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 /* User modifiable flags */
 438 #define EXT4_FL_USER_MODIFIABLE         (EXT4_SECRM_FL | \
 439                                          EXT4_UNRM_FL | \
 440                                          EXT4_COMPR_FL | \
 441                                          EXT4_SYNC_FL | \
 442                                          EXT4_IMMUTABLE_FL | \
 443                                          EXT4_APPEND_FL | \
 444                                          EXT4_NODUMP_FL | \
 445                                          EXT4_NOATIME_FL | \
 446                                          EXT4_JOURNAL_DATA_FL | \
 447                                          EXT4_NOTAIL_FL | \
 448                                          EXT4_DIRSYNC_FL | \
 449                                          EXT4_TOPDIR_FL | \
 450                                          EXT4_EXTENTS_FL | \
 451                                          0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
 452                                          EXT4_DAX_FL | \
 453                                          EXT4_PROJINHERIT_FL | \
 454                                          EXT4_CASEFOLD_FL)
 455
 456 /* User visible flags */
 457 #define EXT4_FL_USER_VISIBLE            (EXT4_FL_USER_MODIFIABLE | \
 458                                          EXT4_DIRTY_FL | \
 459                                          EXT4_COMPRBLK_FL | \
 460                                          EXT4_NOCOMPR_FL | \
 461                                          EXT4_ENCRYPT_FL | \
 462                                          EXT4_INDEX_FL | \
 463                                          EXT4_VERITY_FL | \
 464                                          EXT4_INLINE_DATA_FL)
 465
 466 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
 467 #define EXT4_FL_XFLAG_VISIBLE           (EXT4_SYNC_FL | \
 468                                          EXT4_IMMUTABLE_FL | \
 469                                          EXT4_APPEND_FL | \
 470                                          EXT4_NODUMP_FL | \
 471                                          EXT4_NOATIME_FL | \
 472                                          EXT4_PROJINHERIT_FL | \
 473                                          EXT4_DAX_FL)
 474
 475 /* Flags that should be inherited by new inodes from their parent. */
 476 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
 477                            EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
 478                            EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
 479                            EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
 480                            EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
 481                            EXT4_DAX_FL)
 482
 483 /* Flags that are appropriate for regular files (all but dir-specific ones). */
 484 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
 485                            EXT4_PROJINHERIT_FL))
 486
 487 /* Flags that are appropriate for non-directories/regular files. */
 488 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
 489
 490 /* The only flags that should be swapped */
 491 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
 492
 493 /* Flags which are mutually exclusive to DAX */
 494 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
 495                            EXT4_JOURNAL_DATA_FL)
 496
 497 /* Mask out flags that are inappropriate for the given type of inode. */
 498 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
 499 {
 500         if (S_ISDIR(mode))
 501                 return flags;
 502         else if (S_ISREG(mode))
 503                 return flags & EXT4_REG_FLMASK;
 504         else
 505                 return flags & EXT4_OTHER_FLMASK;
 506 }
 507
 508 /*
 509  * Inode flags used for atomic set/get
 510  */
 511 enum {
 512         EXT4_INODE_SECRM        = 0,    /* Secure deletion */
 513         EXT4_INODE_UNRM         = 1,    /* Undelete */
 514         EXT4_INODE_COMPR        = 2,    /* Compress file */
 515         EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
 516         EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
 517         EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
 518         EXT4_INODE_NODUMP       = 6,    /* do not dump file */
 519         EXT4_INODE_NOATIME      = 7,    /* do not update atime */
 520 /* Reserved for compression usage... */
 521         EXT4_INODE_DIRTY        = 8,
 522         EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
 523         EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
 524         EXT4_INODE_ENCRYPT      = 11,   /* Encrypted file */
 525 /* End compression flags --- maybe not all used */
 526         EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
 527         EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
 528         EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
 529         EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
 530         EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
 531         EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
 532         EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
 533         EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
 534         EXT4_INODE_VERITY       = 20,   /* Verity protected inode */
 535         EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
 536 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
 537         EXT4_INODE_DAX          = 25,   /* Inode is DAX */
 538         EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
 539         EXT4_INODE_PROJINHERIT  = 29,   /* Create with parents projid */
 540         EXT4_INODE_CASEFOLD     = 30,   /* Casefolded directory */
 541         EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
 542 };
 543
 544 /*
 545  * Since it's pretty easy to mix up bit numbers and hex values, we use a
 546  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
 547  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
 548  * any extra space in the compiled kernel image, otherwise, the build will fail.
 549  * It's important that these values are the same, since we are using
 550  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
 551  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
 552  * values found in ext2, ext3 and ext4 filesystems, and of course the values
 553  * defined in e2fsprogs.
 554  *
 555  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
 556  */
 557 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
 558 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
 559
 560 static inline void ext4_check_flag_values(void)
 561 {
 562         CHECK_FLAG_VALUE(SECRM);
 563         CHECK_FLAG_VALUE(UNRM);
 564         CHECK_FLAG_VALUE(COMPR);
 565         CHECK_FLAG_VALUE(SYNC);
 566         CHECK_FLAG_VALUE(IMMUTABLE);
 567         CHECK_FLAG_VALUE(APPEND);
 568         CHECK_FLAG_VALUE(NODUMP);
 569         CHECK_FLAG_VALUE(NOATIME);
 570         CHECK_FLAG_VALUE(DIRTY);
 571         CHECK_FLAG_VALUE(COMPRBLK);
 572         CHECK_FLAG_VALUE(NOCOMPR);
 573         CHECK_FLAG_VALUE(ENCRYPT);
 574         CHECK_FLAG_VALUE(INDEX);
 575         CHECK_FLAG_VALUE(IMAGIC);
 576         CHECK_FLAG_VALUE(JOURNAL_DATA);
 577         CHECK_FLAG_VALUE(NOTAIL);
 578         CHECK_FLAG_VALUE(DIRSYNC);
 579         CHECK_FLAG_VALUE(TOPDIR);
 580         CHECK_FLAG_VALUE(HUGE_FILE);
 581         CHECK_FLAG_VALUE(EXTENTS);
 582         CHECK_FLAG_VALUE(VERITY);
 583         CHECK_FLAG_VALUE(EA_INODE);
 584         CHECK_FLAG_VALUE(INLINE_DATA);
 585         CHECK_FLAG_VALUE(PROJINHERIT);
 586         CHECK_FLAG_VALUE(CASEFOLD);
 587         CHECK_FLAG_VALUE(RESERVED);
 588 }
 589
 590 /* Used to pass group descriptor data when online resize is done */
 591 struct ext4_new_group_input {
 592         __u32 group;            /* Group number for this data */
 593         __u64 block_bitmap;     /* Absolute block number of block bitmap */
 594         __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
 595         __u64 inode_table;      /* Absolute block number of inode table start */
 596         __u32 blocks_count;     /* Total number of blocks in this group */
 597         __u16 reserved_blocks;  /* Number of reserved blocks in this group */
 598         __u16 unused;
 599 };
 600
 601 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 602 struct compat_ext4_new_group_input {
 603         u32 group;
 604         compat_u64 block_bitmap;
 605         compat_u64 inode_bitmap;
 606         compat_u64 inode_table;
 607         u32 blocks_count;
 608         u16 reserved_blocks;
 609         u16 unused;
 610 };
 611 #endif
 612
 613 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
 614 struct ext4_new_group_data {
 615         __u32 group;
 616         __u64 block_bitmap;
 617         __u64 inode_bitmap;
 618         __u64 inode_table;
 619         __u32 blocks_count;
 620         __u16 reserved_blocks;
 621         __u16 mdata_blocks;
 622         __u32 free_clusters_count;
 623 };
 624
 625 /* Indexes used to index group tables in ext4_new_group_data */
 626 enum {
 627         BLOCK_BITMAP = 0,       /* block bitmap */
 628         INODE_BITMAP,           /* inode bitmap */
 629         INODE_TABLE,            /* inode tables */
 630         GROUP_TABLE_COUNT,
 631 };
 632
 633 /*
 634  * Flags used by ext4_map_blocks()
 635  */
 636         /* Allocate any needed blocks and/or convert an unwritten
 637            extent to be an initialized ext4 */
 638 #define EXT4_GET_BLOCKS_CREATE                  0x0001
 639         /* Request the creation of an unwritten extent */
 640 #define EXT4_GET_BLOCKS_UNWRIT_EXT              0x0002
 641 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT       (EXT4_GET_BLOCKS_UNWRIT_EXT|\
 642                                                  EXT4_GET_BLOCKS_CREATE)
 643         /* Caller is from the delayed allocation writeout path
 644          * finally doing the actual allocation of delayed blocks */
 645 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
 646         /* caller is from the direct IO path, request to creation of an
 647         unwritten extents if not allocated, split the unwritten
 648         extent if blocks has been preallocated already*/
 649 #define EXT4_GET_BLOCKS_PRE_IO                  0x0008
 650 #define EXT4_GET_BLOCKS_CONVERT                 0x0010
 651 #define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
 652                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
 653         /* Convert extent to initialized after IO complete */
 654 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
 655                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
 656         /* Eventual metadata allocation (due to growing extent tree)
 657          * should not fail, so try to use reserved blocks for that.*/
 658 #define EXT4_GET_BLOCKS_METADATA_NOFAIL         0x0020
 659         /* Don't normalize allocation size (used for fallocate) */
 660 #define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
 661         /* Convert written extents to unwritten */
 662 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN       0x0100
 663         /* Write zeros to newly created written extents */
 664 #define EXT4_GET_BLOCKS_ZERO                    0x0200
 665 #define EXT4_GET_BLOCKS_CREATE_ZERO             (EXT4_GET_BLOCKS_CREATE |\
 666                                         EXT4_GET_BLOCKS_ZERO)
 667         /* Caller will submit data before dropping transaction handle. This
 668          * allows jbd2 to avoid submitting data before commit. */
 669 #define EXT4_GET_BLOCKS_IO_SUBMIT               0x0400
 670
 671 /*
 672  * The bit position of these flags must not overlap with any of the
 673  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
 674  * read_extent_tree_block(), ext4_split_extent_at(),
 675  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
 676  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
 677  * caching the extents when reading from the extent tree while a
 678  * truncate or punch hole operation is in progress.
 679  */
 680 #define EXT4_EX_NOCACHE                         0x40000000
 681 #define EXT4_EX_FORCE_CACHE                     0x20000000
 682 #define EXT4_EX_NOFAIL                          0x10000000
 683
 684 /*
 685  * Flags used by ext4_free_blocks
 686  */
 687 #define EXT4_FREE_BLOCKS_METADATA               0x0001
 688 #define EXT4_FREE_BLOCKS_FORGET                 0x0002
 689 #define EXT4_FREE_BLOCKS_VALIDATED              0x0004
 690 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE         0x0008
 691 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
 692 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
 693 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
 694
 695 /*
 696  * ioctl commands
 697  */
 698 #define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
 699 #define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
 700 #define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
 701 #define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
 702 #define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
 703 #define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
 704 #define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
 705 #define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
 706 #define EXT4_IOC_MIGRATE                _IO('f', 9)
 707  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
 708  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
 709 #define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
 710 #define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
 711 #define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
 712 #define EXT4_IOC_SWAP_BOOT              _IO('f', 17)
 713 #define EXT4_IOC_PRECACHE_EXTENTS       _IO('f', 18)
 714 /* ioctl codes 19--39 are reserved for fscrypt */
 715 #define EXT4_IOC_CLEAR_ES_CACHE         _IO('f', 40)
 716 #define EXT4_IOC_GETSTATE               _IOW('f', 41, __u32)
 717 #define EXT4_IOC_GET_ES_CACHE           _IOWR('f', 42, struct fiemap)
 718
 719 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
 720
 721 /*
 722  * Flags for going down operation
 723  */
 724 #define EXT4_GOING_FLAGS_DEFAULT                0x0     /* going down */
 725 #define EXT4_GOING_FLAGS_LOGFLUSH               0x1     /* flush log but not data */
 726 #define EXT4_GOING_FLAGS_NOLOGFLUSH             0x2     /* don't flush log nor data */
 727
 728 /*
 729  * Flags returned by EXT4_IOC_GETSTATE
 730  *
 731  * We only expose to userspace a subset of the state flags in
 732  * i_state_flags
 733  */
 734 #define EXT4_STATE_FLAG_EXT_PRECACHED   0x00000001
 735 #define EXT4_STATE_FLAG_NEW             0x00000002
 736 #define EXT4_STATE_FLAG_NEWENTRY        0x00000004
 737 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE  0x00000008
 738
 739 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 740 /*
 741  * ioctl commands in 32 bit emulation
 742  */
 743 #define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
 744 #define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
 745 #define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
 746 #define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
 747 #define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
 748 #define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
 749 #define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
 750 #define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
 751 #endif
 752
 753 /*
 754  * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
 755  * It indicates that the entry in extent status cache is for a hole.
 756  */
 757 #define EXT4_FIEMAP_EXTENT_HOLE         0x08000000
 758
 759 /* Max physical block we can address w/o extents */
 760 #define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
 761
 762 /* Max logical block we can support */
 763 #define EXT4_MAX_LOGICAL_BLOCK          0xFFFFFFFE
 764
 765 /*
 766  * Structure of an inode on the disk
 767  */
 768 struct ext4_inode {
 769         __le16  i_mode;         /* File mode */
 770         __le16  i_uid;          /* Low 16 bits of Owner Uid */
 771         __le32  i_size_lo;      /* Size in bytes */
 772         __le32  i_atime;        /* Access time */
 773         __le32  i_ctime;        /* Inode Change time */
 774         __le32  i_mtime;        /* Modification time */
 775         __le32  i_dtime;        /* Deletion Time */
 776         __le16  i_gid;          /* Low 16 bits of Group Id */
 777         __le16  i_links_count;  /* Links count */
 778         __le32  i_blocks_lo;    /* Blocks count */
 779         __le32  i_flags;        /* File flags */
 780         union {
 781                 struct {
 782                         __le32  l_i_version;
 783                 } linux1;
 784                 struct {
 785                         __u32  h_i_translator;
 786                 } hurd1;
 787                 struct {
 788                         __u32  m_i_reserved1;
 789                 } masix1;
 790         } osd1;                         /* OS dependent 1 */
 791         __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
 792         __le32  i_generation;   /* File version (for NFS) */
 793         __le32  i_file_acl_lo;  /* File ACL */
 794         __le32  i_size_high;
 795         __le32  i_obso_faddr;   /* Obsoleted fragment address */
 796         union {
 797                 struct {
 798                         __le16  l_i_blocks_high; /* were l_i_reserved1 */
 799                         __le16  l_i_file_acl_high;
 800                         __le16  l_i_uid_high;   /* these 2 fields */
 801                         __le16  l_i_gid_high;   /* were reserved2[0] */
 802                         __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
 803                         __le16  l_i_reserved;
 804                 } linux2;
 805                 struct {
 806                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 807                         __u16   h_i_mode_high;
 808                         __u16   h_i_uid_high;
 809                         __u16   h_i_gid_high;
 810                         __u32   h_i_author;
 811                 } hurd2;
 812                 struct {
 813                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
 814                         __le16  m_i_file_acl_high;
 815                         __u32   m_i_reserved2[2];
 816                 } masix2;
 817         } osd2;                         /* OS dependent 2 */
 818         __le16  i_extra_isize;
 819         __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
 820         __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
 821         __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
 822         __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
 823         __le32  i_crtime;       /* File Creation time */
 824         __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
 825         __le32  i_version_hi;   /* high 32 bits for 64-bit version */
 826         __le32  i_projid;       /* Project ID */
 827 };
 828
 829 struct move_extent {
 830         __u32 reserved;         /* should be zero */
 831         __u32 donor_fd;         /* donor file descriptor */
 832         __u64 orig_start;       /* logical start offset in block for orig */
 833         __u64 donor_start;      /* logical start offset in block for donor */
 834         __u64 len;              /* block length to be moved */
 835         __u64 moved_len;        /* moved block length */
 836 };
 837
 838 #define EXT4_EPOCH_BITS 2
 839 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
 840 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
 841
 842 /*
 843  * Extended fields will fit into an inode if the filesystem was formatted
 844  * with large inodes (-I 256 or larger) and there are not currently any EAs
 845  * consuming all of the available space. For new inodes we always reserve
 846  * enough space for the kernel's known extended fields, but for inodes
 847  * created with an old kernel this might not have been the case. None of
 848  * the extended inode fields is critical for correct filesystem operation.
 849  * This macro checks if a certain field fits in the inode. Note that
 850  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
 851  */
 852 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
 853         ((offsetof(typeof(*ext4_inode), field) +        \
 854           sizeof((ext4_inode)->field))                  \
 855         <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
 856             (einode)->i_extra_isize))                   \
 857
 858 /*
 859  * We use an encoding that preserves the times for extra epoch "00":
 860  *
 861  * extra  msb of                         adjust for signed
 862  * epoch  32-bit                         32-bit tv_sec to
 863  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
 864  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
 865  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
 866  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
 867  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
 868  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
 869  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
 870  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
 871  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
 872  *
 873  * Note that previous versions of the kernel on 64-bit systems would
 874  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
 875  * 1970.  e2fsck will correct this, assuming that it is run on the
 876  * affected filesystem before 2242.
 877  */
 878
 879 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
 880 {
 881         u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
 882         return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
 883 }
 884
 885 static inline void ext4_decode_extra_time(struct timespec64 *time,
 886                                           __le32 extra)
 887 {
 888         if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
 889                 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
 890         time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
 891 }
 892
 893 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                           \
 894 do {                                                                            \
 895         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
 896                 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);        \
 897                 (raw_inode)->xtime ## _extra =                                  \
 898                                 ext4_encode_extra_time(&(inode)->xtime);        \
 899                 }                                                               \
 900         else    \
 901                 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));    \
 902 } while (0)
 903
 904 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
 905 do {                                                                           \
 906         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 907                 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
 908         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 909                 (raw_inode)->xtime ## _extra =                                 \
 910                                 ext4_encode_extra_time(&(einode)->xtime);      \
 911 } while (0)
 912
 913 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                           \
 914 do {                                                                            \
 915         (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);        \
 916         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {    \
 917                 ext4_decode_extra_time(&(inode)->xtime,                         \
 918                                        raw_inode->xtime ## _extra);             \
 919                 }                                                               \
 920         else                                                                    \
 921                 (inode)->xtime.tv_nsec = 0;                                     \
 922 } while (0)
 923
 924
 925 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
 926 do {                                                                           \
 927         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
 928                 (einode)->xtime.tv_sec =                                       \
 929                         (signed)le32_to_cpu((raw_inode)->xtime);               \
 930         else                                                                   \
 931                 (einode)->xtime.tv_sec = 0;                                    \
 932         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
 933                 ext4_decode_extra_time(&(einode)->xtime,                       \
 934                                        raw_inode->xtime ## _extra);            \
 935         else                                                                   \
 936                 (einode)->xtime.tv_nsec = 0;                                   \
 937 } while (0)
 938
 939 #define i_disk_version osd1.linux1.l_i_version
 940
 941 #if defined(__KERNEL__) || defined(__linux__)
 942 #define i_reserved1     osd1.linux1.l_i_reserved1
 943 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
 944 #define i_blocks_high   osd2.linux2.l_i_blocks_high
 945 #define i_uid_low       i_uid
 946 #define i_gid_low       i_gid
 947 #define i_uid_high      osd2.linux2.l_i_uid_high
 948 #define i_gid_high      osd2.linux2.l_i_gid_high
 949 #define i_checksum_lo   osd2.linux2.l_i_checksum_lo
 950
 951 #elif defined(__GNU__)
 952
 953 #define i_translator    osd1.hurd1.h_i_translator
 954 #define i_uid_high      osd2.hurd2.h_i_uid_high
 955 #define i_gid_high      osd2.hurd2.h_i_gid_high
 956 #define i_author        osd2.hurd2.h_i_author
 957
 958 #elif defined(__masix__)
 959
 960 #define i_reserved1     osd1.masix1.m_i_reserved1
 961 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
 962 #define i_reserved2     osd2.masix2.m_i_reserved2
 963
 964 #endif /* defined(__KERNEL__) || defined(__linux__) */
 965
 966 #include "extents_status.h"
 967
 968 /*
 969  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
 970  *
 971  * These are needed to avoid lockdep false positives when we need to
 972  * allocate blocks to the quota inode during ext4_map_blocks(), while
 973  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
 974  * do quota tracking for the quota inode, this avoids deadlock (as
 975  * well as infinite recursion, since it isn't turtles all the way
 976  * down...)
 977  *
 978  *  I_DATA_SEM_NORMAL - Used for most inodes
 979  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
 980  *                        where the second inode has larger inode number
 981  *                        than the first
 982  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
 983  */
 984 enum {
 985         I_DATA_SEM_NORMAL = 0,
 986         I_DATA_SEM_OTHER,
 987         I_DATA_SEM_QUOTA,
 988 };
 989
 990
 991 /*
 992  * fourth extended file system inode data in memory
 993  */
 994 struct ext4_inode_info {
 995         __le32  i_data[15];     /* unconverted */
 996         __u32   i_dtime;
 997         ext4_fsblk_t    i_file_acl;
 998
 999         /*
1000          * i_block_group is the number of the block group which contains
1001          * this file's inode.  Constant across the lifetime of the inode,
1002          * it is used for making block allocation decisions - we try to
1003          * place a file's data blocks near its inode block, and new inodes
1004          * near to their parent directory's inode.
1005          */
1006         ext4_group_t    i_block_group;
1007         ext4_lblk_t     i_dir_start_lookup;
1008 #if (BITS_PER_LONG < 64)
1009         unsigned long   i_state_flags;          /* Dynamic state flags */
1010 #endif
1011         unsigned long   i_flags;
1012
1013         /*
1014          * Extended attributes can be read independently of the main file
1015          * data. Taking i_mutex even when reading would cause contention
1016          * between readers of EAs and writers of regular file data, so
1017          * instead we synchronize on xattr_sem when reading or changing
1018          * EAs.
1019          */
1020         struct rw_semaphore xattr_sem;
1021
1022         struct list_head i_orphan;      /* unlinked but open inodes */
1023
1024         /*
1025          * i_disksize keeps track of what the inode size is ON DISK, not
1026          * in memory.  During truncate, i_size is set to the new size by
1027          * the VFS prior to calling ext4_truncate(), but the filesystem won't
1028          * set i_disksize to 0 until the truncate is actually under way.
1029          *
1030          * The intent is that i_disksize always represents the blocks which
1031          * are used by this file.  This allows recovery to restart truncate
1032          * on orphans if we crash during truncate.  We actually write i_disksize
1033          * into the on-disk inode when writing inodes out, instead of i_size.
1034          *
1035          * The only time when i_disksize and i_size may be different is when
1036          * a truncate is in progress.  The only things which change i_disksize
1037          * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1038          */
1039         loff_t  i_disksize;
1040
1041         /*
1042          * i_data_sem is for serialising ext4_truncate() against
1043          * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1044          * data tree are chopped off during truncate. We can't do that in
1045          * ext4 because whenever we perform intermediate commits during
1046          * truncate, the inode and all the metadata blocks *must* be in a
1047          * consistent state which allows truncation of the orphans to restart
1048          * during recovery.  Hence we must fix the get_block-vs-truncate race
1049          * by other means, so we have i_data_sem.
1050          */
1051         struct rw_semaphore i_data_sem;
1052         /*
1053          * i_mmap_sem is for serializing page faults with truncate / punch hole
1054          * operations. We have to make sure that new page cannot be faulted in
1055          * a section of the inode that is being punched. We cannot easily use
1056          * i_data_sem for this since we need protection for the whole punch
1057          * operation and i_data_sem ranks below transaction start so we have
1058          * to occasionally drop it.
1059          */
1060         struct rw_semaphore i_mmap_sem;
1061         struct inode vfs_inode;
1062         struct jbd2_inode *jinode;
1063
1064         spinlock_t i_raw_lock;  /* protects updates to the raw inode */
1065
1066         /*
1067          * File creation time. Its function is same as that of
1068          * struct timespec64 i_{a,c,m}time in the generic inode.
1069          */
1070         struct timespec64 i_crtime;
1071
1072         /* mballoc */
1073         struct list_head i_prealloc_list;
1074         spinlock_t i_prealloc_lock;
1075
1076         /* extents status tree */
1077         struct ext4_es_tree i_es_tree;
1078         rwlock_t i_es_lock;
1079         struct list_head i_es_list;
1080         unsigned int i_es_all_nr;       /* protected by i_es_lock */
1081         unsigned int i_es_shk_nr;       /* protected by i_es_lock */
1082         ext4_lblk_t i_es_shrink_lblk;   /* Offset where we start searching for
1083                                            extents to shrink. Protected by
1084                                            i_es_lock  */
1085
1086         /* ialloc */
1087         ext4_group_t    i_last_alloc_group;
1088
1089         /* allocation reservation info for delalloc */
1090         /* In case of bigalloc, this refer to clusters rather than blocks */
1091         unsigned int i_reserved_data_blocks;
1092
1093         /* pending cluster reservations for bigalloc file systems */
1094         struct ext4_pending_tree i_pending_tree;
1095
1096         /* on-disk additional length */
1097         __u16 i_extra_isize;
1098
1099         /* Indicate the inline data space. */
1100         u16 i_inline_off;
1101         u16 i_inline_size;
1102
1103 #ifdef CONFIG_QUOTA
1104         /* quota space reservation, managed internally by quota code */
1105         qsize_t i_reserved_quota;
1106 #endif
1107
1108         /* Lock protecting lists below */
1109         spinlock_t i_completed_io_lock;
1110         /*
1111          * Completed IOs that need unwritten extents handling and have
1112          * transaction reserved
1113          */
1114         struct list_head i_rsv_conversion_list;
1115         struct work_struct i_rsv_conversion_work;
1116         atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1117
1118         spinlock_t i_block_reservation_lock;
1119
1120         /*
1121          * Transactions that contain inode's metadata needed to complete
1122          * fsync and fdatasync, respectively.
1123          */
1124         tid_t i_sync_tid;
1125         tid_t i_datasync_tid;
1126
1127 #ifdef CONFIG_QUOTA
1128         struct dquot *i_dquot[MAXQUOTAS];
1129 #endif
1130
1131         /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1132         __u32 i_csum_seed;
1133
1134         kprojid_t i_projid;
1135 };
1136
1137 /*
1138  * File system states
1139  */
1140 #define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
1141 #define EXT4_ERROR_FS                   0x0002  /* Errors detected */
1142 #define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
1143
1144 /*
1145  * Misc. filesystem flags
1146  */
1147 #define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
1148 #define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
1149 #define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
1150
1151 /*
1152  * Mount flags set via mount options or defaults
1153  */
1154 #define EXT4_MOUNT_NO_MBCACHE           0x00001 /* Do not use mbcache */
1155 #define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
1156 #define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
1157 #define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
1158 #define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
1159 #define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
1160 #define EXT4_MOUNT_ERRORS_MASK          0x00070
1161 #define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
1162 #define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
1163 #ifdef CONFIG_FS_DAX
1164 #define EXT4_MOUNT_DAX_ALWAYS           0x00200 /* Direct Access */
1165 #else
1166 #define EXT4_MOUNT_DAX_ALWAYS           0
1167 #endif
1168 #define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
1169 #define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
1170 #define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
1171 #define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
1172 #define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
1173 #define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
1174 #define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
1175 #define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
1176 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
1177 #define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
1178 #define EXT4_MOUNT_QUOTA                0x40000 /* Some quota option set */
1179 #define EXT4_MOUNT_USRQUOTA             0x80000 /* "old" user quota,
1180                                                  * enable enforcement for hidden
1181                                                  * quota files */
1182 #define EXT4_MOUNT_GRPQUOTA             0x100000 /* "old" group quota, enable
1183                                                   * enforcement for hidden quota
1184                                                   * files */
1185 #define EXT4_MOUNT_PRJQUOTA             0x200000 /* Enable project quota
1186                                                   * enforcement */
1187 #define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
1188 #define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
1189 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1190 #define EXT4_MOUNT_WARN_ON_ERROR        0x2000000 /* Trigger WARN_ON on error */
1191 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1192 #define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
1193 #define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
1194 #define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
1195 #define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
1196 #define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
1197
1198 /*
1199  * Mount flags set either automatically (could not be set by mount option)
1200  * based on per file system feature or property or in special cases such as
1201  * distinguishing between explicit mount option definition and default.
1202  */
1203 #define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
1204                                                       specified delalloc */
1205 #define EXT4_MOUNT2_STD_GROUP_SIZE      0x00000002 /* We have standard group
1206                                                       size of blocksize * 8
1207                                                       blocks */
1208 #define EXT4_MOUNT2_HURD_COMPAT         0x00000004 /* Support HURD-castrated
1209                                                       file systems */
1210 #define EXT4_MOUNT2_DAX_NEVER           0x00000008 /* Do not allow Direct Access */
1211 #define EXT4_MOUNT2_DAX_INODE           0x00000010 /* For printing options only */
1212
1213 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM   0x00000008 /* User explicitly
1214                                                 specified journal checksum */
1215
1216 #define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
1217                                                 ~EXT4_MOUNT_##opt
1218 #define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1219                                                 EXT4_MOUNT_##opt
1220 #define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1221                                          EXT4_MOUNT_##opt)
1222
1223 #define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1224                                                 ~EXT4_MOUNT2_##opt
1225 #define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1226                                                 EXT4_MOUNT2_##opt
1227 #define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1228                                          EXT4_MOUNT2_##opt)
1229
1230 #define ext4_test_and_set_bit           __test_and_set_bit_le
1231 #define ext4_set_bit                    __set_bit_le
1232 #define ext4_set_bit_atomic             ext2_set_bit_atomic
1233 #define ext4_test_and_clear_bit         __test_and_clear_bit_le
1234 #define ext4_clear_bit                  __clear_bit_le
1235 #define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1236 #define ext4_test_bit                   test_bit_le
1237 #define ext4_find_next_zero_bit         find_next_zero_bit_le
1238 #define ext4_find_next_bit              find_next_bit_le
1239
1240 extern void ext4_set_bits(void *bm, int cur, int len);
1241
1242 /*
1243  * Maximal mount counts between two filesystem checks
1244  */
1245 #define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1246 #define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1247
1248 /*
1249  * Behaviour when detecting errors
1250  */
1251 #define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1252 #define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1253 #define EXT4_ERRORS_PANIC               3       /* Panic */
1254 #define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1255
1256 /* Metadata checksum algorithm codes */
1257 #define EXT4_CRC32C_CHKSUM              1
1258
1259 /*
1260  * Structure of the super block
1261  */
1262 struct ext4_super_block {
1263 /*00*/  __le32  s_inodes_count;         /* Inodes count */
1264         __le32  s_blocks_count_lo;      /* Blocks count */
1265         __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1266         __le32  s_free_blocks_count_lo; /* Free blocks count */
1267 /*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1268         __le32  s_first_data_block;     /* First Data Block */
1269         __le32  s_log_block_size;       /* Block size */
1270         __le32  s_log_cluster_size;     /* Allocation cluster size */
1271 /*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1272         __le32  s_clusters_per_group;   /* # Clusters per group */
1273         __le32  s_inodes_per_group;     /* # Inodes per group */
1274         __le32  s_mtime;                /* Mount time */
1275 /*30*/  __le32  s_wtime;                /* Write time */
1276         __le16  s_mnt_count;            /* Mount count */
1277         __le16  s_max_mnt_count;        /* Maximal mount count */
1278         __le16  s_magic;                /* Magic signature */
1279         __le16  s_state;                /* File system state */
1280         __le16  s_errors;               /* Behaviour when detecting errors */
1281         __le16  s_minor_rev_level;      /* minor revision level */
1282 /*40*/  __le32  s_lastcheck;            /* time of last check */
1283         __le32  s_checkinterval;        /* max. time between checks */
1284         __le32  s_creator_os;           /* OS */
1285         __le32  s_rev_level;            /* Revision level */
1286 /*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1287         __le16  s_def_resgid;           /* Default gid for reserved blocks */
1288         /*
1289          * These fields are for EXT4_DYNAMIC_REV superblocks only.
1290          *
1291          * Note: the difference between the compatible feature set and
1292          * the incompatible feature set is that if there is a bit set
1293          * in the incompatible feature set that the kernel doesn't
1294          * know about, it should refuse to mount the filesystem.
1295          *
1296          * e2fsck's requirements are more strict; if it doesn't know
1297          * about a feature in either the compatible or incompatible
1298          * feature set, it must abort and not try to meddle with
1299          * things it doesn't understand...
1300          */
1301         __le32  s_first_ino;            /* First non-reserved inode */
1302         __le16  s_inode_size;           /* size of inode structure */
1303         __le16  s_block_group_nr;       /* block group # of this superblock */
1304         __le32  s_feature_compat;       /* compatible feature set */
1305 /*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1306         __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1307 /*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1308 /*78*/  char    s_volume_name[16];      /* volume name */
1309 /*88*/  char    s_last_mounted[64] __nonstring; /* directory where last mounted */
1310 /*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1311         /*
1312          * Performance hints.  Directory preallocation should only
1313          * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1314          */
1315         __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1316         __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1317         __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1318         /*
1319          * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1320          */
1321 /*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1322 /*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1323         __le32  s_journal_dev;          /* device number of journal file */
1324         __le32  s_last_orphan;          /* start of list of inodes to delete */
1325         __le32  s_hash_seed[4];         /* HTREE hash seed */
1326         __u8    s_def_hash_version;     /* Default hash version to use */
1327         __u8    s_jnl_backup_type;
1328         __le16  s_desc_size;            /* size of group descriptor */
1329 /*100*/ __le32  s_default_mount_opts;
1330         __le32  s_first_meta_bg;        /* First metablock block group */
1331         __le32  s_mkfs_time;            /* When the filesystem was created */
1332         __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1333         /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1334 /*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1335         __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1336         __le32  s_free_blocks_count_hi; /* Free blocks count */
1337         __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1338         __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1339         __le32  s_flags;                /* Miscellaneous flags */
1340         __le16  s_raid_stride;          /* RAID stride */
1341         __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1342         __le64  s_mmp_block;            /* Block for multi-mount protection */
1343         __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1344         __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1345         __u8    s_checksum_type;        /* metadata checksum algorithm used */
1346         __u8    s_encryption_level;     /* versioning level for encryption */
1347         __u8    s_reserved_pad;         /* Padding to next 32bits */
1348         __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1349         __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1350         __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1351         __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1352                                               snapshot's future use */
1353         __le32  s_snapshot_list;        /* inode number of the head of the
1354                                            on-disk snapshot list */
1355 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1356         __le32  s_error_count;          /* number of fs errors */
1357         __le32  s_first_error_time;     /* first time an error happened */
1358         __le32  s_first_error_ino;      /* inode involved in first error */
1359         __le64  s_first_error_block;    /* block involved of first error */
1360         __u8    s_first_error_func[32] __nonstring;     /* function where the error happened */
1361         __le32  s_first_error_line;     /* line number where error happened */
1362         __le32  s_last_error_time;      /* most recent time of an error */
1363         __le32  s_last_error_ino;       /* inode involved in last error */
1364         __le32  s_last_error_line;      /* line number where error happened */
1365         __le64  s_last_error_block;     /* block involved of last error */
1366         __u8    s_last_error_func[32] __nonstring;      /* function where the error happened */
1367 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1368         __u8    s_mount_opts[64];
1369         __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1370         __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1371         __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1372         __le32  s_backup_bgs[2];        /* groups with sparse_super2 SBs */
1373         __u8    s_encrypt_algos[4];     /* Encryption algorithms in use  */
1374         __u8    s_encrypt_pw_salt[16];  /* Salt used for string2key algorithm */
1375         __le32  s_lpf_ino;              /* Location of the lost+found inode */
1376         __le32  s_prj_quota_inum;       /* inode for tracking project quota */
1377         __le32  s_checksum_seed;        /* crc32c(uuid) if csum_seed set */
1378         __u8    s_wtime_hi;
1379         __u8    s_mtime_hi;
1380         __u8    s_mkfs_time_hi;
1381         __u8    s_lastcheck_hi;
1382         __u8    s_first_error_time_hi;
1383         __u8    s_last_error_time_hi;
1384         __u8    s_first_error_errcode;
1385         __u8    s_last_error_errcode;
1386         __le16  s_encoding;             /* Filename charset encoding */
1387         __le16  s_encoding_flags;       /* Filename charset encoding flags */
1388         __le32  s_reserved[95];         /* Padding to the end of the block */
1389         __le32  s_checksum;             /* crc32c(superblock) */
1390 };
1391
1392 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1393
1394 #ifdef __KERNEL__
1395
1396 /*
1397  * run-time mount flags
1398  */
1399 #define EXT4_MF_MNTDIR_SAMPLED          0x0001
1400 #define EXT4_MF_FS_ABORTED              0x0002  /* Fatal error detected */
1401
1402 #ifdef CONFIG_FS_ENCRYPTION
1403 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL)
1404 #else
1405 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1406 #endif
1407
1408 /* Number of quota types we support */
1409 #define EXT4_MAXQUOTAS 3
1410
1411 #define EXT4_ENC_UTF8_12_1      1
1412
1413 /*
1414  * Flags for ext4_sb_info.s_encoding_flags.
1415  */
1416 #define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1417
1418 #define ext4_has_strict_mode(sbi) \
1419         (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1420
1421 /*
1422  * fourth extended-fs super-block data in memory
1423  */
1424 struct ext4_sb_info {
1425         unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1426         unsigned long s_inodes_per_block;/* Number of inodes per block */
1427         unsigned long s_blocks_per_group;/* Number of blocks in a group */
1428         unsigned long s_clusters_per_group; /* Number of clusters in a group */
1429         unsigned long s_inodes_per_group;/* Number of inodes in a group */
1430         unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1431         unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1432         unsigned long s_desc_per_block; /* Number of group descriptors per block */
1433         ext4_group_t s_groups_count;    /* Number of groups in the fs */
1434         ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1435         unsigned long s_overhead;  /* # of fs overhead clusters */
1436         unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1437         unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1438         loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1439         struct buffer_head * s_sbh;     /* Buffer containing the super block */
1440         struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1441         struct buffer_head * __rcu *s_group_desc;
1442         unsigned int s_mount_opt;
1443         unsigned int s_mount_opt2;
1444         unsigned int s_mount_flags;
1445         unsigned int s_def_mount_opt;
1446         ext4_fsblk_t s_sb_block;
1447         atomic64_t s_resv_clusters;
1448         kuid_t s_resuid;
1449         kgid_t s_resgid;
1450         unsigned short s_mount_state;
1451         unsigned short s_pad;
1452         int s_addr_per_block_bits;
1453         int s_desc_per_block_bits;
1454         int s_inode_size;
1455         int s_first_ino;
1456         unsigned int s_inode_readahead_blks;
1457         unsigned int s_inode_goal;
1458         u32 s_hash_seed[4];
1459         int s_def_hash_version;
1460         int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1461         struct percpu_counter s_freeclusters_counter;
1462         struct percpu_counter s_freeinodes_counter;
1463         struct percpu_counter s_dirs_counter;
1464         struct percpu_counter s_dirtyclusters_counter;
1465         struct blockgroup_lock *s_blockgroup_lock;
1466         struct proc_dir_entry *s_proc;
1467         struct kobject s_kobj;
1468         struct completion s_kobj_unregister;
1469         struct super_block *s_sb;
1470 #ifdef CONFIG_UNICODE
1471         struct unicode_map *s_encoding;
1472         __u16 s_encoding_flags;
1473 #endif
1474
1475         /* Journaling */
1476         struct journal_s *s_journal;
1477         struct list_head s_orphan;
1478         struct mutex s_orphan_lock;
1479         unsigned long s_ext4_flags;             /* Ext4 superblock flags */
1480         unsigned long s_commit_interval;
1481         u32 s_max_batch_time;
1482         u32 s_min_batch_time;
1483         struct block_device *journal_bdev;
1484 #ifdef CONFIG_QUOTA
1485         /* Names of quota files with journalled quota */
1486         char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1487         int s_jquota_fmt;                       /* Format of quota to use */
1488 #endif
1489         unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1490         struct ext4_system_blocks __rcu *system_blks;
1491
1492 #ifdef EXTENTS_STATS
1493         /* ext4 extents stats */
1494         unsigned long s_ext_min;
1495         unsigned long s_ext_max;
1496         unsigned long s_depth_max;
1497         spinlock_t s_ext_stats_lock;
1498         unsigned long s_ext_blocks;
1499         unsigned long s_ext_extents;
1500 #endif
1501
1502         /* for buddy allocator */
1503         struct ext4_group_info ** __rcu *s_group_info;
1504         struct inode *s_buddy_cache;
1505         spinlock_t s_md_lock;
1506         unsigned short *s_mb_offsets;
1507         unsigned int *s_mb_maxs;
1508         unsigned int s_group_info_size;
1509         unsigned int s_mb_free_pending;
1510         struct list_head s_freed_data_list;     /* List of blocks to be freed
1511                                                    after commit completed */
1512
1513         /* tunables */
1514         unsigned long s_stripe;
1515         unsigned int s_mb_stream_request;
1516         unsigned int s_mb_max_to_scan;
1517         unsigned int s_mb_min_to_scan;
1518         unsigned int s_mb_stats;
1519         unsigned int s_mb_order2_reqs;
1520         unsigned int s_mb_group_prealloc;
1521         unsigned int s_max_dir_size_kb;
1522         /* where last allocation was done - for stream allocation */
1523         unsigned long s_mb_last_group;
1524         unsigned long s_mb_last_start;
1525         unsigned int s_mb_prefetch;
1526         unsigned int s_mb_prefetch_limit;
1527
1528         /* stats for buddy allocator */
1529         atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1530         atomic_t s_bal_success; /* we found long enough chunks */
1531         atomic_t s_bal_allocated;       /* in blocks */
1532         atomic_t s_bal_ex_scanned;      /* total extents scanned */
1533         atomic_t s_bal_goals;   /* goal hits */
1534         atomic_t s_bal_breaks;  /* too long searches */
1535         atomic_t s_bal_2orders; /* 2^order hits */
1536         spinlock_t s_bal_lock;
1537         unsigned long s_mb_buddies_generated;
1538         unsigned long long s_mb_generation_time;
1539         atomic_t s_mb_lost_chunks;
1540         atomic_t s_mb_preallocated;
1541         atomic_t s_mb_discarded;
1542         atomic_t s_lock_busy;
1543
1544         /* locality groups */
1545         struct ext4_locality_group __percpu *s_locality_groups;
1546
1547         /* for write statistics */
1548         unsigned long s_sectors_written_start;
1549         u64 s_kbytes_written;
1550
1551         /* the size of zero-out chunk */
1552         unsigned int s_extent_max_zeroout_kb;
1553
1554         unsigned int s_log_groups_per_flex;
1555         struct flex_groups * __rcu *s_flex_groups;
1556         ext4_group_t s_flex_groups_allocated;
1557
1558         /* workqueue for reserved extent conversions (buffered io) */
1559         struct workqueue_struct *rsv_conversion_wq;
1560
1561         /* timer for periodic error stats printing */
1562         struct timer_list s_err_report;
1563
1564         /* Lazy inode table initialization info */
1565         struct ext4_li_request *s_li_request;
1566         /* Wait multiplier for lazy initialization thread */
1567         unsigned int s_li_wait_mult;
1568
1569         /* Kernel thread for multiple mount protection */
1570         struct task_struct *s_mmp_tsk;
1571
1572         /* record the last minlen when FITRIM is called. */
1573         atomic_t s_last_trim_minblks;
1574
1575         /* Reference to checksum algorithm driver via cryptoapi */
1576         struct crypto_shash *s_chksum_driver;
1577
1578         /* Precomputed FS UUID checksum for seeding other checksums */
1579         __u32 s_csum_seed;
1580
1581         /* Reclaim extents from extent status tree */
1582         struct shrinker s_es_shrinker;
1583         struct list_head s_es_list;     /* List of inodes with reclaimable extents */
1584         long s_es_nr_inode;
1585         struct ext4_es_stats s_es_stats;
1586         struct mb_cache *s_ea_block_cache;
1587         struct mb_cache *s_ea_inode_cache;
1588         spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1589
1590         /* Ratelimit ext4 messages. */
1591         struct ratelimit_state s_err_ratelimit_state;
1592         struct ratelimit_state s_warning_ratelimit_state;
1593         struct ratelimit_state s_msg_ratelimit_state;
1594         atomic_t s_warning_count;
1595         atomic_t s_msg_count;
1596
1597         /* Encryption context for '-o test_dummy_encryption' */
1598         struct fscrypt_dummy_context s_dummy_enc_ctx;
1599
1600         /*
1601          * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1602          * or EXTENTS flag.
1603          */
1604         struct percpu_rw_semaphore s_writepages_rwsem;
1605         struct dax_device *s_daxdev;
1606 #ifdef CONFIG_EXT4_DEBUG
1607         unsigned long s_simulate_fail;
1608 #endif
1609         /* Record the errseq of the backing block device */
1610         errseq_t s_bdev_wb_err;
1611         spinlock_t s_bdev_wb_lock;
1612 };
1613
1614 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1615 {
1616         return sb->s_fs_info;
1617 }
1618 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1619 {
1620         return container_of(inode, struct ext4_inode_info, vfs_inode);
1621 }
1622
1623 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1624 {
1625         return ino == EXT4_ROOT_INO ||
1626                 (ino >= EXT4_FIRST_INO(sb) &&
1627                  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1628 }
1629
1630 /*
1631  * Returns: sbi->field[index]
1632  * Used to access an array element from the following sbi fields which require
1633  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1634  * - s_group_desc
1635  * - s_group_info
1636  * - s_flex_group
1637  */
1638 #define sbi_array_rcu_deref(sbi, field, index)                             \
1639 ({                                                                         \
1640         typeof(*((sbi)->field)) _v;                                        \
1641         rcu_read_lock();                                                   \
1642         _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];          \
1643         rcu_read_unlock();                                                 \
1644         _v;                                                                \
1645 })
1646
1647 /*
1648  * Simulate_fail codes
1649  */
1650 #define EXT4_SIM_BBITMAP_EIO    1
1651 #define EXT4_SIM_BBITMAP_CRC    2
1652 #define EXT4_SIM_IBITMAP_EIO    3
1653 #define EXT4_SIM_IBITMAP_CRC    4
1654 #define EXT4_SIM_INODE_EIO      5
1655 #define EXT4_SIM_INODE_CRC      6
1656 #define EXT4_SIM_DIRBLOCK_EIO   7
1657 #define EXT4_SIM_DIRBLOCK_CRC   8
1658
1659 static inline bool ext4_simulate_fail(struct super_block *sb,
1660                                      unsigned long code)
1661 {
1662 #ifdef CONFIG_EXT4_DEBUG
1663         struct ext4_sb_info *sbi = EXT4_SB(sb);
1664
1665         if (unlikely(sbi->s_simulate_fail == code)) {
1666                 sbi->s_simulate_fail = 0;
1667                 return true;
1668         }
1669 #endif
1670         return false;
1671 }
1672
1673 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1674                                          struct buffer_head *bh,
1675                                          unsigned long code)
1676 {
1677         if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1678                 clear_buffer_uptodate(bh);
1679 }
1680
1681 /*
1682  * Error number codes for s_{first,last}_error_errno
1683  *
1684  * Linux errno numbers are architecture specific, so we need to translate
1685  * them into something which is architecture independent.   We don't define
1686  * codes for all errno's; just the ones which are most likely to be the cause
1687  * of an ext4_error() call.
1688  */
1689 #define EXT4_ERR_UNKNOWN         1
1690 #define EXT4_ERR_EIO             2
1691 #define EXT4_ERR_ENOMEM          3
1692 #define EXT4_ERR_EFSBADCRC       4
1693 #define EXT4_ERR_EFSCORRUPTED    5
1694 #define EXT4_ERR_ENOSPC          6
1695 #define EXT4_ERR_ENOKEY          7
1696 #define EXT4_ERR_EROFS           8
1697 #define EXT4_ERR_EFBIG           9
1698 #define EXT4_ERR_EEXIST         10
1699 #define EXT4_ERR_ERANGE         11
1700 #define EXT4_ERR_EOVERFLOW      12
1701 #define EXT4_ERR_EBUSY          13
1702 #define EXT4_ERR_ENOTDIR        14
1703 #define EXT4_ERR_ENOTEMPTY      15
1704 #define EXT4_ERR_ESHUTDOWN      16
1705 #define EXT4_ERR_EFAULT         17
1706
1707 /*
1708  * Inode dynamic state flags
1709  */
1710 enum {
1711         EXT4_STATE_JDATA,               /* journaled data exists */
1712         EXT4_STATE_NEW,                 /* inode is newly created */
1713         EXT4_STATE_XATTR,               /* has in-inode xattrs */
1714         EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1715         EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1716         EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1717         EXT4_STATE_NEWENTRY,            /* File just added to dir */
1718         EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1719         EXT4_STATE_EXT_PRECACHED,       /* extents have been precached */
1720         EXT4_STATE_LUSTRE_EA_INODE,     /* Lustre-style ea_inode */
1721         EXT4_STATE_VERITY_IN_PROGRESS,  /* building fs-verity Merkle tree */
1722 };
1723
1724 #define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1725 static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1726 {                                                                       \
1727         return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1728 }                                                                       \
1729 static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1730 {                                                                       \
1731         set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1732 }                                                                       \
1733 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1734 {                                                                       \
1735         clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1736 }
1737
1738 /* Add these declarations here only so that these functions can be
1739  * found by name.  Otherwise, they are very hard to locate. */
1740 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1741 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1742 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1743 EXT4_INODE_BIT_FNS(flag, flags, 0)
1744
1745 /* Add these declarations here only so that these functions can be
1746  * found by name.  Otherwise, they are very hard to locate. */
1747 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1748 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1749 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1750 #if (BITS_PER_LONG < 64)
1751 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1752
1753 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1754 {
1755         (ei)->i_state_flags = 0;
1756 }
1757 #else
1758 EXT4_INODE_BIT_FNS(state, flags, 32)
1759
1760 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1761 {
1762         /* We depend on the fact that callers will set i_flags */
1763 }
1764 #endif
1765 #else
1766 /* Assume that user mode programs are passing in an ext4fs superblock, not
1767  * a kernel struct super_block.  This will allow us to call the feature-test
1768  * macros from user land. */
1769 #define EXT4_SB(sb)     (sb)
1770 #endif
1771
1772 static inline bool ext4_verity_in_progress(struct inode *inode)
1773 {
1774         return IS_ENABLED(CONFIG_FS_VERITY) &&
1775                ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1776 }
1777
1778 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1779
1780 /*
1781  * Codes for operating systems
1782  */
1783 #define EXT4_OS_LINUX           0
1784 #define EXT4_OS_HURD            1
1785 #define EXT4_OS_MASIX           2
1786 #define EXT4_OS_FREEBSD         3
1787 #define EXT4_OS_LITES           4
1788
1789 /*
1790  * Revision levels
1791  */
1792 #define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1793 #define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1794
1795 #define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1796 #define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1797
1798 #define EXT4_GOOD_OLD_INODE_SIZE 128
1799
1800 #define EXT4_EXTRA_TIMESTAMP_MAX        (((s64)1 << 34) - 1  + S32_MIN)
1801 #define EXT4_NON_EXTRA_TIMESTAMP_MAX    S32_MAX
1802 #define EXT4_TIMESTAMP_MIN              S32_MIN
1803
1804 /*
1805  * Feature set definitions
1806  */
1807
1808 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1809 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1810 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1811 #define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1812 #define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1813 #define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1814 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2       0x0200
1815 #define EXT4_FEATURE_COMPAT_STABLE_INODES       0x0800
1816
1817 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1818 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1819 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1820 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1821 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1822 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1823 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1824 #define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1825 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1826 /*
1827  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1828  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1829  * all other data structures' checksums.  However, the METADATA_CSUM and
1830  * GDT_CSUM bits are mutually exclusive.
1831  */
1832 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1833 #define EXT4_FEATURE_RO_COMPAT_READONLY         0x1000
1834 #define EXT4_FEATURE_RO_COMPAT_PROJECT          0x2000
1835 #define EXT4_FEATURE_RO_COMPAT_VERITY           0x8000
1836
1837 #define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1838 #define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1839 #define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1840 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1841 #define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1842 #define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1843 #define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1844 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1845 #define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1846 #define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1847 #define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1848 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED         0x2000
1849 #define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1850 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1851 #define EXT4_FEATURE_INCOMPAT_ENCRYPT           0x10000
1852 #define EXT4_FEATURE_INCOMPAT_CASEFOLD          0x20000
1853
1854 extern void ext4_update_dynamic_rev(struct super_block *sb);
1855
1856 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1857 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1858 { \
1859         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1860                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1861 } \
1862 static inline void ext4_set_feature_##name(struct super_block *sb) \
1863 { \
1864         ext4_update_dynamic_rev(sb); \
1865         EXT4_SB(sb)->s_es->s_feature_compat |= \
1866                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1867 } \
1868 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1869 { \
1870         EXT4_SB(sb)->s_es->s_feature_compat &= \
1871                 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1872 }
1873
1874 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1875 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1876 { \
1877         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1878                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1879 } \
1880 static inline void ext4_set_feature_##name(struct super_block *sb) \
1881 { \
1882         ext4_update_dynamic_rev(sb); \
1883         EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1884                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1885 } \
1886 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1887 { \
1888         EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1889                 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1890 }
1891
1892 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1893 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1894 { \
1895         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1896                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1897 } \
1898 static inline void ext4_set_feature_##name(struct super_block *sb) \
1899 { \
1900         ext4_update_dynamic_rev(sb); \
1901         EXT4_SB(sb)->s_es->s_feature_incompat |= \
1902                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1903 } \
1904 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1905 { \
1906         EXT4_SB(sb)->s_es->s_feature_incompat &= \
1907                 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1908 }
1909
1910 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,         DIR_PREALLOC)
1911 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,        IMAGIC_INODES)
1912 EXT4_FEATURE_COMPAT_FUNCS(journal,              HAS_JOURNAL)
1913 EXT4_FEATURE_COMPAT_FUNCS(xattr,                EXT_ATTR)
1914 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,         RESIZE_INODE)
1915 EXT4_FEATURE_COMPAT_FUNCS(dir_index,            DIR_INDEX)
1916 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,        SPARSE_SUPER2)
1917 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,        STABLE_INODES)
1918
1919 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,      SPARSE_SUPER)
1920 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,        LARGE_FILE)
1921 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,         BTREE_DIR)
1922 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,         HUGE_FILE)
1923 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,          GDT_CSUM)
1924 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,         DIR_NLINK)
1925 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,       EXTRA_ISIZE)
1926 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,             QUOTA)
1927 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,          BIGALLOC)
1928 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,     METADATA_CSUM)
1929 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,          READONLY)
1930 EXT4_FEATURE_RO_COMPAT_FUNCS(project,           PROJECT)
1931 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,            VERITY)
1932
1933 EXT4_FEATURE_INCOMPAT_FUNCS(compression,        COMPRESSION)
1934 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,           FILETYPE)
1935 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,     RECOVER)
1936 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,        JOURNAL_DEV)
1937 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,            META_BG)
1938 EXT4_FEATURE_INCOMPAT_FUNCS(extents,            EXTENTS)
1939 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,              64BIT)
1940 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,                MMP)
1941 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,            FLEX_BG)
1942 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,           EA_INODE)
1943 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,            DIRDATA)
1944 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,          CSUM_SEED)
1945 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,           LARGEDIR)
1946 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,        INLINE_DATA)
1947 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,            ENCRYPT)
1948 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,           CASEFOLD)
1949
1950 #define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1951 #define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1952                                          EXT4_FEATURE_INCOMPAT_META_BG)
1953 #define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1954                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1955                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1956
1957 #define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1958 #define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1959                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1960                                          EXT4_FEATURE_INCOMPAT_META_BG)
1961 #define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1962                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1963                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1964
1965 #define EXT4_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1966 #define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1967                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1968                                          EXT4_FEATURE_INCOMPAT_META_BG| \
1969                                          EXT4_FEATURE_INCOMPAT_EXTENTS| \
1970                                          EXT4_FEATURE_INCOMPAT_64BIT| \
1971                                          EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1972                                          EXT4_FEATURE_INCOMPAT_EA_INODE| \
1973                                          EXT4_FEATURE_INCOMPAT_MMP | \
1974                                          EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1975                                          EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1976                                          EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1977                                          EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1978                                          EXT4_FEATURE_INCOMPAT_LARGEDIR)
1979 #define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1980                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1981                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1982                                          EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1983                                          EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1984                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1985                                          EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1986                                          EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1987                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1988                                          EXT4_FEATURE_RO_COMPAT_QUOTA |\
1989                                          EXT4_FEATURE_RO_COMPAT_PROJECT |\
1990                                          EXT4_FEATURE_RO_COMPAT_VERITY)
1991
1992 #define EXTN_FEATURE_FUNCS(ver) \
1993 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1994 { \
1995         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1996                 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1997 } \
1998 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1999 { \
2000         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2001                 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2002 } \
2003 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2004 { \
2005         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2006                 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2007 }
2008
2009 EXTN_FEATURE_FUNCS(2)
2010 EXTN_FEATURE_FUNCS(3)
2011 EXTN_FEATURE_FUNCS(4)
2012
2013 static inline bool ext4_has_compat_features(struct super_block *sb)
2014 {
2015         return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2016 }
2017 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2018 {
2019         return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2020 }
2021 static inline bool ext4_has_incompat_features(struct super_block *sb)
2022 {
2023         return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2024 }
2025
2026 /*
2027  * Superblock flags
2028  */
2029 #define EXT4_FLAGS_RESIZING     0
2030 #define EXT4_FLAGS_SHUTDOWN     1
2031 #define EXT4_FLAGS_BDEV_IS_DAX  2
2032
2033 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2034 {
2035         return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2036 }
2037
2038
2039 /*
2040  * Default values for user and/or group using reserved blocks
2041  */
2042 #define EXT4_DEF_RESUID         0
2043 #define EXT4_DEF_RESGID         0
2044
2045 /*
2046  * Default project ID
2047  */
2048 #define EXT4_DEF_PROJID         0
2049
2050 #define EXT4_DEF_INODE_READAHEAD_BLKS   32
2051
2052 /*
2053  * Default mount options
2054  */
2055 #define EXT4_DEFM_DEBUG         0x0001
2056 #define EXT4_DEFM_BSDGROUPS     0x0002
2057 #define EXT4_DEFM_XATTR_USER    0x0004
2058 #define EXT4_DEFM_ACL           0x0008
2059 #define EXT4_DEFM_UID16         0x0010
2060 #define EXT4_DEFM_JMODE         0x0060
2061 #define EXT4_DEFM_JMODE_DATA    0x0020
2062 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2063 #define EXT4_DEFM_JMODE_WBACK   0x0060
2064 #define EXT4_DEFM_NOBARRIER     0x0100
2065 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2066 #define EXT4_DEFM_DISCARD       0x0400
2067 #define EXT4_DEFM_NODELALLOC    0x0800
2068
2069 /*
2070  * Default journal batch times
2071  */
2072 #define EXT4_DEF_MIN_BATCH_TIME 0
2073 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2074
2075 /*
2076  * Minimum number of groups in a flexgroup before we separate out
2077  * directories into the first block group of a flexgroup
2078  */
2079 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2080
2081 /*
2082  * Structure of a directory entry
2083  */
2084 #define EXT4_NAME_LEN 255
2085
2086 struct ext4_dir_entry {
2087         __le32  inode;                  /* Inode number */
2088         __le16  rec_len;                /* Directory entry length */
2089         __le16  name_len;               /* Name length */
2090         char    name[EXT4_NAME_LEN];    /* File name */
2091 };
2092
2093 /*
2094  * The new version of the directory entry.  Since EXT4 structures are
2095  * stored in intel byte order, and the name_len field could never be
2096  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2097  * file_type field.
2098  */
2099 struct ext4_dir_entry_2 {
2100         __le32  inode;                  /* Inode number */
2101         __le16  rec_len;                /* Directory entry length */
2102         __u8    name_len;               /* Name length */
2103         __u8    file_type;              /* See file type macros EXT4_FT_* below */
2104         char    name[EXT4_NAME_LEN];    /* File name */
2105 };
2106
2107 /*
2108  * This is a bogus directory entry at the end of each leaf block that
2109  * records checksums.
2110  */
2111 struct ext4_dir_entry_tail {
2112         __le32  det_reserved_zero1;     /* Pretend to be unused */
2113         __le16  det_rec_len;            /* 12 */
2114         __u8    det_reserved_zero2;     /* Zero name length */
2115         __u8    det_reserved_ft;        /* 0xDE, fake file type */
2116         __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
2117 };
2118
2119 #define EXT4_DIRENT_TAIL(block, blocksize) \
2120         ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2121                                         ((blocksize) - \
2122                                          sizeof(struct ext4_dir_entry_tail))))
2123
2124 /*
2125  * Ext4 directory file types.  Only the low 3 bits are used.  The
2126  * other bits are reserved for now.
2127  */
2128 #define EXT4_FT_UNKNOWN         0
2129 #define EXT4_FT_REG_FILE        1
2130 #define EXT4_FT_DIR             2
2131 #define EXT4_FT_CHRDEV          3
2132 #define EXT4_FT_BLKDEV          4
2133 #define EXT4_FT_FIFO            5
2134 #define EXT4_FT_SOCK            6
2135 #define EXT4_FT_SYMLINK         7
2136
2137 #define EXT4_FT_MAX             8
2138
2139 #define EXT4_FT_DIR_CSUM        0xDE
2140
2141 /*
2142  * EXT4_DIR_PAD defines the directory entries boundaries
2143  *
2144  * NOTE: It must be a multiple of 4
2145  */
2146 #define EXT4_DIR_PAD                    4
2147 #define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
2148 #define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
2149                                          ~EXT4_DIR_ROUND)
2150 #define EXT4_MAX_REC_LEN                ((1<<16)-1)
2151
2152 /*
2153  * If we ever get support for fs block sizes > page_size, we'll need
2154  * to remove the #if statements in the next two functions...
2155  */
2156 static inline unsigned int
2157 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2158 {
2159         unsigned len = le16_to_cpu(dlen);
2160
2161 #if (PAGE_SIZE >= 65536)
2162         if (len == EXT4_MAX_REC_LEN || len == 0)
2163                 return blocksize;
2164         return (len & 65532) | ((len & 3) << 16);
2165 #else
2166         return len;
2167 #endif
2168 }
2169
2170 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2171 {
2172         if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2173                 BUG();
2174 #if (PAGE_SIZE >= 65536)
2175         if (len < 65536)
2176                 return cpu_to_le16(len);
2177         if (len == blocksize) {
2178                 if (blocksize == 65536)
2179                         return cpu_to_le16(EXT4_MAX_REC_LEN);
2180                 else
2181                         return cpu_to_le16(0);
2182         }
2183         return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2184 #else
2185         return cpu_to_le16(len);
2186 #endif
2187 }
2188
2189 /*
2190  * Hash Tree Directory indexing
2191  * (c) Daniel Phillips, 2001
2192  */
2193
2194 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2195                     ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2196 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2197                     !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2198 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2199
2200 /* Legal values for the dx_root hash_version field: */
2201
2202 #define DX_HASH_LEGACY                  0
2203 #define DX_HASH_HALF_MD4                1
2204 #define DX_HASH_TEA                     2
2205 #define DX_HASH_LEGACY_UNSIGNED         3
2206 #define DX_HASH_HALF_MD4_UNSIGNED       4
2207 #define DX_HASH_TEA_UNSIGNED            5
2208
2209 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2210                               const void *address, unsigned int length)
2211 {
2212         struct {
2213                 struct shash_desc shash;
2214                 char ctx[4];
2215         } desc;
2216
2217         BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2218
2219         desc.shash.tfm = sbi->s_chksum_driver;
2220         *(u32 *)desc.ctx = crc;
2221
2222         BUG_ON(crypto_shash_update(&desc.shash, address, length));
2223
2224         return *(u32 *)desc.ctx;
2225 }
2226
2227 #ifdef __KERNEL__
2228
2229 /* hash info structure used by the directory hash */
2230 struct dx_hash_info
2231 {
2232         u32             hash;
2233         u32             minor_hash;
2234         int             hash_version;
2235         u32             *seed;
2236 };
2237
2238
2239 /* 32 and 64 bit signed EOF for dx directories */
2240 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2241 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2242
2243
2244 /*
2245  * Control parameters used by ext4_htree_next_block
2246  */
2247 #define HASH_NB_ALWAYS          1
2248
2249 struct ext4_filename {
2250         const struct qstr *usr_fname;
2251         struct fscrypt_str disk_name;
2252         struct dx_hash_info hinfo;
2253 #ifdef CONFIG_FS_ENCRYPTION
2254         struct fscrypt_str crypto_buf;
2255 #endif
2256 #ifdef CONFIG_UNICODE
2257         struct fscrypt_str cf_name;
2258 #endif
2259 };
2260
2261 #define fname_name(p) ((p)->disk_name.name)
2262 #define fname_len(p)  ((p)->disk_name.len)
2263
2264 /*
2265  * Describe an inode's exact location on disk and in memory
2266  */
2267 struct ext4_iloc
2268 {
2269         struct buffer_head *bh;
2270         unsigned long offset;
2271         ext4_group_t block_group;
2272 };
2273
2274 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2275 {
2276         return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2277 }
2278
2279 static inline bool ext4_is_quota_file(struct inode *inode)
2280 {
2281         return IS_NOQUOTA(inode) &&
2282                !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2283 }
2284
2285 /*
2286  * This structure is stuffed into the struct file's private_data field
2287  * for directories.  It is where we put information so that we can do
2288  * readdir operations in hash tree order.
2289  */
2290 struct dir_private_info {
2291         struct rb_root  root;
2292         struct rb_node  *curr_node;
2293         struct fname    *extra_fname;
2294         loff_t          last_pos;
2295         __u32           curr_hash;
2296         __u32           curr_minor_hash;
2297         __u32           next_hash;
2298 };
2299
2300 /* calculate the first block number of the group */
2301 static inline ext4_fsblk_t
2302 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2303 {
2304         return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2305                 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2306 }
2307
2308 /*
2309  * Special error return code only used by dx_probe() and its callers.
2310  */
2311 #define ERR_BAD_DX_DIR  (-(MAX_ERRNO - 1))
2312
2313 /* htree levels for ext4 */
2314 #define EXT4_HTREE_LEVEL_COMPAT 2
2315 #define EXT4_HTREE_LEVEL        3
2316
2317 static inline int ext4_dir_htree_level(struct super_block *sb)
2318 {
2319         return ext4_has_feature_largedir(sb) ?
2320                 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2321 }
2322
2323 /*
2324  * Timeout and state flag for lazy initialization inode thread.
2325  */
2326 #define EXT4_DEF_LI_WAIT_MULT                   10
2327 #define EXT4_DEF_LI_MAX_START_DELAY             5
2328 #define EXT4_LAZYINIT_QUIT                      0x0001
2329 #define EXT4_LAZYINIT_RUNNING                   0x0002
2330
2331 /*
2332  * Lazy inode table initialization info
2333  */
2334 struct ext4_lazy_init {
2335         unsigned long           li_state;
2336         struct list_head        li_request_list;
2337         struct mutex            li_list_mtx;
2338 };
2339
2340 enum ext4_li_mode {
2341         EXT4_LI_MODE_PREFETCH_BBITMAP,
2342         EXT4_LI_MODE_ITABLE,
2343 };
2344
2345 struct ext4_li_request {
2346         struct super_block      *lr_super;
2347         enum ext4_li_mode       lr_mode;
2348         ext4_group_t            lr_first_not_zeroed;
2349         ext4_group_t            lr_next_group;
2350         struct list_head        lr_request;
2351         unsigned long           lr_next_sched;
2352         unsigned long           lr_timeout;
2353 };
2354
2355 struct ext4_features {
2356         struct kobject f_kobj;
2357         struct completion f_kobj_unregister;
2358 };
2359
2360 /*
2361  * This structure will be used for multiple mount protection. It will be
2362  * written into the block number saved in the s_mmp_block field in the
2363  * superblock. Programs that check MMP should assume that if
2364  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2365  * to use the filesystem, regardless of how old the timestamp is.
2366  */
2367 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2368 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2369 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2370 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2371
2372 struct mmp_struct {
2373         __le32  mmp_magic;              /* Magic number for MMP */
2374         __le32  mmp_seq;                /* Sequence no. updated periodically */
2375
2376         /*
2377          * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2378          * purposes and do not affect the correctness of the algorithm
2379          */
2380         __le64  mmp_time;               /* Time last updated */
2381         char    mmp_nodename[64];       /* Node which last updated MMP block */
2382         char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
2383
2384         /*
2385          * mmp_check_interval is used to verify if the MMP block has been
2386          * updated on the block device. The value is updated based on the
2387          * maximum time to write the MMP block during an update cycle.
2388          */
2389         __le16  mmp_check_interval;
2390
2391         __le16  mmp_pad1;
2392         __le32  mmp_pad2[226];
2393         __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
2394 };
2395
2396 /* arguments passed to the mmp thread */
2397 struct mmpd_data {
2398         struct buffer_head *bh; /* bh from initial read_mmp_block() */
2399         struct super_block *sb;  /* super block of the fs */
2400 };
2401
2402 /*
2403  * Check interval multiplier
2404  * The MMP block is written every update interval and initially checked every
2405  * update interval x the multiplier (the value is then adapted based on the
2406  * write latency). The reason is that writes can be delayed under load and we
2407  * don't want readers to incorrectly assume that the filesystem is no longer
2408  * in use.
2409  */
2410 #define EXT4_MMP_CHECK_MULT             2UL
2411
2412 /*
2413  * Minimum interval for MMP checking in seconds.
2414  */
2415 #define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
2416
2417 /*
2418  * Maximum interval for MMP checking in seconds.
2419  */
2420 #define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
2421
2422 /*
2423  * Function prototypes
2424  */
2425
2426 /*
2427  * Ok, these declarations are also in <linux/kernel.h> but none of the
2428  * ext4 source programs needs to include it so they are duplicated here.
2429  */
2430 # define NORET_TYPE     /**/
2431 # define ATTRIB_NORET   __attribute__((noreturn))
2432 # define NORET_AND      noreturn,
2433
2434 /* bitmap.c */
2435 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2436 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2437                                 struct ext4_group_desc *gdp,
2438                                 struct buffer_head *bh, int sz);
2439 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2440                                   struct ext4_group_desc *gdp,
2441                                   struct buffer_head *bh, int sz);
2442 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2443                                 struct ext4_group_desc *gdp,
2444                                 struct buffer_head *bh);
2445 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2446                                   struct ext4_group_desc *gdp,
2447                                   struct buffer_head *bh);
2448
2449 /* balloc.c */
2450 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2451                                          ext4_fsblk_t blocknr,
2452                                          ext4_group_t *blockgrpp,
2453                                          ext4_grpblk_t *offsetp);
2454 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2455                                           ext4_fsblk_t block);
2456
2457 extern unsigned int ext4_block_group(struct super_block *sb,
2458                         ext4_fsblk_t blocknr);
2459 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2460                         ext4_fsblk_t blocknr);
2461 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2462 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2463                         ext4_group_t group);
2464 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2465                                          ext4_fsblk_t goal,
2466                                          unsigned int flags,
2467                                          unsigned long *count,
2468                                          int *errp);
2469 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2470                                     s64 nclusters, unsigned int flags);
2471 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2472 extern void ext4_check_blocks_bitmap(struct super_block *);
2473 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2474                                                     ext4_group_t block_group,
2475                                                     struct buffer_head ** bh);
2476 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2477
2478 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2479                                                 ext4_group_t block_group,
2480                                                 bool ignore_locked);
2481 extern int ext4_wait_block_bitmap(struct super_block *sb,
2482                                   ext4_group_t block_group,
2483                                   struct buffer_head *bh);
2484 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2485                                                   ext4_group_t block_group);
2486 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2487                                               ext4_group_t block_group,
2488                                               struct ext4_group_desc *gdp);
2489 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2490
2491 #ifdef CONFIG_UNICODE
2492 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2493                                          const struct qstr *iname,
2494                                          struct fscrypt_str *fname);
2495 #endif
2496
2497 #ifdef CONFIG_FS_ENCRYPTION
2498 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2499                                                 const struct fscrypt_name *src)
2500 {
2501         memset(dst, 0, sizeof(*dst));
2502
2503         dst->usr_fname = src->usr_fname;
2504         dst->disk_name = src->disk_name;
2505         dst->hinfo.hash = src->hash;
2506         dst->hinfo.minor_hash = src->minor_hash;
2507         dst->crypto_buf = src->crypto_buf;
2508 }
2509
2510 static inline int ext4_fname_setup_filename(struct inode *dir,
2511                                             const struct qstr *iname,
2512                                             int lookup,
2513                                             struct ext4_filename *fname)
2514 {
2515         struct fscrypt_name name;
2516         int err;
2517
2518         err = fscrypt_setup_filename(dir, iname, lookup, &name);
2519         if (err)
2520                 return err;
2521
2522         ext4_fname_from_fscrypt_name(fname, &name);
2523
2524 #ifdef CONFIG_UNICODE
2525         ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2526 #endif
2527         return 0;
2528 }
2529
2530 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2531                                             struct dentry *dentry,
2532                                             struct ext4_filename *fname)
2533 {
2534         struct fscrypt_name name;
2535         int err;
2536
2537         err = fscrypt_prepare_lookup(dir, dentry, &name);
2538         if (err)
2539                 return err;
2540
2541         ext4_fname_from_fscrypt_name(fname, &name);
2542
2543 #ifdef CONFIG_UNICODE
2544         ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2545 #endif
2546         return 0;
2547 }
2548
2549 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2550 {
2551         struct fscrypt_name name;
2552
2553         name.crypto_buf = fname->crypto_buf;
2554         fscrypt_free_filename(&name);
2555
2556         fname->crypto_buf.name = NULL;
2557         fname->usr_fname = NULL;
2558         fname->disk_name.name = NULL;
2559
2560 #ifdef CONFIG_UNICODE
2561         kfree(fname->cf_name.name);
2562         fname->cf_name.name = NULL;
2563 #endif
2564 }
2565 #else /* !CONFIG_FS_ENCRYPTION */
2566 static inline int ext4_fname_setup_filename(struct inode *dir,
2567                                             const struct qstr *iname,
2568                                             int lookup,
2569                                             struct ext4_filename *fname)
2570 {
2571         fname->usr_fname = iname;
2572         fname->disk_name.name = (unsigned char *) iname->name;
2573         fname->disk_name.len = iname->len;
2574
2575 #ifdef CONFIG_UNICODE
2576         ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2577 #endif
2578
2579         return 0;
2580 }
2581
2582 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2583                                             struct dentry *dentry,
2584                                             struct ext4_filename *fname)
2585 {
2586         return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2587 }
2588
2589 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2590 {
2591 #ifdef CONFIG_UNICODE
2592         kfree(fname->cf_name.name);
2593         fname->cf_name.name = NULL;
2594 #endif
2595 }
2596 #endif /* !CONFIG_FS_ENCRYPTION */
2597
2598 /* dir.c */
2599 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2600                                   struct file *,
2601                                   struct ext4_dir_entry_2 *,
2602                                   struct buffer_head *, char *, int,
2603                                   unsigned int);
2604 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
2605         unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2606                                         (de), (bh), (buf), (size), (offset)))
2607 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2608                                 __u32 minor_hash,
2609                                 struct ext4_dir_entry_2 *dirent,
2610                                 struct fscrypt_str *ent_name);
2611 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2612 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2613                              struct buffer_head *bh,
2614                              void *buf, int buf_size,
2615                              struct ext4_filename *fname,
2616                              struct ext4_dir_entry_2 **dest_de);
2617 void ext4_insert_dentry(struct inode *inode,
2618                         struct ext4_dir_entry_2 *de,
2619                         int buf_size,
2620                         struct ext4_filename *fname);
2621 static inline void ext4_update_dx_flag(struct inode *inode)
2622 {
2623         if (!ext4_has_feature_dir_index(inode->i_sb)) {
2624                 /* ext4_iget() should have caught this... */
2625                 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2626                 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2627         }
2628 }
2629 static const unsigned char ext4_filetype_table[] = {
2630         DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2631 };
2632
2633 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2634 {
2635         if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2636                 return DT_UNKNOWN;
2637
2638         return ext4_filetype_table[filetype];
2639 }
2640 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2641                              void *buf, int buf_size);
2642
2643 /* fsync.c */
2644 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2645
2646 /* hash.c */
2647 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2648                           struct dx_hash_info *hinfo);
2649
2650 /* ialloc.c */
2651 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2652                                       const struct qstr *qstr, __u32 goal,
2653                                       uid_t *owner, __u32 i_flags,
2654                                       int handle_type, unsigned int line_no,
2655                                       int nblocks);
2656
2657 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2658         __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2659                          i_flags, 0, 0, 0)
2660 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2661                                     type, nblocks)                  \
2662         __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2663                          0, (type), __LINE__, (nblocks))
2664
2665
2666 extern void ext4_free_inode(handle_t *, struct inode *);
2667 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2668 extern unsigned long ext4_count_free_inodes(struct super_block *);
2669 extern unsigned long ext4_count_dirs(struct super_block *);
2670 extern void ext4_check_inodes_bitmap(struct super_block *);
2671 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2672 extern int ext4_init_inode_table(struct super_block *sb,
2673                                  ext4_group_t group, int barrier);
2674 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2675
2676 /* mballoc.c */
2677 extern const struct seq_operations ext4_mb_seq_groups_ops;
2678 extern long ext4_mb_stats;
2679 extern long ext4_mb_max_to_scan;
2680 extern int ext4_mb_init(struct super_block *);
2681 extern int ext4_mb_release(struct super_block *);
2682 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2683                                 struct ext4_allocation_request *, int *);
2684 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2685 extern void ext4_discard_preallocations(struct inode *);
2686 extern int __init ext4_init_mballoc(void);
2687 extern void ext4_exit_mballoc(void);
2688 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2689                                      ext4_group_t group,
2690                                      unsigned int nr, int *cnt);
2691 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2692                                   unsigned int nr);
2693
2694 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2695                              struct buffer_head *bh, ext4_fsblk_t block,
2696                              unsigned long count, int flags);
2697 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2698                                    ext4_group_t ngroups);
2699 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2700                 ext4_group_t i, struct ext4_group_desc *desc);
2701 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2702                                 ext4_fsblk_t block, unsigned long count);
2703 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2704 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2705
2706 /* inode.c */
2707 int ext4_inode_is_fast_symlink(struct inode *inode);
2708 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2709 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2710 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2711                      bool wait, struct buffer_head **bhs);
2712 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2713                              struct buffer_head *bh_result, int create);
2714 int ext4_get_block(struct inode *inode, sector_t iblock,
2715                    struct buffer_head *bh_result, int create);
2716 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2717                            struct buffer_head *bh, int create);
2718 int ext4_walk_page_buffers(handle_t *handle,
2719                            struct buffer_head *head,
2720                            unsigned from,
2721                            unsigned to,
2722                            int *partial,
2723                            int (*fn)(handle_t *handle,
2724                                      struct buffer_head *bh));
2725 int do_journal_get_write_access(handle_t *handle,
2726                                 struct buffer_head *bh);
2727 #define FALL_BACK_TO_NONDELALLOC 1
2728 #define CONVERT_INLINE_DATA      2
2729
2730 typedef enum {
2731         EXT4_IGET_NORMAL =      0,
2732         EXT4_IGET_SPECIAL =     0x0001, /* OK to iget a system inode */
2733         EXT4_IGET_HANDLE =      0x0002  /* Inode # is from a handle */
2734 } ext4_iget_flags;
2735
2736 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2737                                  ext4_iget_flags flags, const char *function,
2738                                  unsigned int line);
2739
2740 #define ext4_iget(sb, ino, flags) \
2741         __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2742
2743 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2744 extern int  ext4_setattr(struct dentry *, struct iattr *);
2745 extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2746 extern void ext4_evict_inode(struct inode *);
2747 extern void ext4_clear_inode(struct inode *);
2748 extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2749 extern int  ext4_sync_inode(handle_t *, struct inode *);
2750 extern void ext4_dirty_inode(struct inode *, int);
2751 extern int ext4_change_inode_journal_flag(struct inode *, int);
2752 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2753 extern int ext4_inode_attach_jinode(struct inode *inode);
2754 extern int ext4_can_truncate(struct inode *inode);
2755 extern int ext4_truncate(struct inode *);
2756 extern int ext4_break_layouts(struct inode *);
2757 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2758 extern void ext4_set_inode_flags(struct inode *, bool init);
2759 extern int ext4_alloc_da_blocks(struct inode *inode);
2760 extern void ext4_set_aops(struct inode *inode);
2761 extern int ext4_writepage_trans_blocks(struct inode *);
2762 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2763 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2764                              loff_t lstart, loff_t lend);
2765 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2766 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2767 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2768 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2769 extern void ext4_da_release_space(struct inode *inode, int to_free);
2770 extern void ext4_da_update_reserve_space(struct inode *inode,
2771                                         int used, int quota_claim);
2772 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2773                               ext4_fsblk_t pblk, ext4_lblk_t len);
2774
2775 /* indirect.c */
2776 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2777                                 struct ext4_map_blocks *map, int flags);
2778 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2779 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2780 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2781                                  ext4_lblk_t start, ext4_lblk_t end);
2782
2783 /* ioctl.c */
2784 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2785 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2786
2787 /* migrate.c */
2788 extern int ext4_ext_migrate(struct inode *);
2789 extern int ext4_ind_migrate(struct inode *inode);
2790
2791 /* namei.c */
2792 extern int ext4_dirblock_csum_verify(struct inode *inode,
2793                                      struct buffer_head *bh);
2794 extern int ext4_orphan_add(handle_t *, struct inode *);
2795 extern int ext4_orphan_del(handle_t *, struct inode *);
2796 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2797                                 __u32 start_minor_hash, __u32 *next_hash);
2798 extern int ext4_search_dir(struct buffer_head *bh,
2799                            char *search_buf,
2800                            int buf_size,
2801                            struct inode *dir,
2802                            struct ext4_filename *fname,
2803                            unsigned int offset,
2804                            struct ext4_dir_entry_2 **res_dir);
2805 extern int ext4_generic_delete_entry(struct inode *dir,
2806                                      struct ext4_dir_entry_2 *de_del,
2807                                      struct buffer_head *bh,
2808                                      void *entry_buf,
2809                                      int buf_size,
2810                                      int csum_size);
2811 extern bool ext4_empty_dir(struct inode *inode);
2812
2813 /* resize.c */
2814 extern void ext4_kvfree_array_rcu(void *to_free);
2815 extern int ext4_group_add(struct super_block *sb,
2816                                 struct ext4_new_group_data *input);
2817 extern int ext4_group_extend(struct super_block *sb,
2818                                 struct ext4_super_block *es,
2819                                 ext4_fsblk_t n_blocks_count);
2820 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2821
2822 /* super.c */
2823 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2824                                          sector_t block, int op_flags);
2825 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2826 extern int ext4_calculate_overhead(struct super_block *sb);
2827 extern void ext4_superblock_csum_set(struct super_block *sb);
2828 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2829                                     ext4_group_t ngroup);
2830 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2831                                      char nbuf[16]);
2832 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2833                                              ext4_group_t block_group,
2834                                              unsigned int flags);
2835
2836 extern __printf(6, 7)
2837 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2838                   const char *, ...);
2839 extern __printf(6, 7)
2840 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2841                         ext4_fsblk_t, int, const char *, ...);
2842 extern __printf(5, 6)
2843 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2844                      const char *, ...);
2845 extern void __ext4_std_error(struct super_block *, const char *,
2846                              unsigned int, int);
2847 extern __printf(5, 6)
2848 void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2849                   const char *, ...);
2850 extern __printf(4, 5)
2851 void __ext4_warning(struct super_block *, const char *, unsigned int,
2852                     const char *, ...);
2853 extern __printf(4, 5)
2854 void __ext4_warning_inode(const struct inode *inode, const char *function,
2855                           unsigned int line, const char *fmt, ...);
2856 extern __printf(3, 4)
2857 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2858 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2859                            const char *, unsigned int, const char *);
2860 extern __printf(7, 8)
2861 void __ext4_grp_locked_error(const char *, unsigned int,
2862                              struct super_block *, ext4_group_t,
2863                              unsigned long, ext4_fsblk_t,
2864                              const char *, ...);
2865
2866 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2867         ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2868
2869 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)                     \
2870         __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2871
2872 #define ext4_error_inode_block(inode, block, err, fmt, a...)            \
2873         __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
2874                            (fmt), ## a)
2875
2876 #define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
2877         ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2878
2879 #ifdef CONFIG_PRINTK
2880
2881 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2882         __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2883 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)   \
2884         __ext4_error_inode((inode), (func), (line), (block),            \
2885                            (err), (fmt), ##__VA_ARGS__)
2886 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2887         __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2888 #define ext4_error(sb, fmt, ...)                                        \
2889         __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2890 #define ext4_error_err(sb, err, fmt, ...)                               \
2891         __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2892 #define ext4_abort(sb, err, fmt, ...)                                   \
2893         __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2894 #define ext4_warning(sb, fmt, ...)                                      \
2895         __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2896 #define ext4_warning_inode(inode, fmt, ...)                             \
2897         __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2898 #define ext4_msg(sb, level, fmt, ...)                           \
2899         __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2900 #define dump_mmp_msg(sb, mmp, msg)                                      \
2901         __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2902 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2903         __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2904                                 fmt, ##__VA_ARGS__)
2905
2906 #else
2907
2908 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2909 do {                                                                    \
2910         no_printk(fmt, ##__VA_ARGS__);                                  \
2911         __ext4_error_inode(inode, "", 0, block, 0, " ");                \
2912 } while (0)
2913 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)   \
2914 do {                                                                    \
2915         no_printk(fmt, ##__VA_ARGS__);                                  \
2916         __ext4_error_inode(inode, "", 0, block, err, " ");              \
2917 } while (0)
2918 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2919 do {                                                                    \
2920         no_printk(fmt, ##__VA_ARGS__);                                  \
2921         __ext4_error_file(file, "", 0, block, " ");                     \
2922 } while (0)
2923 #define ext4_error(sb, fmt, ...)                                        \
2924 do {                                                                    \
2925         no_printk(fmt, ##__VA_ARGS__);                                  \
2926         __ext4_error(sb, "", 0, 0, 0, " ");                             \
2927 } while (0)
2928 #define ext4_error_err(sb, err, fmt, ...)                               \
2929 do {                                                                    \
2930         no_printk(fmt, ##__VA_ARGS__);                                  \
2931         __ext4_error(sb, "", 0, err, 0, " ");                           \
2932 } while (0)
2933 #define ext4_abort(sb, err, fmt, ...)                                   \
2934 do {                                                                    \
2935         no_printk(fmt, ##__VA_ARGS__);                                  \
2936         __ext4_abort(sb, "", 0, err, " ");                              \
2937 } while (0)
2938 #define ext4_warning(sb, fmt, ...)                                      \
2939 do {                                                                    \
2940         no_printk(fmt, ##__VA_ARGS__);                                  \
2941         __ext4_warning(sb, "", 0, " ");                                 \
2942 } while (0)
2943 #define ext4_warning_inode(inode, fmt, ...)                             \
2944 do {                                                                    \
2945         no_printk(fmt, ##__VA_ARGS__);                                  \
2946         __ext4_warning_inode(inode, "", 0, " ");                        \
2947 } while (0)
2948 #define ext4_msg(sb, level, fmt, ...)                                   \
2949 do {                                                                    \
2950         no_printk(fmt, ##__VA_ARGS__);                                  \
2951         __ext4_msg(sb, "", " ");                                        \
2952 } while (0)
2953 #define dump_mmp_msg(sb, mmp, msg)                                      \
2954         __dump_mmp_msg(sb, mmp, "", 0, "")
2955 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2956 do {                                                                    \
2957         no_printk(fmt, ##__VA_ARGS__);                          \
2958         __ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");       \
2959 } while (0)
2960
2961 #endif
2962
2963 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2964                                       struct ext4_group_desc *bg);
2965 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2966                                       struct ext4_group_desc *bg);
2967 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2968                                      struct ext4_group_desc *bg);
2969 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2970                                       struct ext4_group_desc *bg);
2971 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2972                                  struct ext4_group_desc *bg);
2973 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2974                                 struct ext4_group_desc *bg);
2975 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2976                                    struct ext4_group_desc *bg);
2977 extern void ext4_block_bitmap_set(struct super_block *sb,
2978                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2979 extern void ext4_inode_bitmap_set(struct super_block *sb,
2980                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2981 extern void ext4_inode_table_set(struct super_block *sb,
2982                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2983 extern void ext4_free_group_clusters_set(struct super_block *sb,
2984                                          struct ext4_group_desc *bg,
2985                                          __u32 count);
2986 extern void ext4_free_inodes_set(struct super_block *sb,
2987                                 struct ext4_group_desc *bg, __u32 count);
2988 extern void ext4_used_dirs_set(struct super_block *sb,
2989                                 struct ext4_group_desc *bg, __u32 count);
2990 extern void ext4_itable_unused_set(struct super_block *sb,
2991                                    struct ext4_group_desc *bg, __u32 count);
2992 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2993                                        struct ext4_group_desc *gdp);
2994 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2995                                      struct ext4_group_desc *gdp);
2996 extern int ext4_register_li_request(struct super_block *sb,
2997                                     ext4_group_t first_not_zeroed);
2998
2999 static inline int ext4_has_metadata_csum(struct super_block *sb)
3000 {
3001         WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3002                      !EXT4_SB(sb)->s_chksum_driver);
3003
3004         return ext4_has_feature_metadata_csum(sb) &&
3005                (EXT4_SB(sb)->s_chksum_driver != NULL);
3006 }
3007
3008 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3009 {
3010         return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3011 }
3012
3013 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3014 {
3015         return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
3016                 le32_to_cpu(es->s_blocks_count_lo);
3017 }
3018
3019 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3020 {
3021         return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
3022                 le32_to_cpu(es->s_r_blocks_count_lo);
3023 }
3024
3025 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3026 {
3027         return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
3028                 le32_to_cpu(es->s_free_blocks_count_lo);
3029 }
3030
3031 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3032                                          ext4_fsblk_t blk)
3033 {
3034         es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3035         es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3036 }
3037
3038 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3039                                               ext4_fsblk_t blk)
3040 {
3041         es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3042         es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3043 }
3044
3045 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3046                                            ext4_fsblk_t blk)
3047 {
3048         es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3049         es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3050 }
3051
3052 static inline loff_t ext4_isize(struct super_block *sb,
3053                                 struct ext4_inode *raw_inode)
3054 {
3055         if (ext4_has_feature_largedir(sb) ||
3056             S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3057                 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3058                         le32_to_cpu(raw_inode->i_size_lo);
3059
3060         return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3061 }
3062
3063 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3064 {
3065         raw_inode->i_size_lo = cpu_to_le32(i_size);
3066         raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3067 }
3068
3069 static inline
3070 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3071                                             ext4_group_t group)
3072 {
3073          struct ext4_group_info **grp_info;
3074          long indexv, indexh;
3075          BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3076          indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3077          indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3078          grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3079          return grp_info[indexh];
3080 }
3081
3082 /*
3083  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3084  * the locking protocol documented in the comments of ext4_group_add()
3085  * in resize.c
3086  */
3087 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3088 {
3089         ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
3090
3091         smp_rmb();
3092         return ngroups;
3093 }
3094
3095 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3096                                              ext4_group_t block_group)
3097 {
3098         return block_group >> sbi->s_log_groups_per_flex;
3099 }
3100
3101 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3102 {
3103         return 1 << sbi->s_log_groups_per_flex;
3104 }
3105
3106 #define ext4_std_error(sb, errno)                               \
3107 do {                                                            \
3108         if ((errno))                                            \
3109                 __ext4_std_error((sb), __func__, __LINE__, (errno));    \
3110 } while (0)
3111
3112 #ifdef CONFIG_SMP
3113 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3114  * counters. So we need to make sure we have free clusters more
3115  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3116  */
3117 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3118 #else
3119 #define EXT4_FREECLUSTERS_WATERMARK 0
3120 #endif
3121
3122 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3123 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3124 {
3125         WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3126                      !inode_is_locked(inode));
3127         down_write(&EXT4_I(inode)->i_data_sem);
3128         if (newsize > EXT4_I(inode)->i_disksize)
3129                 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3130         up_write(&EXT4_I(inode)->i_data_sem);
3131 }
3132
3133 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3134 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3135 {
3136         int changed = 0;
3137
3138         if (newsize > inode->i_size) {
3139                 i_size_write(inode, newsize);
3140                 changed = 1;
3141         }
3142         if (newsize > EXT4_I(inode)->i_disksize) {
3143                 ext4_update_i_disksize(inode, newsize);
3144                 changed |= 2;
3145         }
3146         return changed;
3147 }
3148
3149 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3150                                       loff_t len);
3151
3152 struct ext4_group_info {
3153         unsigned long   bb_state;
3154         struct rb_root  bb_free_root;
3155         ext4_grpblk_t   bb_first_free;  /* first free block */
3156         ext4_grpblk_t   bb_free;        /* total free blocks */
3157         ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
3158         ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
3159         struct          list_head bb_prealloc_list;
3160 #ifdef DOUBLE_CHECK
3161         void            *bb_bitmap;
3162 #endif
3163         struct rw_semaphore alloc_sem;
3164         ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
3165                                          * regions, index is order.
3166                                          * bb_counters[3] = 5 means
3167                                          * 5 free 8-block regions. */
3168 };
3169
3170 #define EXT4_GROUP_INFO_NEED_INIT_BIT           0
3171 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
3172 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT     2
3173 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT     3
3174 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT         \
3175         (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3176 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT         \
3177         (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3178 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT        4
3179
3180 #define EXT4_MB_GRP_NEED_INIT(grp)      \
3181         (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3182 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)        \
3183         (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3184 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)        \
3185         (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3186
3187 #define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
3188         (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3189 #define EXT4_MB_GRP_SET_TRIMMED(grp)    \
3190         (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3191 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
3192         (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3193 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)      \
3194         (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3195
3196 #define EXT4_MAX_CONTENTION             8
3197 #define EXT4_CONTENTION_THRESHOLD       2
3198
3199 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3200                                               ext4_group_t group)
3201 {
3202         return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3203 }
3204
3205 /*
3206  * Returns true if the filesystem is busy enough that attempts to
3207  * access the block group locks has run into contention.
3208  */
3209 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3210 {
3211         return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3212 }
3213
3214 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3215 {
3216         spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3217         if (spin_trylock(lock))
3218                 /*
3219                  * We're able to grab the lock right away, so drop the
3220                  * lock contention counter.
3221                  */
3222                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3223         else {
3224                 /*
3225                  * The lock is busy, so bump the contention counter,
3226                  * and then wait on the spin lock.
3227                  */
3228                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3229                                   EXT4_MAX_CONTENTION);
3230                 spin_lock(lock);
3231         }
3232 }
3233
3234 static inline void ext4_unlock_group(struct super_block *sb,
3235                                         ext4_group_t group)
3236 {
3237         spin_unlock(ext4_group_lock_ptr(sb, group));
3238 }
3239
3240 /*
3241  * Block validity checking
3242  */
3243 #define ext4_check_indirect_blockref(inode, bh)                         \
3244         ext4_check_blockref(__func__, __LINE__, inode,                  \
3245                             (__le32 *)(bh)->b_data,                     \
3246                             EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3247
3248 #define ext4_ind_check_inode(inode)                                     \
3249         ext4_check_blockref(__func__, __LINE__, inode,                  \
3250                             EXT4_I(inode)->i_data,                      \
3251                             EXT4_NDIR_BLOCKS)
3252
3253 /*
3254  * Inodes and files operations
3255  */
3256
3257 /* dir.c */
3258 extern const struct file_operations ext4_dir_operations;
3259
3260 #ifdef CONFIG_UNICODE
3261 extern const struct dentry_operations ext4_dentry_ops;
3262 #endif
3263
3264 /* file.c */
3265 extern const struct inode_operations ext4_file_inode_operations;
3266 extern const struct file_operations ext4_file_operations;
3267 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3268
3269 /* inline.c */
3270 extern int ext4_get_max_inline_size(struct inode *inode);
3271 extern int ext4_find_inline_data_nolock(struct inode *inode);
3272 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3273                                  unsigned int len);
3274 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3275
3276 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3277 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3278                                          struct inode *inode,
3279                                          loff_t pos, unsigned len,
3280                                          unsigned flags,
3281                                          struct page **pagep);
3282 extern int ext4_write_inline_data_end(struct inode *inode,
3283                                       loff_t pos, unsigned len,
3284                                       unsigned copied,
3285                                       struct page *page);
3286 extern struct buffer_head *
3287 ext4_journalled_write_inline_data(struct inode *inode,
3288                                   unsigned len,
3289                                   struct page *page);
3290 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3291                                            struct inode *inode,
3292                                            loff_t pos, unsigned len,
3293                                            unsigned flags,
3294                                            struct page **pagep,
3295                                            void **fsdata);
3296 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3297                                          unsigned len, unsigned copied,
3298                                          struct page *page);
3299 extern int ext4_try_add_inline_entry(handle_t *handle,
3300                                      struct ext4_filename *fname,
3301                                      struct inode *dir, struct inode *inode);
3302 extern int ext4_try_create_inline_dir(handle_t *handle,
3303                                       struct inode *parent,
3304                                       struct inode *inode);
3305 extern int ext4_read_inline_dir(struct file *filp,
3306                                 struct dir_context *ctx,
3307                                 int *has_inline_data);
3308 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3309                                   struct inode *dir, ext4_lblk_t block,
3310                                   struct dx_hash_info *hinfo,
3311                                   __u32 start_hash, __u32 start_minor_hash,
3312                                   int *has_inline_data);
3313 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3314                                         struct ext4_filename *fname,
3315                                         struct ext4_dir_entry_2 **res_dir,
3316                                         int *has_inline_data);
3317 extern int ext4_delete_inline_entry(handle_t *handle,
3318                                     struct inode *dir,
3319                                     struct ext4_dir_entry_2 *de_del,
3320                                     struct buffer_head *bh,
3321                                     int *has_inline_data);
3322 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3323 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3324                                         struct ext4_dir_entry_2 **parent_de,
3325                                         int *retval);
3326 extern int ext4_inline_data_fiemap(struct inode *inode,
3327                                    struct fiemap_extent_info *fieinfo,
3328                                    int *has_inline, __u64 start, __u64 len);
3329
3330 struct iomap;
3331 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3332
3333 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3334
3335 extern int ext4_convert_inline_data(struct inode *inode);
3336
3337 static inline int ext4_has_inline_data(struct inode *inode)
3338 {
3339         return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3340                EXT4_I(inode)->i_inline_off;
3341 }
3342
3343 /* namei.c */
3344 extern const struct inode_operations ext4_dir_inode_operations;
3345 extern const struct inode_operations ext4_special_inode_operations;
3346 extern struct dentry *ext4_get_parent(struct dentry *child);
3347 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3348                                  struct ext4_dir_entry_2 *de,
3349                                  int blocksize, int csum_size,
3350                                  unsigned int parent_ino, int dotdot_real_len);
3351 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3352                                         unsigned int blocksize);
3353 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3354                                       struct buffer_head *bh);
3355 extern int ext4_ci_compare(const struct inode *parent,
3356                            const struct qstr *fname,
3357                            const struct qstr *entry, bool quick);
3358
3359 #define S_SHIFT 12
3360 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3361         [S_IFREG >> S_SHIFT]    = EXT4_FT_REG_FILE,
3362         [S_IFDIR >> S_SHIFT]    = EXT4_FT_DIR,
3363         [S_IFCHR >> S_SHIFT]    = EXT4_FT_CHRDEV,
3364         [S_IFBLK >> S_SHIFT]    = EXT4_FT_BLKDEV,
3365         [S_IFIFO >> S_SHIFT]    = EXT4_FT_FIFO,
3366         [S_IFSOCK >> S_SHIFT]   = EXT4_FT_SOCK,
3367         [S_IFLNK >> S_SHIFT]    = EXT4_FT_SYMLINK,
3368 };
3369
3370 static inline void ext4_set_de_type(struct super_block *sb,
3371                                 struct ext4_dir_entry_2 *de,
3372                                 umode_t mode) {
3373         if (ext4_has_feature_filetype(sb))
3374                 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3375 }
3376
3377 /* readpages.c */
3378 extern int ext4_mpage_readpages(struct inode *inode,
3379                 struct readahead_control *rac, struct page *page);
3380 extern int __init ext4_init_post_read_processing(void);
3381 extern void ext4_exit_post_read_processing(void);
3382
3383 /* symlink.c */
3384 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3385 extern const struct inode_operations ext4_symlink_inode_operations;
3386 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3387
3388 /* sysfs.c */
3389 extern int ext4_register_sysfs(struct super_block *sb);
3390 extern void ext4_unregister_sysfs(struct super_block *sb);
3391 extern int __init ext4_init_sysfs(void);
3392 extern void ext4_exit_sysfs(void);
3393
3394 /* block_validity */
3395 extern void ext4_release_system_zone(struct super_block *sb);
3396 extern int ext4_setup_system_zone(struct super_block *sb);
3397 extern int __init ext4_init_system_zone(void);
3398 extern void ext4_exit_system_zone(void);
3399 extern int ext4_inode_block_valid(struct inode *inode,
3400                                   ext4_fsblk_t start_blk,
3401                                   unsigned int count);
3402 extern int ext4_check_blockref(const char *, unsigned int,
3403                                struct inode *, __le32 *, unsigned int);
3404
3405 /* extents.c */
3406 struct ext4_ext_path;
3407 struct ext4_extent;
3408
3409 /*
3410  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3411  * __le32.
3412  */
3413 #define EXT_MAX_BLOCKS  0xffffffff
3414
3415 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3416 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3417 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3418                                struct ext4_map_blocks *map, int flags);
3419 extern int ext4_ext_truncate(handle_t *, struct inode *);
3420 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3421                                  ext4_lblk_t end);
3422 extern void ext4_ext_init(struct super_block *);
3423 extern void ext4_ext_release(struct super_block *);
3424 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3425                           loff_t len);
3426 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3427                                           loff_t offset, ssize_t len);
3428 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3429                                              ext4_io_end_t *io_end);
3430 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3431                            struct ext4_map_blocks *map, int flags);
3432 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3433                                                    int num,
3434                                                    struct ext4_ext_path *path);
3435 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3436                                   struct ext4_ext_path **,
3437                                   struct ext4_extent *, int);
3438 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3439                                               struct ext4_ext_path **,
3440                                               int flags);
3441 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3442 extern int ext4_ext_check_inode(struct inode *inode);
3443 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3444 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3445                         __u64 start, __u64 len);
3446 extern int ext4_get_es_cache(struct inode *inode,
3447                              struct fiemap_extent_info *fieinfo,
3448                              __u64 start, __u64 len);
3449 extern int ext4_ext_precache(struct inode *inode);
3450 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3451                                 struct inode *inode2, ext4_lblk_t lblk1,
3452                              ext4_lblk_t lblk2,  ext4_lblk_t count,
3453                              int mark_unwritten,int *err);
3454 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3455 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3456                                        int check_cred, int restart_cred,
3457                                        int revoke_cred);
3458
3459
3460 /* move_extent.c */
3461 extern void ext4_double_down_write_data_sem(struct inode *first,
3462                                             struct inode *second);
3463 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3464                                           struct inode *donor_inode);
3465 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3466                              __u64 start_orig, __u64 start_donor,
3467                              __u64 len, __u64 *moved_len);
3468
3469 /* page-io.c */
3470 extern int __init ext4_init_pageio(void);
3471 extern void ext4_exit_pageio(void);
3472 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3473 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3474 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3475 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3476 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3477                                 struct writeback_control *wbc);
3478 extern void ext4_end_io_rsv_work(struct work_struct *work);
3479 extern void ext4_io_submit(struct ext4_io_submit *io);
3480 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3481                                struct page *page,
3482                                int len,
3483                                struct writeback_control *wbc,
3484                                bool keep_towrite);
3485 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3486 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3487
3488 /* mmp.c */
3489 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3490
3491 /* verity.c */
3492 extern const struct fsverity_operations ext4_verityops;
3493
3494 /*
3495  * Add new method to test whether block and inode bitmaps are properly
3496  * initialized. With uninit_bg reading the block from disk is not enough
3497  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3498  */
3499 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3500
3501 static inline int bitmap_uptodate(struct buffer_head *bh)
3502 {
3503         return (buffer_uptodate(bh) &&
3504                         test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3505 }
3506 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3507 {
3508         set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3509 }
3510
3511 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3512
3513 /* For ioend & aio unwritten conversion wait queues */
3514 #define EXT4_WQ_HASH_SZ         37
3515 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3516                                             EXT4_WQ_HASH_SZ])
3517 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3518
3519 extern int ext4_resize_begin(struct super_block *sb);
3520 extern void ext4_resize_end(struct super_block *sb);
3521
3522 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3523                                               struct ext4_io_end *io_end)
3524 {
3525         if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3526                 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3527                 atomic_inc(&EXT4_I(inode)->i_unwritten);
3528         }
3529 }
3530
3531 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3532 {
3533         struct inode *inode = io_end->inode;
3534
3535         if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3536                 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3537                 /* Wake up anyone waiting on unwritten extent conversion */
3538                 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3539                         wake_up_all(ext4_ioend_wq(inode));
3540         }
3541 }
3542
3543 extern const struct iomap_ops ext4_iomap_ops;
3544 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3545 extern const struct iomap_ops ext4_iomap_report_ops;
3546
3547 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3548 {
3549         /*
3550          * If the buffer has the write error flag, we have failed
3551          * to write out data in the block.  In this  case, we don't
3552          * have to read the block because we may read the old data
3553          * successfully.
3554          */
3555         if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3556                 set_buffer_uptodate(bh);
3557         return buffer_uptodate(bh);
3558 }
3559
3560 #endif  /* __KERNEL__ */
3561
3562 #define EFSBADCRC       EBADMSG         /* Bad CRC detected */
3563 #define EFSCORRUPTED    EUCLEAN         /* Filesystem is corrupted */
3564
3565 #endif  /* _EXT4_H */