LXR linux/fs/erofs/erofs

   1/* SPDX-License-Identifier: GPL-2.0-only OR Apache-2.0 */
   2/*
   3 * EROFS (Enhanced ROM File System) on-disk format definition
   4 *
   5 * Copyright (C) 2017-2018 HUAWEI, Inc.
   6 *             https://www.huawei.com/
   7 * Copyright (C) 2021, Alibaba Cloud
   8 */
   9#ifndef __EROFS_FS_H
  10#define __EROFS_FS_H
  11
  12#define EROFS_SUPER_OFFSET      1024
  13
  14#define EROFS_FEATURE_COMPAT_SB_CHKSUM          0x00000001
  15
  16/*
  17 * Any bits that aren't in EROFS_ALL_FEATURE_INCOMPAT should
  18 * be incompatible with this kernel version.
  19 */
  20#define EROFS_FEATURE_INCOMPAT_ZERO_PADDING     0x00000001
  21#define EROFS_FEATURE_INCOMPAT_COMPR_CFGS       0x00000002
  22#define EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER     0x00000002
  23#define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE     0x00000004
  24#define EROFS_FEATURE_INCOMPAT_DEVICE_TABLE     0x00000008
  25#define EROFS_FEATURE_INCOMPAT_COMPR_HEAD2      0x00000008
  26#define EROFS_FEATURE_INCOMPAT_ZTAILPACKING     0x00000010
  27#define EROFS_ALL_FEATURE_INCOMPAT              \
  28        (EROFS_FEATURE_INCOMPAT_ZERO_PADDING | \
  29         EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
  30         EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER | \
  31         EROFS_FEATURE_INCOMPAT_CHUNKED_FILE | \
  32         EROFS_FEATURE_INCOMPAT_DEVICE_TABLE | \
  33         EROFS_FEATURE_INCOMPAT_COMPR_HEAD2 | \
  34         EROFS_FEATURE_INCOMPAT_ZTAILPACKING)
  35
  36#define EROFS_SB_EXTSLOT_SIZE   16
  37
  38struct erofs_deviceslot {
  39        union {
  40                u8 uuid[16];            /* used for device manager later */
  41                u8 userdata[64];        /* digest(sha256), etc. */
  42        } u;
  43        __le32 blocks;                  /* total fs blocks of this device */
  44        __le32 mapped_blkaddr;          /* map starting at mapped_blkaddr */
  45        u8 reserved[56];
  46};
  47#define EROFS_DEVT_SLOT_SIZE    sizeof(struct erofs_deviceslot)
  48
  49/* erofs on-disk super block (currently 128 bytes) */
  50struct erofs_super_block {
  51        __le32 magic;           /* file system magic number */
  52        __le32 checksum;        /* crc32c(super_block) */
  53        __le32 feature_compat;
  54        __u8 blkszbits;         /* support block_size == PAGE_SIZE only */
  55        __u8 sb_extslots;       /* superblock size = 128 + sb_extslots * 16 */
  56
  57        __le16 root_nid;        /* nid of root directory */
  58        __le64 inos;            /* total valid ino # (== f_files - f_favail) */
  59
  60        __le64 build_time;      /* inode v1 time derivation */
  61        __le32 build_time_nsec; /* inode v1 time derivation in nano scale */
  62        __le32 blocks;          /* used for statfs */
  63        __le32 meta_blkaddr;    /* start block address of metadata area */
  64        __le32 xattr_blkaddr;   /* start block address of shared xattr area */
  65        __u8 uuid[16];          /* 128-bit uuid for volume */
  66        __u8 volume_name[16];   /* volume name */
  67        __le32 feature_incompat;
  68        union {
  69                /* bitmap for available compression algorithms */
  70                __le16 available_compr_algs;
  71                /* customized sliding window size instead of 64k by default */
  72                __le16 lz4_max_distance;
  73        } __packed u1;
  74        __le16 extra_devices;   /* # of devices besides the primary device */
  75        __le16 devt_slotoff;    /* startoff = devt_slotoff * devt_slotsize */
  76        __u8 reserved2[38];
  77};
  78
  79/*
  80 * erofs inode datalayout (i_format in on-disk inode):
  81 * 0 - inode plain without inline data A:
  82 * inode, [xattrs], ... | ... | no-holed data
  83 * 1 - inode VLE compression B (legacy):
  84 * inode, [xattrs], extents ... | ...
  85 * 2 - inode plain with inline data C:
  86 * inode, [xattrs], last_inline_data, ... | ... | no-holed data
  87 * 3 - inode compression D:
  88 * inode, [xattrs], map_header, extents ... | ...
  89 * 4 - inode chunk-based E:
  90 * inode, [xattrs], chunk indexes ... | ...
  91 * 5~7 - reserved
  92 */
  93enum {
  94        EROFS_INODE_FLAT_PLAIN                  = 0,
  95        EROFS_INODE_FLAT_COMPRESSION_LEGACY     = 1,
  96        EROFS_INODE_FLAT_INLINE                 = 2,
  97        EROFS_INODE_FLAT_COMPRESSION            = 3,
  98        EROFS_INODE_CHUNK_BASED                 = 4,
  99        EROFS_INODE_DATALAYOUT_MAX
 100};
 101
 102static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
 103{
 104        return datamode == EROFS_INODE_FLAT_COMPRESSION ||
 105                datamode == EROFS_INODE_FLAT_COMPRESSION_LEGACY;
 106}
 107
 108/* bit definitions of inode i_advise */
 109#define EROFS_I_VERSION_BITS            1
 110#define EROFS_I_DATALAYOUT_BITS         3
 111
 112#define EROFS_I_VERSION_BIT             0
 113#define EROFS_I_DATALAYOUT_BIT          1
 114
 115#define EROFS_I_ALL     \
 116        ((1 << (EROFS_I_DATALAYOUT_BIT + EROFS_I_DATALAYOUT_BITS)) - 1)
 117
 118/* indicate chunk blkbits, thus 'chunksize = blocksize << chunk blkbits' */
 119#define EROFS_CHUNK_FORMAT_BLKBITS_MASK         0x001F
 120/* with chunk indexes or just a 4-byte blkaddr array */
 121#define EROFS_CHUNK_FORMAT_INDEXES              0x0020
 122
 123#define EROFS_CHUNK_FORMAT_ALL  \
 124        (EROFS_CHUNK_FORMAT_BLKBITS_MASK | EROFS_CHUNK_FORMAT_INDEXES)
 125
 126struct erofs_inode_chunk_info {
 127        __le16 format;          /* chunk blkbits, etc. */
 128        __le16 reserved;
 129};
 130
 131/* 32-byte reduced form of an ondisk inode */
 132struct erofs_inode_compact {
 133        __le16 i_format;        /* inode format hints */
 134
 135/* 1 header + n-1 * 4 bytes inline xattr to keep continuity */
 136        __le16 i_xattr_icount;
 137        __le16 i_mode;
 138        __le16 i_nlink;
 139        __le32 i_size;
 140        __le32 i_reserved;
 141        union {
 142                /* file total compressed blocks for data mapping 1 */
 143                __le32 compressed_blocks;
 144                __le32 raw_blkaddr;
 145
 146                /* for device files, used to indicate old/new device # */
 147                __le32 rdev;
 148
 149                /* for chunk-based files, it contains the summary info */
 150                struct erofs_inode_chunk_info c;
 151        } i_u;
 152        __le32 i_ino;           /* only used for 32-bit stat compatibility */
 153        __le16 i_uid;
 154        __le16 i_gid;
 155        __le32 i_reserved2;
 156};
 157
 158/* 32 bytes on-disk inode */
 159#define EROFS_INODE_LAYOUT_COMPACT      0
 160/* 64 bytes on-disk inode */
 161#define EROFS_INODE_LAYOUT_EXTENDED     1
 162
 163/* 64-byte complete form of an ondisk inode */
 164struct erofs_inode_extended {
 165        __le16 i_format;        /* inode format hints */
 166
 167/* 1 header + n-1 * 4 bytes inline xattr to keep continuity */
 168        __le16 i_xattr_icount;
 169        __le16 i_mode;
 170        __le16 i_reserved;
 171        __le64 i_size;
 172        union {
 173                /* file total compressed blocks for data mapping 1 */
 174                __le32 compressed_blocks;
 175                __le32 raw_blkaddr;
 176
 177                /* for device files, used to indicate old/new device # */
 178                __le32 rdev;
 179
 180                /* for chunk-based files, it contains the summary info */
 181                struct erofs_inode_chunk_info c;
 182        } i_u;
 183
 184        /* only used for 32-bit stat compatibility */
 185        __le32 i_ino;
 186
 187        __le32 i_uid;
 188        __le32 i_gid;
 189        __le64 i_ctime;
 190        __le32 i_ctime_nsec;
 191        __le32 i_nlink;
 192        __u8   i_reserved2[16];
 193};
 194
 195#define EROFS_MAX_SHARED_XATTRS         (128)
 196/* h_shared_count between 129 ... 255 are special # */
 197#define EROFS_SHARED_XATTR_EXTENT       (255)
 198
 199/*
 200 * inline xattrs (n == i_xattr_icount):
 201 * erofs_xattr_ibody_header(1) + (n - 1) * 4 bytes
 202 *          12 bytes           /                   \
 203 *                            /                     \
 204 *                           /-----------------------\
 205 *                           |  erofs_xattr_entries+ |
 206 *                           +-----------------------+
 207 * inline xattrs must starts in erofs_xattr_ibody_header,
 208 * for read-only fs, no need to introduce h_refcount
 209 */
 210struct erofs_xattr_ibody_header {
 211        __le32 h_reserved;
 212        __u8   h_shared_count;
 213        __u8   h_reserved2[7];
 214        __le32 h_shared_xattrs[];       /* shared xattr id array */
 215};
 216
 217/* Name indexes */
 218#define EROFS_XATTR_INDEX_USER              1
 219#define EROFS_XATTR_INDEX_POSIX_ACL_ACCESS  2
 220#define EROFS_XATTR_INDEX_POSIX_ACL_DEFAULT 3
 221#define EROFS_XATTR_INDEX_TRUSTED           4
 222#define EROFS_XATTR_INDEX_LUSTRE            5
 223#define EROFS_XATTR_INDEX_SECURITY          6
 224
 225/* xattr entry (for both inline & shared xattrs) */
 226struct erofs_xattr_entry {
 227        __u8   e_name_len;      /* length of name */
 228        __u8   e_name_index;    /* attribute name index */
 229        __le16 e_value_size;    /* size of attribute value */
 230        /* followed by e_name and e_value */
 231        char   e_name[];        /* attribute name */
 232};
 233
 234static inline unsigned int erofs_xattr_ibody_size(__le16 i_xattr_icount)
 235{
 236        if (!i_xattr_icount)
 237                return 0;
 238
 239        return sizeof(struct erofs_xattr_ibody_header) +
 240                sizeof(__u32) * (le16_to_cpu(i_xattr_icount) - 1);
 241}
 242
 243#define EROFS_XATTR_ALIGN(size) round_up(size, sizeof(struct erofs_xattr_entry))
 244
 245static inline unsigned int erofs_xattr_entry_size(struct erofs_xattr_entry *e)
 246{
 247        return EROFS_XATTR_ALIGN(sizeof(struct erofs_xattr_entry) +
 248                                 e->e_name_len + le16_to_cpu(e->e_value_size));
 249}
 250
 251/* represent a zeroed chunk (hole) */
 252#define EROFS_NULL_ADDR                 -1
 253
 254/* 4-byte block address array */
 255#define EROFS_BLOCK_MAP_ENTRY_SIZE      sizeof(__le32)
 256
 257/* 8-byte inode chunk indexes */
 258struct erofs_inode_chunk_index {
 259        __le16 advise;          /* always 0, don't care for now */
 260        __le16 device_id;       /* back-end storage id (with bits masked) */
 261        __le32 blkaddr;         /* start block address of this inode chunk */
 262};
 263
 264/* maximum supported size of a physical compression cluster */
 265#define Z_EROFS_PCLUSTER_MAX_SIZE       (1024 * 1024)
 266
 267/* available compression algorithm types (for h_algorithmtype) */
 268enum {
 269        Z_EROFS_COMPRESSION_LZ4         = 0,
 270        Z_EROFS_COMPRESSION_LZMA        = 1,
 271        Z_EROFS_COMPRESSION_MAX
 272};
 273#define Z_EROFS_ALL_COMPR_ALGS          ((1 << Z_EROFS_COMPRESSION_MAX) - 1)
 274
 275/* 14 bytes (+ length field = 16 bytes) */
 276struct z_erofs_lz4_cfgs {
 277        __le16 max_distance;
 278        __le16 max_pclusterblks;
 279        u8 reserved[10];
 280} __packed;
 281
 282/* 14 bytes (+ length field = 16 bytes) */
 283struct z_erofs_lzma_cfgs {
 284        __le32 dict_size;
 285        __le16 format;
 286        u8 reserved[8];
 287} __packed;
 288
 289#define Z_EROFS_LZMA_MAX_DICT_SIZE      (8 * Z_EROFS_PCLUSTER_MAX_SIZE)
 290
 291/*
 292 * bit 0 : COMPACTED_2B indexes (0 - off; 1 - on)
 293 *  e.g. for 4k logical cluster size,      4B        if compacted 2B is off;
 294 *                                  (4B) + 2B + (4B) if compacted 2B is on.
 295 * bit 1 : HEAD1 big pcluster (0 - off; 1 - on)
 296 * bit 2 : HEAD2 big pcluster (0 - off; 1 - on)
 297 * bit 3 : tailpacking inline pcluster (0 - off; 1 - on)
 298 */
 299#define Z_EROFS_ADVISE_COMPACTED_2B             0x0001
 300#define Z_EROFS_ADVISE_BIG_PCLUSTER_1           0x0002
 301#define Z_EROFS_ADVISE_BIG_PCLUSTER_2           0x0004
 302#define Z_EROFS_ADVISE_INLINE_PCLUSTER          0x0008
 303
 304struct z_erofs_map_header {
 305        __le16  h_reserved1;
 306        /* indicates the encoded size of tailpacking data */
 307        __le16  h_idata_size;
 308        __le16  h_advise;
 309        /*
 310         * bit 0-3 : algorithm type of head 1 (logical cluster type 01);
 311         * bit 4-7 : algorithm type of head 2 (logical cluster type 11).
 312         */
 313        __u8    h_algorithmtype;
 314        /*
 315         * bit 0-2 : logical cluster bits - 12, e.g. 0 for 4096;
 316         * bit 3-7 : reserved.
 317         */
 318        __u8    h_clusterbits;
 319};
 320
 321#define Z_EROFS_VLE_LEGACY_HEADER_PADDING       8
 322
 323/*
 324 * Fixed-sized output compression on-disk logical cluster type:
 325 *    0   - literal (uncompressed) lcluster
 326 *    1,3 - compressed lcluster (for HEAD lclusters)
 327 *    2   - compressed lcluster (for NONHEAD lclusters)
 328 *
 329 * In detail,
 330 *    0 - literal (uncompressed) lcluster,
 331 *        di_advise = 0
 332 *        di_clusterofs = the literal data offset of the lcluster
 333 *        di_blkaddr = the blkaddr of the literal pcluster
 334 *
 335 *    1,3 - compressed lcluster (for HEAD lclusters)
 336 *        di_advise = 1 or 3
 337 *        di_clusterofs = the decompressed data offset of the lcluster
 338 *        di_blkaddr = the blkaddr of the compressed pcluster
 339 *
 340 *    2 - compressed lcluster (for NONHEAD lclusters)
 341 *        di_advise = 2
 342 *        di_clusterofs =
 343 *           the decompressed data offset in its own HEAD lcluster
 344 *        di_u.delta[0] = distance to this HEAD lcluster
 345 *        di_u.delta[1] = distance to the next HEAD lcluster
 346 */
 347enum {
 348        Z_EROFS_VLE_CLUSTER_TYPE_PLAIN          = 0,
 349        Z_EROFS_VLE_CLUSTER_TYPE_HEAD1          = 1,
 350        Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD        = 2,
 351        Z_EROFS_VLE_CLUSTER_TYPE_HEAD2          = 3,
 352        Z_EROFS_VLE_CLUSTER_TYPE_MAX
 353};
 354
 355#define Z_EROFS_VLE_DI_CLUSTER_TYPE_BITS        2
 356#define Z_EROFS_VLE_DI_CLUSTER_TYPE_BIT         0
 357
 358/*
 359 * D0_CBLKCNT will be marked _only_ at the 1st non-head lcluster to store the
 360 * compressed block count of a compressed extent (in logical clusters, aka.
 361 * block count of a pcluster).
 362 */
 363#define Z_EROFS_VLE_DI_D0_CBLKCNT               (1 << 11)
 364
 365struct z_erofs_vle_decompressed_index {
 366        __le16 di_advise;
 367        /* where to decompress in the head cluster */
 368        __le16 di_clusterofs;
 369
 370        union {
 371                /* for the head cluster */
 372                __le32 blkaddr;
 373                /*
 374                 * for the rest clusters
 375                 * eg. for 4k page-sized cluster, maximum 4K*64k = 256M)
 376                 * [0] - pointing to the head cluster
 377                 * [1] - pointing to the tail cluster
 378                 */
 379                __le16 delta[2];
 380        } di_u;
 381};
 382
 383#define Z_EROFS_VLE_LEGACY_INDEX_ALIGN(size) \
 384        (round_up(size, sizeof(struct z_erofs_vle_decompressed_index)) + \
 385         sizeof(struct z_erofs_map_header) + Z_EROFS_VLE_LEGACY_HEADER_PADDING)
 386
 387/* dirent sorts in alphabet order, thus we can do binary search */
 388struct erofs_dirent {
 389        __le64 nid;     /* node number */
 390        __le16 nameoff; /* start offset of file name */
 391        __u8 file_type; /* file type */
 392        __u8 reserved;  /* reserved */
 393} __packed;
 394
 395/*
 396 * EROFS file types should match generic FT_* types and
 397 * it seems no need to add BUILD_BUG_ONs since potential
 398 * unmatchness will break other fses as well...
 399 */
 400
 401#define EROFS_NAME_LEN      255
 402
 403/* check the EROFS on-disk layout strictly at compile time */
 404static inline void erofs_check_ondisk_layout_definitions(void)
 405{
 406        BUILD_BUG_ON(sizeof(struct erofs_super_block) != 128);
 407        BUILD_BUG_ON(sizeof(struct erofs_inode_compact) != 32);
 408        BUILD_BUG_ON(sizeof(struct erofs_inode_extended) != 64);
 409        BUILD_BUG_ON(sizeof(struct erofs_xattr_ibody_header) != 12);
 410        BUILD_BUG_ON(sizeof(struct erofs_xattr_entry) != 4);
 411        BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_info) != 4);
 412        BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) != 8);
 413        BUILD_BUG_ON(sizeof(struct z_erofs_map_header) != 8);
 414        BUILD_BUG_ON(sizeof(struct z_erofs_vle_decompressed_index) != 8);
 415        BUILD_BUG_ON(sizeof(struct erofs_dirent) != 12);
 416        /* keep in sync between 2 index structures for better extendibility */
 417        BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) !=
 418                     sizeof(struct z_erofs_vle_decompressed_index));
 419        BUILD_BUG_ON(sizeof(struct erofs_deviceslot) != 128);
 420
 421        BUILD_BUG_ON(BIT(Z_EROFS_VLE_DI_CLUSTER_TYPE_BITS) <
 422                     Z_EROFS_VLE_CLUSTER_TYPE_MAX - 1);
 423}
 424
 425#endif
 426