blocksize) - \
2098                                          sizeof(struct ext4_dir_entry_tail))))
2099
2100 /*
2101  * Ext4 directory file types.  Only the low 3 bits are used.  The
2102  * other bits are reserved for now.
2103  */
2104 #define EXT4_FT_UNKNOWN         0
2105 #define EXT4_FT_REG_FILE        1
2106 #define EXT4_FT_DIR             2
2107 #define EXT4_FT_CHRDEV          3
2108 #define EXT4_FT_BLKDEV          4
2109 #define EXT4_FT_FIFO            5
2110 #define EXT4_FT_SOCK            6
2111 #define EXT4_FT_SYMLINK         7
2112
2113 #define EXT4_FT_MAX             8
2114
2115 #define EXT4_FT_DIR_CSUM        0xDE
2116
2117 /*
2118  * EXT4_DIR_PAD defines the directory entries boundaries
2119  *
2120  * NOTE: It must be a multiple of 4
2121  */
2122 #define EXT4_DIR_PAD                    4
2123 #define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
2124 #define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
2125                                          ~EXT4_DIR_ROUND)
2126 #define EXT4_MAX_REC_LEN                ((1<<16)-1)
2127
2128 /*
2129  * If we ever get support for fs block sizes > page_size, we'll need
2130  * to remove the #if statements in the next two functions...
2131  */
2132 static inline unsigned int
2133 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2134 {
2135         unsigned len = le16_to_cpu(dlen);
2136
2137 #if (PAGE_SIZE >= 65536)
2138         if (len == EXT4_MAX_REC_LEN || len == 0)
2139                 return blocksize;
2140         return (len & 65532) | ((len & 3) << 16);
2141 #else
2142         return len;
2143 #endif
2144 }
2145
2146 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2147 {
2148         if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2149                 BUG();
2150 #if (PAGE_SIZE >= 65536)
2151         if (len < 65536)
2152                 return cpu_to_le16(len);
2153         if (len == blocksize) {
2154                 if (blocksize == 65536)
2155                         return cpu_to_le16(EXT4_MAX_REC_LEN);
2156                 else
2157                         return cpu_to_le16(0);
2158         }
2159         return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2160 #else
2161         return cpu_to_le16(len);
2162 #endif
2163 }
2164
2165 /*
2166  * Hash Tree Directory indexing
2167  * (c) Daniel Phillips, 2001
2168  */
2169
2170 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2171                     ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2172 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2173                     !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2174 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2175
2176 /* Legal values for the dx_root hash_version field: */
2177
2178 #define DX_HASH_LEGACY                  0
2179 #define DX_HASH_HALF_MD4                1
2180 #define DX_HASH_TEA                     2
2181 #define DX_HASH_LEGACY_UNSIGNED         3
2182 #define DX_HASH_HALF_MD4_UNSIGNED       4
2183 #define DX_HASH_TEA_UNSIGNED            5
2184
2185 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2186                               const void *address, unsigned int length)
2187 {
2188         struct {
2189                 struct shash_desc shash;
2190                 char ctx[4];
2191         } desc;
2192
2193         BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2194
2195         desc.shash.tfm = sbi->s_chksum_driver;
2196         *(u32 *)desc.ctx = crc;
2197
2198         BUG_ON(crypto_shash_update(&desc.shash, address, length));
2199
2200         return *(u32 *)desc.ctx;
2201 }
2202
2203 #ifdef __KERNEL__
2204
2205 /* hash info structure used by the directory hash */
2206 struct dx_hash_info
2207 {
2208         u32             hash;
2209         u32             minor_hash;
2210         int             hash_version;
2211         u32             *seed;
2212 };
2213
2214
2215 /* 32 and 64 bit signed EOF for dx directories */
2216 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2217 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2218
2219
2220 /*
2221  * Control parameters used by ext4_htree_next_block
2222  */
2223 #define HASH_NB_ALWAYS          1
2224
2225 struct ext4_filename {
2226         const struct qstr *usr_fname;
2227         struct fscrypt_str disk_name;
2228         struct dx_hash_info hinfo;
2229 #ifdef CONFIG_FS_ENCRYPTION
2230         struct fscrypt_str crypto_buf;
2231 #endif
2232 #ifdef CONFIG_UNICODE
2233         struct fscrypt_str cf_name;
2234 #endif
2235 };
2236
2237 #define fname_name(p) ((p)->disk_name.name)
2238 #define fname_len(p)  ((p)->disk_name.len)
2239
2240 /*
2241  * Describe an inode's exact location on disk and in memory
2242  */
2243 struct ext4_iloc
2244 {
2245         struct buffer_head *bh;
2246         unsigned long offset;
2247         ext4_group_t block_group;
2248 };
2249
2250 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2251 {
2252         return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2253 }
2254
2255 static inline bool ext4_is_quota_file(struct inode *inode)
2256 {
2257         return IS_NOQUOTA(inode) &&
2258                !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2259 }
2260
2261 /*
2262  * This structure is stuffed into the struct file's private_data field
2263  * for directories.  It is where we put information so that we can do
2264  * readdir operations in hash tree order.
2265  */
2266 struct dir_private_info {
2267         struct rb_root  root;
2268         struct rb_node  *curr_node;
2269         struct fname    *extra_fname;
2270         loff_t          last_pos;
2271         __u32           curr_hash;
2272         __u32           curr_minor_hash;
2273         __u32           next_hash;
2274 };
2275
2276 /* calculate the first block number of the group */
2277 static inline ext4_fsblk_t
2278 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2279 {
2280         return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2281                 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2282 }
2283
2284 /*
2285  * Special error return code only used by dx_probe() and its callers.
2286  */
2287 #define ERR_BAD_DX_DIR  (-(MAX_ERRNO - 1))
2288
2289 /* htree levels for ext4 */
2290 #define EXT4_HTREE_LEVEL_COMPAT 2
2291 #define EXT4_HTREE_LEVEL        3
2292
2293 static inline int ext4_dir_htree_level(struct super_block *sb)
2294 {
2295         return ext4_has_feature_largedir(sb) ?
2296                 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2297 }
2298
2299 /*
2300  * Timeout and state flag for lazy initialization inode thread.
2301  */
2302 #define EXT4_DEF_LI_WAIT_MULT                   10
2303 #define EXT4_DEF_LI_MAX_START_DELAY             5
2304 #define EXT4_LAZYINIT_QUIT                      0x0001
2305 #define EXT4_LAZYINIT_RUNNING                   0x0002
2306
2307 /*
2308  * Lazy inode table initialization info
2309  */
2310 struct ext4_lazy_init {
2311         unsigned long           li_state;
2312         struct list_head        li_request_list;
2313         struct mutex            li_list_mtx;
2314 };
2315
2316 struct ext4_li_request {
2317         struct super_block      *lr_super;
2318         struct ext4_sb_info     *lr_sbi;
2319         ext4_group_t            lr_next_group;
2320         struct list_head        lr_request;
2321         unsigned long           lr_next_sched;
2322         unsigned long           lr_timeout;
2323 };
2324
2325 struct ext4_features {
2326         struct kobject f_kobj;
2327         struct completion f_kobj_unregister;
2328 };
2329
2330 /*
2331  * This structure will be used for multiple mount protection. It will be
2332  * written into the block number saved in the s_mmp_block field in the
2333  * superblock. Programs that check MMP should assume that if
2334  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2335  * to use the filesystem, regardless of how old the timestamp is.
2336  */
2337 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2338 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2339 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2340 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2341
2342 struct mmp_struct {
2343         __le32  mmp_magic;              /* Magic number for MMP */
2344         __le32  mmp_seq;                /* Sequence no. updated periodically */
2345
2346         /*
2347          * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2348          * purposes and do not affect the correctness of the algorithm
2349          */
2350         __le64  mmp_time;               /* Time last updated */
2351         char    mmp_nodename[64];       /* Node which last updated MMP block */
2352         char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
2353
2354         /*
2355          * mmp_check_interval is used to verify if the MMP block has been
2356          * updated on the block device. The value is updated based on the
2357          * maximum time to write the MMP block during an update cycle.
2358          */
2359         __le16  mmp_check_interval;
2360
2361         __le16  mmp_pad1;
2362         __le32  mmp_pad2[226];
2363         __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
2364 };
2365
2366 /* arguments passed to the mmp thread */
2367 struct mmpd_data {
2368         struct buffer_head *bh; /* bh from initial read_mmp_block() */
2369         struct super_block *sb;  /* super block of the fs */
2370 };
2371
2372 /*
2373  * Check interval multiplier
2374  * The MMP block is written every update interval and initially checked every
2375  * update interval x the multiplier (the value is then adapted based on the
2376  * write latency). The reason is that writes can be delayed under load and we
2377  * don't want readers to incorrectly assume that the filesystem is no longer
2378  * in use.
2379  */
2380 #define EXT4_MMP_CHECK_MULT             2UL
2381
2382 /*
2383  * Minimum interval for MMP checking in seconds.
2384  */
2385 #define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
2386
2387 /*
2388  * Maximum interval for MMP checking in seconds.
2389  */
2390 #define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
2391
2392 /*
2393  * Function prototypes
2394  */
2395
2396 /*
2397  * Ok, these declarations are also in <linux/kernel.h> but none of the
2398  * ext4 source programs needs to include it so they are duplicated here.
2399  */
2400 # define NORET_TYPE     /**/
2401 # define ATTRIB_NORET   __attribute__((noreturn))
2402 # define NORET_AND      noreturn,
2403
2404 /* bitmap.c */
2405 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2406 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2407                                 struct ext4_group_desc *gdp,
2408                                 struct buffer_head *bh, int sz);
2409 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2410                                   struct ext4_group_desc *gdp,
2411                                   struct buffer_head *bh, int sz);
2412 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2413                                 struct ext4_group_desc *gdp,
2414                                 struct buffer_head *bh);
2415 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2416                                   struct ext4_group_desc *gdp,
2417                                   struct buffer_head *bh);
2418
2419 /* balloc.c */
2420 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2421                                          ext4_fsblk_t blocknr,
2422                                          ext4_group_t *blockgrpp,
2423                                          ext4_grpblk_t *offsetp);
2424 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2425                                           ext4_fsblk_t block);
2426
2427 extern unsigned int ext4_block_group(struct super_block *sb,
2428                         ext4_fsblk_t blocknr);
2429 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2430                         ext4_fsblk_t blocknr);
2431 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2432 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2433                         ext4_group_t group);
2434 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2435                                          ext4_fsblk_t goal,
2436                                          unsigned int flags,
2437                                          unsigned long *count,
2438                                          int *errp);
2439 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2440                                     s64 nclusters, unsigned int flags);
2441 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2442 extern void ext4_check_blocks_bitmap(struct super_block *);
2443 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2444                                                     ext4_group_t block_group,
2445                                                     struct buffer_head ** bh);
2446 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2447
2448 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2449                                                 ext4_group_t block_group);
2450 extern int ext4_wait_block_bitmap(struct super_block *sb,
2451                                   ext4_group_t block_group,
2452                                   struct buffer_head *bh);
2453 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2454                                                   ext4_group_t block_group);
2455 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2456                                               ext4_group_t block_group,
2457                                               struct ext4_group_desc *gdp);
2458 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2459
2460 #ifdef CONFIG_UNICODE
2461 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2462                                          const struct qstr *iname,
2463                                          struct fscrypt_str *fname);
2464 #endif
2465
2466 #ifdef CONFIG_FS_ENCRYPTION
2467 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2468                                                 const struct fscrypt_name *src)
2469 {
2470         memset(dst, 0, sizeof(*dst));
2471
2472         dst->usr_fname = src->usr_fname;
2473         dst->disk_name = src->disk_name;
2474         dst->hinfo.hash = src->hash;
2475         dst->hinfo.minor_hash = src->minor_hash;
2476         dst->crypto_buf = src->crypto_buf;
2477 }
2478
2479 static inline int ext4_fname_setup_filename(struct inode *dir,
2480                                             const struct qstr *iname,
2481                                             int lookup,
2482                                             struct ext4_filename *fname)
2483 {
2484         struct fscrypt_name name;
2485         int err;
2486
2487         err = fscrypt_setup_filename(dir, iname, lookup, &name);
2488         if (err)
2489                 return err;
2490
2491         ext4_fname_from_fscrypt_name(fname, &name);
2492
2493 #ifdef CONFIG_UNICODE
2494         ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2495 #endif
2496         return 0;
2497 }
2498
2499 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2500                                             struct dentry *dentry,
2501                                             struct ext4_filename *fname)
2502 {
2503         struct fscrypt_name name;
2504         int err;
2505
2506         err = fscrypt_prepare_lookup(dir, dentry, &name);
2507         if (err)
2508                 return err;
2509
2510         ext4_fname_from_fscrypt_name(fname, &name);
2511
2512 #ifdef CONFIG_UNICODE
2513         ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2514 #endif
2515         return 0;
2516 }
2517
2518 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2519 {
2520         struct fscrypt_name name;
2521
2522         name.crypto_buf = fname->crypto_buf;
2523         fscrypt_free_filename(&name);
2524
2525         fname->crypto_buf.name = NULL;
2526         fname->usr_fname = NULL;
2527         fname->disk_name.name = NULL;
2528
2529 #ifdef CONFIG_UNICODE
2530         kfree(fname->cf_name.name);
2531         fname->cf_name.name = NULL;
2532 #endif
2533 }
2534 #else /* !CONFIG_FS_ENCRYPTION */
2535 static inline int ext4_fname_setup_filename(struct inode *dir,
2536                                             const struct qstr *iname,
2537                                             int lookup,
2538                                             struct ext4_filename *fname)
2539 {
2540         fname->usr_fname = iname;
2541         fname->disk_name.name = (unsigned char *) iname->name;
2542         fname->disk_name.len = iname->len;
2543
2544 #ifdef CONFIG_UNICODE
2545         ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2546 #endif
2547
2548         return 0;
2549 }
2550
2551 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2552                                             struct dentry *dentry,
2553                                             struct ext4_filename *fname)
2554 {
2555         return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2556 }
2557
2558 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2559 {
2560 #ifdef CONFIG_UNICODE
2561         kfree(fname->cf_name.name);
2562         fname->cf_name.name = NULL;
2563 #endif
2564 }
2565 #endif /* !CONFIG_FS_ENCRYPTION */
2566
2567 /* dir.c */
2568 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2569                                   struct file *,
2570                                   struct ext4_dir_entry_2 *,
2571                                   struct buffer_head *, char *, int,
2572                                   unsigned int);
2573 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
2574         unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2575                                         (de), (bh), (buf), (size), (offset)))
2576 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2577                                 __u32 minor_hash,
2578                                 struct ext4_dir_entry_2 *dirent,
2579                                 struct fscrypt_str *ent_name);
2580 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2581 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2582                              struct buffer_head *bh,
2583                              void *buf, int buf_size,
2584                              struct ext4_filename *fname,
2585                              struct ext4_dir_entry_2 **dest_de);
2586 void ext4_insert_dentry(struct inode *inode,
2587                         struct ext4_dir_entry_2 *de,
2588                         int buf_size,
2589                         struct ext4_filename *fname);
2590 static inline void ext4_update_dx_flag(struct inode *inode)
2591 {
2592         if (!ext4_has_feature_dir_index(inode->i_sb)) {
2593                 /* ext4_iget() should have caught this... */
2594                 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2595                 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2596         }
2597 }
2598 static const unsigned char ext4_filetype_table[] = {
2599         DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2600 };
2601
2602 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2603 {
2604         if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2605                 return DT_UNKNOWN;
2606
2607         return ext4_filetype_table[filetype];
2608 }
2609 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2610                              void *buf, int buf_size);
2611
2612 /* fsync.c */
2613 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2614
2615 /* hash.c */
2616 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2617                           struct dx_hash_info *hinfo);
2618
2619 /* ialloc.c */
2620 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2621                                       const struct qstr *qstr, __u32 goal,
2622                                       uid_t *owner, __u32 i_flags,
2623                                       int handle_type, unsigned int line_no,
2624                                       int nblocks);
2625
2626 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2627         __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2628                          i_flags, 0, 0, 0)
2629 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2630                                     type, nblocks)                  \
2631         __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2632                          0, (type), __LINE__, (nblocks))
2633
2634
2635 extern void ext4_free_inode(handle_t *, struct inode *);
2636 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2637 extern unsigned long ext4_count_free_inodes(struct super_block *);
2638 extern unsigned long ext4_count_dirs(struct super_block *);
2639 extern void ext4_check_inodes_bitmap(struct super_block *);
2640 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2641 extern int ext4_init_inode_table(struct super_block *sb,
2642                                  ext4_group_t group, int barrier);
2643 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2644
2645 /* mballoc.c */
2646 extern const struct seq_operations ext4_mb_seq_groups_ops;
2647 extern long ext4_mb_stats;
2648 extern long ext4_mb_max_to_scan;
2649 extern int ext4_mb_init(struct super_block *);
2650 extern int ext4_mb_release(struct super_block *);
2651 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2652                                 struct ext4_allocation_request *, int *);
2653 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2654 extern void ext4_discard_preallocations(struct inode *);
2655 extern int __init ext4_init_mballoc(void);
2656 extern void ext4_exit_mballoc(void);
2657 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2658                              struct buffer_head *bh, ext4_fsblk_t block,
2659                              unsigned long count, int flags);
2660 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2661                                    ext4_group_t ngroups);
2662 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2663                 ext4_group_t i, struct ext4_group_desc *desc);
2664 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2665                                 ext4_fsblk_t block, unsigned long count);
2666 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2667 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2668
2669 /* inode.c */
2670 int ext4_inode_is_fast_symlink(struct inode *inode);
2671 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2672 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2673 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2674                      bool wait, struct buffer_head **bhs);
2675 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2676                              struct buffer_head *bh_result, int create);
2677 int ext4_get_block(struct inode *inode, sector_t iblock,
2678                    struct buffer_head *bh_result, int create);
2679 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2680                            struct buffer_head *bh, int create);
2681 int ext4_walk_page_buffers(handle_t *handle,
2682                            struct buffer_head *head,
2683                            unsigned from,
2684                            unsigned to,
2685                            int *partial,
2686                            int (*fn)(handle_t *handle,
2687                                      struct buffer_head *bh));
2688 int do_journal_get_write_access(handle_t *handle,
2689                                 struct buffer_head *bh);
2690 #define FALL_BACK_TO_NONDELALLOC 1
2691 #define CONVERT_INLINE_DATA      2
2692
2693 typedef enum {
2694         EXT4_IGET_NORMAL =      0,
2695         EXT4_IGET_SPECIAL =     0x0001, /* OK to iget a system inode */
2696         EXT4_IGET_HANDLE =      0x0002  /* Inode # is from a handle */
2697 } ext4_iget_flags;
2698
2699 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2700                                  ext4_iget_flags flags, const char *function,
2701                                  unsigned int line);
2702
2703 #define ext4_iget(sb, ino, flags) \
2704         __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2705
2706 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2707 extern int  ext4_setattr(struct dentry *, struct iattr *);
2708 extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2709 extern void ext4_evict_inode(struct inode *);
2710 extern void ext4_clear_inode(struct inode *);
2711 extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2712 extern int  ext4_sync_inode(handle_t *, struct inode *);
2713 extern void ext4_dirty_inode(struct inode *, int);
2714 extern int ext4_change_inode_journal_flag(struct inode *, int);
2715 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2716 extern int ext4_inode_attach_jinode(struct inode *inode);
2717 extern int ext4_can_truncate(struct inode *inode);
2718 extern int ext4_truncate(struct inode *);
2719 extern int ext4_break_layouts(struct inode *);
2720 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2721 extern void ext4_set_inode_flags(struct inode *, bool init);
2722 extern int ext4_alloc_da_blocks(struct inode *inode);
2723 extern void ext4_set_aops(struct inode *inode);
2724 extern int ext4_writepage_trans_blocks(struct inode *);
2725 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2726 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2727                              loff_t lstart, loff_t lend);
2728 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2729 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2730 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2731 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2732 extern void ext4_da_release_space(struct inode *inode, int to_free);
2733 extern void ext4_da_update_reserve_space(struct inode *inode,
2734                                         int used, int quota_claim);
2735 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2736                               ext4_fsblk_t pblk, ext4_lblk_t len);
2737
2738 /* indirect.c */
2739 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2740                                 struct ext4_map_blocks *map, int flags);
2741 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2742 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2743 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2744                                  ext4_lblk_t start, ext4_lblk_t end);
2745
2746 /* ioctl.c */
2747 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2748 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2749
2750 /* migrate.c */
2751 extern int ext4_ext_migrate(struct inode *);
2752 extern int ext4_ind_migrate(struct inode *inode);
2753
2754 /* namei.c */
2755 extern int ext4_dirblock_csum_verify(struct inode *inode,
2756                                      struct buffer_head *bh);
2757 extern int ext4_orphan_add(handle_t *, struct inode *);
2758 extern int ext4_orphan_del(handle_t *, struct inode *);
2759 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2760                                 __u32 start_minor_hash, __u32 *next_hash);
2761 extern int ext4_search_dir(struct buffer_head *bh,
2762                            char *search_buf,
2763                            int buf_size,
2764                            struct inode *dir,
2765                            struct ext4_filename *fname,
2766                            unsigned int offset,
2767                            struct ext4_dir_entry_2 **res_dir);
2768 extern int ext4_generic_delete_entry(handle_t *handle,
2769                                      struct inode *dir,
2770                                      struct ext4_dir_entry_2 *de_del,
2771                                      struct buffer_head *bh,
2772                                      void *entry_buf,
2773                                      int buf_size,
2774                                      int csum_size);
2775 extern bool ext4_empty_dir(struct inode *inode);
2776
2777 /* resize.c */
2778 extern void ext4_kvfree_array_rcu(void *to_free);
2779 extern int ext4_group_add(struct super_block *sb,
2780                                 struct ext4_new_group_data *input);
2781 extern int ext4_group_extend(struct super_block *sb,
2782                                 struct ext4_super_block *es,
2783                                 ext4_fsblk_t n_blocks_count);
2784 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2785
2786 /* super.c */
2787 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2788                                          sector_t block, int op_flags);
2789 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2790 extern int ext4_calculate_overhead(struct super_block *sb);
2791 extern void ext4_superblock_csum_set(struct super_block *sb);
2792 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2793                                     ext4_group_t ngroup);
2794 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2795                                      char nbuf[16]);
2796 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2797                                              ext4_group_t block_group,
2798                                              unsigned int flags);
2799
2800 extern __printf(6, 7)
2801 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2802                   const char *, ...);
2803 extern __printf(6, 7)
2804 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2805                         ext4_fsblk_t, int, const char *, ...);
2806 extern __printf(5, 6)
2807 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2808                      const char *, ...);
2809 extern void __ext4_std_error(struct super_block *, const char *,
2810                              unsigned int, int);
2811 extern __printf(5, 6)
2812 void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2813                   const char *, ...);
2814 extern __printf(4, 5)
2815 void __ext4_warning(struct super_block *, const char *, unsigned int,
2816                     const char *, ...);
2817 extern __printf(4, 5)
2818 void __ext4_warning_inode(const struct inode *inode, const char *function,
2819                           unsigned int line, const char *fmt, ...);
2820 extern __printf(3, 4)
2821 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2822 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2823                            const char *, unsigned int, const char *);
2824 extern __printf(7, 8)
2825 void __ext4_grp_locked_error(const char *, unsigned int,
2826                              struct super_block *, ext4_group_t,
2827                              unsigned long, ext4_fsblk_t,
2828                              const char *, ...);
2829
2830 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2831         ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2832
2833 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)                     \
2834         __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2835
2836 #define ext4_error_inode_block(inode, block, err, fmt, a...)            \
2837         __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
2838                            (fmt), ## a)
2839
2840 #define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
2841         ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2842
2843 #ifdef CONFIG_PRINTK
2844
2845 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2846         __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2847 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)   \
2848         __ext4_error_inode((inode), (func), (line), (block),            \
2849                            (err), (fmt), ##__VA_ARGS__)
2850 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2851         __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2852 #define ext4_error(sb, fmt, ...)                                        \
2853         __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2854 #define ext4_error_err(sb, err, fmt, ...)                               \
2855         __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2856 #define ext4_abort(sb, err, fmt, ...)                                   \
2857         __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2858 #define ext4_warning(sb, fmt, ...)                                      \
2859         __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2860 #define ext4_warning_inode(inode, fmt, ...)                             \
2861         __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2862 #define ext4_msg(sb, level, fmt, ...)                           \
2863         __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2864 #define dump_mmp_msg(sb, mmp, msg)                                      \
2865         __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2866 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2867         __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2868                                 fmt, ##__VA_ARGS__)
2869
2870 #else
2871
2872 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2873 do {                                                                    \
2874         no_printk(fmt, ##__VA_ARGS__);                                  \
2875         __ext4_error_inode(inode, "", 0, block, 0, " ");                \
2876 } while (0)
2877 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)   \
2878 do {                                                                    \
2879         no_printk(fmt, ##__VA_ARGS__);                                  \
2880         __ext4_error_inode(inode, "", 0, block, err, " ");              \
2881 } while (0)
2882 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2883 do {                                                                    \
2884         no_printk(fmt, ##__VA_ARGS__);                                  \
2885         __ext4_error_file(file, "", 0, block, " ");                     \
2886 } while (0)
2887 #define ext4_error(sb, fmt, ...)                                        \
2888 do {                                                                    \
2889         no_printk(fmt, ##__VA_ARGS__);                                  \
2890         __ext4_error(sb, "", 0, 0, 0, " ");                             \
2891 } while (0)
2892 #define ext4_error_err(sb, err, fmt, ...)                               \
2893 do {                                                                    \
2894         no_printk(fmt, ##__VA_ARGS__);                                  \
2895         __ext4_error(sb, "", 0, err, 0, " ");                           \
2896 } while (0)
2897 #define ext4_abort(sb, err, fmt, ...)                                   \
2898 do {                                                                    \
2899         no_printk(fmt, ##__VA_ARGS__);                                  \
2900         __ext4_abort(sb, "", 0, err, " ");                              \
2901 } while (0)
2902 #define ext4_warning(sb, fmt, ...)                                      \
2903 do {                                                                    \
2904         no_printk(fmt, ##__VA_ARGS__);                                  \
2905         __ext4_warning(sb, "", 0, " ");                                 \
2906 } while (0)
2907 #define ext4_warning_inode(inode, fmt, ...)                             \
2908 do {                                                                    \
2909         no_printk(fmt, ##__VA_ARGS__);                                  \
2910         __ext4_warning_inode(inode, "", 0, " ");                        \
2911 } while (0)
2912 #define ext4_msg(sb, level, fmt, ...)                                   \
2913 do {                                                                    \
2914         no_printk(fmt, ##__VA_ARGS__);                                  \
2915         __ext4_msg(sb, "", " ");                                        \
2916 } while (0)
2917 #define dump_mmp_msg(sb, mmp, msg)                                      \
2918         __dump_mmp_msg(sb, mmp, "", 0, "")
2919 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2920 do {                                                                    \
2921         no_printk(fmt, ##__VA_ARGS__);                          \
2922         __ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");       \
2923 } while (0)
2924
2925 #endif
2926
2927 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2928                                         __u32 compat);
2929 extern int ext4_update_rocompat_feature(handle_t *handle,
2930                                         struct super_block *sb, __u32 rocompat);
2931 extern int ext4_update_incompat_feature(handle_t *handle,
2932                                         struct super_block *sb, __u32 incompat);
2933 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2934                                       struct ext4_group_desc *bg);
2935 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2936                                       struct ext4_group_desc *bg);
2937 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2938                                      struct ext4_group_desc *bg);
2939 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2940                                       struct ext4_group_desc *bg);
2941 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2942                                  struct ext4_group_desc *bg);
2943 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2944                                 struct ext4_group_desc *bg);
2945 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2946                                    struct ext4_group_desc *bg);
2947 extern void ext4_block_bitmap_set(struct super_block *sb,
2948                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2949 extern void ext4_inode_bitmap_set(struct super_block *sb,
2950                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2951 extern void ext4_inode_table_set(struct super_block *sb,
2952                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2953 extern void ext4_free_group_clusters_set(struct super_block *sb,
2954                                          struct ext4_group_desc *bg,
2955                                          __u32 count);
2956 extern void ext4_free_inodes_set(struct super_block *sb,
2957                                 struct ext4_group_desc *bg, __u32 count);
2958 extern void ext4_used_dirs_set(struct super_block *sb,
2959                                 struct ext4_group_desc *bg, __u32 count);
2960 extern void ext4_itable_unused_set(struct super_block *sb,
2961                                    struct ext4_group_desc *bg, __u32 count);
2962 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2963                                        struct ext4_group_desc *gdp);
2964 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2965                                      struct ext4_group_desc *gdp);
2966 extern int ext4_register_li_request(struct super_block *sb,
2967                                     ext4_group_t first_not_zeroed);
2968
2969 static inline int ext4_has_metadata_csum(struct super_block *sb)
2970 {
2971         WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2972                      !EXT4_SB(sb)->s_chksum_driver);
2973
2974         return ext4_has_feature_metadata_csum(sb) &&
2975                (EXT4_SB(sb)->s_chksum_driver != NULL);
2976 }
2977
2978 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2979 {
2980         return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2981 }
2982
2983 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2984 {
2985         return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2986                 le32_to_cpu(es->s_blocks_count_lo);
2987 }
2988
2989 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2990 {
2991         return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2992                 le32_to_cpu(es->s_r_blocks_count_lo);
2993 }
2994
2995 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2996 {
2997         return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2998                 le32_to_cpu(es->s_free_blocks_count_lo);
2999 }
3000
3001 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3002                                          ext4_fsblk_t blk)
3003 {
3004         es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3005         es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3006 }
3007
3008 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3009                                               ext4_fsblk_t blk)
3010 {
3011         es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3012         es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3013 }
3014
3015 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3016                                            ext4_fsblk_t blk)
3017 {
3018         es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3019         es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3020 }
3021
3022 static inline loff_t ext4_isize(struct super_block *sb,
3023                                 struct ext4_inode *raw_inode)
3024 {
3025         if (ext4_has_feature_largedir(sb) ||
3026             S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3027                 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3028                         le32_to_cpu(raw_inode->i_size_lo);
3029
3030         return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3031 }
3032
3033 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3034 {
3035         raw_inode->i_size_lo = cpu_to_le32(i_size);
3036         raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3037 }
3038
3039 static inline
3040 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3041                                             ext4_group_t group)
3042 {
3043          struct ext4_group_info **grp_info;
3044          long indexv, indexh;
3045          BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3046          indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3047          indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3048          grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3049          return grp_info[indexh];
3050 }
3051
3052 /*
3053  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3054  * the locking protocol documented in the comments of ext4_group_add()
3055  * in resize.c
3056  */
3057 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3058 {
3059         ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
3060
3061         smp_rmb();
3062         return ngroups;
3063 }
3064
3065 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3066                                              ext4_group_t block_group)
3067 {
3068         return block_group >> sbi->s_log_groups_per_flex;
3069 }
3070
3071 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3072 {
3073         return 1 << sbi->s_log_groups_per_flex;
3074 }
3075
3076 #define ext4_std_error(sb, errno)                               \
3077 do {                                                            \
3078         if ((errno))                                            \
3079                 __ext4_std_error((sb), __func__, __LINE__, (errno));    \
3080 } while (0)
3081
3082 #ifdef CONFIG_SMP
3083 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3084  * counters. So we need to make sure we have free clusters more
3085  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3086  */
3087 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3088 #else
3089 #define EXT4_FREECLUSTERS_WATERMARK 0
3090 #endif
3091
3092 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3093 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3094 {
3095         WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3096                      !inode_is_locked(inode));
3097         down_write(&EXT4_I(inode)->i_data_sem);
3098         if (newsize > EXT4_I(inode)->i_disksize)
3099                 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3100         up_write(&EXT4_I(inode)->i_data_sem);
3101 }
3102
3103 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3104 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3105 {
3106         int changed = 0;
3107
3108         if (newsize > inode->i_size) {
3109                 i_size_write(inode, newsize);
3110                 changed = 1;
3111         }
3112         if (newsize > EXT4_I(inode)->i_disksize) {
3113                 ext4_update_i_disksize(inode, newsize);
3114                 changed |= 2;
3115         }
3116         return changed;
3117 }
3118
3119 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3120                                       loff_t len);
3121
3122 struct ext4_group_info {
3123         unsigned long   bb_state;
3124         struct rb_root  bb_free_root;
3125         ext4_grpblk_t   bb_first_free;  /* first free block */
3126         ext4_grpblk_t   bb_free;        /* total free blocks */
3127         ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
3128         ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
3129         struct          list_head bb_prealloc_list;
3130 #ifdef DOUBLE_CHECK
3131         void            *bb_bitmap;
3132 #endif
3133         struct rw_semaphore alloc_sem;
3134         ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
3135                                          * regions, index is order.
3136                                          * bb_counters[3] = 5 means
3137                                          * 5 free 8-block regions. */
3138 };
3139
3140 #define EXT4_GROUP_INFO_NEED_INIT_BIT           0
3141 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
3142 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT     2
3143 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT     3
3144 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT         \
3145         (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3146 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT         \
3147         (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3148
3149 #define EXT4_MB_GRP_NEED_INIT(grp)      \
3150         (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3151 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)        \
3152         (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3153 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)        \
3154         (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3155
3156 #define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
3157         (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3158 #define EXT4_MB_GRP_SET_TRIMMED(grp)    \
3159         (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3160 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
3161         (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3162
3163 #define EXT4_MAX_CONTENTION             8
3164 #define EXT4_CONTENTION_THRESHOLD       2
3165
3166 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3167                                               ext4_group_t group)
3168 {
3169         return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3170 }
3171
3172 /*
3173  * Returns true if the filesystem is busy enough that attempts to
3174  * access the block group locks has run into contention.
3175  */
3176 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3177 {
3178         return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3179 }
3180
3181 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3182 {
3183         spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3184         if (spin_trylock(lock))
3185                 /*
3186                  * We're able to grab the lock right away, so drop the
3187                  * lock contention counter.
3188                  */
3189                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3190         else {
3191                 /*
3192                  * The lock is busy, so bump the contention counter,
3193                  * and then wait on the spin lock.
3194                  */
3195                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3196                                   EXT4_MAX_CONTENTION);
3197                 spin_lock(lock);
3198         }
3199 }
3200
3201 static inline void ext4_unlock_group(struct super_block *sb,
3202                                         ext4_group_t group)
3203 {
3204         spin_unlock(ext4_group_lock_ptr(sb, group));
3205 }
3206
3207 /*
3208  * Block validity checking
3209  */
3210 #define ext4_check_indirect_blockref(inode, bh)                         \
3211         ext4_check_blockref(__func__, __LINE__, inode,                  \
3212                             (__le32 *)(bh)->b_data,                     \
3213                             EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3214
3215 #define ext4_ind_check_inode(inode)                                     \
3216         ext4_check_blockref(__func__, __LINE__, inode,                  \
3217                             EXT4_I(inode)->i_data,                      \
3218                             EXT4_NDIR_BLOCKS)
3219
3220 /*
3221  * Inodes and files operations
3222  */
3223
3224 /* dir.c */
3225 extern const struct file_operations ext4_dir_operations;
3226
3227 #ifdef CONFIG_UNICODE
3228 extern const struct dentry_operations ext4_dentry_ops;
3229 #endif
3230
3231 /* file.c */
3232 extern const struct inode_operations ext4_file_inode_operations;
3233 extern const struct file_operations ext4_file_operations;
3234 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3235
3236 /* inline.c */
3237 extern int ext4_get_max_inline_size(struct inode *inode);
3238 extern int ext4_find_inline_data_nolock(struct inode *inode);
3239 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3240                                  unsigned int len);
3241 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3242
3243 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3244 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3245                                          struct inode *inode,
3246                                          loff_t pos, unsigned len,
3247                                          unsigned flags,
3248                                          struct page **pagep);
3249 extern int ext4_write_inline_data_end(struct inode *inode,
3250                                       loff_t pos, unsigned len,
3251                                       unsigned copied,
3252                                       struct page *page);
3253 extern struct buffer_head *
3254 ext4_journalled_write_inline_data(struct inode *inode,
3255                                   unsigned len,
3256                                   struct page *page);
3257 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3258                                            struct inode *inode,
3259                                            loff_t pos, unsigned len,
3260                                            unsigned flags,
3261                                            struct page **pagep,
3262                                            void **fsdata);
3263 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3264                                          unsigned len, unsigned copied,
3265                                          struct page *page);
3266 extern int ext4_try_add_inline_entry(handle_t *handle,
3267                                      struct ext4_filename *fname,
3268                                      struct inode *dir, struct inode *inode);
3269 extern int ext4_try_create_inline_dir(handle_t *handle,
3270                                       struct inode *parent,
3271                                       struct inode *inode);
3272 extern int ext4_read_inline_dir(struct file *filp,
3273                                 struct dir_context *ctx,
3274                                 int *has_inline_data);
3275 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3276                                   struct inode *dir, ext4_lblk_t block,
3277                                   struct dx_hash_info *hinfo,
3278                                   __u32 start_hash, __u32 start_minor_hash,
3279                                   int *has_inline_data);
3280 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3281                                         struct ext4_filename *fname,
3282                                         struct ext4_dir_entry_2 **res_dir,
3283                                         int *has_inline_data);
3284 extern int ext4_delete_inline_entry(handle_t *handle,
3285                                     struct inode *dir,
3286                                     struct ext4_dir_entry_2 *de_del,
3287                                     struct buffer_head *bh,
3288                                     int *has_inline_data);
3289 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3290 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3291                                         struct ext4_dir_entry_2 **parent_de,
3292                                         int *retval);
3293 extern int ext4_inline_data_fiemap(struct inode *inode,
3294                                    struct fiemap_extent_info *fieinfo,
3295                                    int *has_inline, __u64 start, __u64 len);
3296
3297 struct iomap;
3298 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3299
3300 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3301
3302 extern int ext4_convert_inline_data(struct inode *inode);
3303
3304 static inline int ext4_has_inline_data(struct inode *inode)
3305 {
3306         return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3307                EXT4_I(inode)->i_inline_off;
3308 }
3309
3310 /* namei.c */
3311 extern const struct inode_operations ext4_dir_inode_operations;
3312 extern const struct inode_operations ext4_special_inode_operations;
3313 extern struct dentry *ext4_get_parent(struct dentry *child);
3314 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3315                                  struct ext4_dir_entry_2 *de,
3316                                  int blocksize, int csum_size,
3317                                  unsigned int parent_ino, int dotdot_real_len);
3318 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3319                                         unsigned int blocksize);
3320 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3321                                       struct buffer_head *bh);
3322 extern int ext4_ci_compare(const struct inode *parent,
3323                            const struct qstr *fname,
3324                            const struct qstr *entry, bool quick);
3325
3326 #define S_SHIFT 12
3327 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3328         [S_IFREG >> S_SHIFT]    = EXT4_FT_REG_FILE,
3329         [S_IFDIR >> S_SHIFT]    = EXT4_FT_DIR,
3330         [S_IFCHR >> S_SHIFT]    = EXT4_FT_CHRDEV,
3331         [S_IFBLK >> S_SHIFT]    = EXT4_FT_BLKDEV,
3332         [S_IFIFO >> S_SHIFT]    = EXT4_FT_FIFO,
3333         [S_IFSOCK >> S_SHIFT]   = EXT4_FT_SOCK,
3334         [S_IFLNK >> S_SHIFT]    = EXT4_FT_SYMLINK,
3335 };
3336
3337 static inline void ext4_set_de_type(struct super_block *sb,
3338                                 struct ext4_dir_entry_2 *de,
3339                                 umode_t mode) {
3340         if (ext4_has_feature_filetype(sb))
3341                 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3342 }
3343
3344 /* readpages.c */
3345 extern int ext4_mpage_readpages(struct inode *inode,
3346                 struct readahead_control *rac, struct page *page);
3347 extern int __init ext4_init_post_read_processing(void);
3348 extern void ext4_exit_post_read_processing(void);
3349
3350 /* symlink.c */
3351 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3352 extern const struct inode_operations ext4_symlink_inode_operations;
3353 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3354
3355 /* sysfs.c */
3356 extern int ext4_register_sysfs(struct super_block *sb);
3357 extern void ext4_unregister_sysfs(struct super_block *sb);
3358 extern int __init ext4_init_sysfs(void);
3359 extern void ext4_exit_sysfs(void);
3360
3361 /* block_validity */
3362 extern void ext4_release_system_zone(struct super_block *sb);
3363 extern int ext4_setup_system_zone(struct super_block *sb);
3364 extern int __init ext4_init_system_zone(void);
3365 extern void ext4_exit_system_zone(void);
3366 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3367                                  ext4_fsblk_t start_blk,
3368                                  unsigned int count);
3369 extern int ext4_check_blockref(const char *, unsigned int,
3370                                struct inode *, __le32 *, unsigned int);
3371
3372 /* extents.c */
3373 struct ext4_ext_path;
3374 struct ext4_extent;
3375
3376 /*
3377  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3378  * __le32.
3379  */
3380 #define EXT_MAX_BLOCKS  0xffffffff
3381
3382 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3383 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3384 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3385                                struct ext4_map_blocks *map, int flags);
3386 extern int ext4_ext_truncate(handle_t *, struct inode *);
3387 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3388                                  ext4_lblk_t end);
3389 extern void ext4_ext_init(struct super_block *);
3390 extern void ext4_ext_release(struct super_block *);
3391 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3392                           loff_t len);
3393 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3394                                           loff_t offset, ssize_t len);
3395 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3396                                              ext4_io_end_t *io_end);
3397 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3398                            struct ext4_map_blocks *map, int flags);
3399 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3400                                                    int num,
3401                                                    struct ext4_ext_path *path);
3402 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3403                                   struct ext4_ext_path **,
3404                                   struct ext4_extent *, int);
3405 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3406                                               struct ext4_ext_path **,
3407                                               int flags);
3408 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3409 extern int ext4_ext_check_inode(struct inode *inode);
3410 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3411 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3412                         __u64 start, __u64 len);
3413 extern int ext4_get_es_cache(struct inode *inode,
3414                              struct fiemap_extent_info *fieinfo,
3415                              __u64 start, __u64 len);
3416 extern int ext4_ext_precache(struct inode *inode);
3417 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3418                                 struct inode *inode2, ext4_lblk_t lblk1,
3419                              ext4_lblk_t lblk2,  ext4_lblk_t count,
3420                              int mark_unwritten,int *err);
3421 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3422 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3423                                        int check_cred, int restart_cred,
3424                                        int revoke_cred);
3425
3426
3427 /* move_extent.c */
3428 extern void ext4_double_down_write_data_sem(struct inode *first,
3429                                             struct inode *second);
3430 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3431                                           struct inode *donor_inode);
3432 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3433                              __u64 start_orig, __u64 start_donor,
3434                              __u64 len, __u64 *moved_len);
3435
3436 /* page-io.c */
3437 extern int __init ext4_init_pageio(void);
3438 extern void ext4_exit_pageio(void);
3439 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3440 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3441 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3442 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3443 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3444                                 struct writeback_control *wbc);
3445 extern void ext4_end_io_rsv_work(struct work_struct *work);
3446 extern void ext4_io_submit(struct ext4_io_submit *io);
3447 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3448                                struct page *page,
3449                                int len,
3450                                struct writeback_control *wbc,
3451                                bool keep_towrite);
3452 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3453 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3454
3455 /* mmp.c */
3456 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3457
3458 /* verity.c */
3459 extern const struct fsverity_operations ext4_verityops;
3460
3461 /*
3462  * Add new method to test whether block and inode bitmaps are properly
3463  * initialized. With uninit_bg reading the block from disk is not enough
3464  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3465  */
3466 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3467
3468 static inline int bitmap_uptodate(struct buffer_head *bh)
3469 {
3470         return (buffer_uptodate(bh) &&
3471                         test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3472 }
3473 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3474 {
3475         set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3476 }
3477
3478 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3479
3480 /* For ioend & aio unwritten conversion wait queues */
3481 #define EXT4_WQ_HASH_SZ         37
3482 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3483                                             EXT4_WQ_HASH_SZ])
3484 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3485
3486 extern int ext4_resize_begin(struct super_block *sb);
3487 extern void ext4_resize_end(struct super_block *sb);
3488
3489 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3490                                               struct ext4_io_end *io_end)
3491 {
3492         if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3493                 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3494                 atomic_inc(&EXT4_I(inode)->i_unwritten);
3495         }
3496 }
3497
3498 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3499 {
3500         struct inode *inode = io_end->inode;
3501
3502         if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3503                 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3504                 /* Wake up anyone waiting on unwritten extent conversion */
3505                 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3506                         wake_up_all(ext4_ioend_wq(inode));
3507         }
3508 }
3509
3510 extern const struct iomap_ops ext4_iomap_ops;
3511 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3512 extern const struct iomap_ops ext4_iomap_report_ops;
3513
3514 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3515 {
3516         /*
3517          * If the buffer has the write error flag, we have failed
3518          * to write out data in the block.  In this  case, we don't
3519          * have to read the block because we may read the old data
3520          * successfully.
3521          */
3522         if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3523                 set_buffer_uptodate(bh);
3524         return buffer_uptodate(bh);
3525 }
3526
3527 #endif  /* __KERNEL__ */
3528
3529 #define EFSBADCRC       EBADMSG         /* Bad CRC detected */
3530 #define EFSCORRUPTED    EUCLEAN         /* Filesystem is corrupted */
3531
3532 #endif  /* _EXT4_H */