fs/ceph/ceph_fs.h

   1 /*
   2  * ceph_fs.h - Ceph constants and data types to share between kernel and
   3  * user space.
   4  *
   5  * Most types in this file are defined as little-endian, and are
   6  * primarily intended to describe data structures that pass over the
   7  * wire or that are stored on disk.
   8  *
   9  * LGPL2
  10  */
  11
  12 #ifndef _FS_CEPH_CEPH_FS_H
  13 #define _FS_CEPH_CEPH_FS_H
  14
  15 #include "msgr.h"
  16 #include "rados.h"
  17
  18 /*
  19  * Ceph release version
  20  */
  21 #define CEPH_VERSION_MAJOR 0
  22 #define CEPH_VERSION_MINOR 20
  23 #define CEPH_VERSION_PATCH 0
  24
  25 #define _CEPH_STRINGIFY(x) #x
  26 #define CEPH_STRINGIFY(x) _CEPH_STRINGIFY(x)
  27 #define CEPH_MAKE_VERSION(x, y, z) CEPH_STRINGIFY(x) "." CEPH_STRINGIFY(y) \
  28         "." CEPH_STRINGIFY(z)
  29 #define CEPH_VERSION CEPH_MAKE_VERSION(CEPH_VERSION_MAJOR, \
  30                                        CEPH_VERSION_MINOR, CEPH_VERSION_PATCH)
  31
  32 /*
  33  * subprotocol versions.  when specific messages types or high-level
  34  * protocols change, bump the affected components.  we keep rev
  35  * internal cluster protocols separately from the public,
  36  * client-facing protocol.
  37  */
  38 #define CEPH_OSD_PROTOCOL     8 /* cluster internal */
  39 #define CEPH_MDS_PROTOCOL    12 /* cluster internal */
  40 #define CEPH_MON_PROTOCOL     5 /* cluster internal */
  41 #define CEPH_OSDC_PROTOCOL   24 /* server/client */
  42 #define CEPH_MDSC_PROTOCOL   32 /* server/client */
  43 #define CEPH_MONC_PROTOCOL   15 /* server/client */
  44
  45
  46 #define CEPH_INO_ROOT  1
  47 #define CEPH_INO_CEPH  2        /* hidden .ceph dir */
  48
  49 /* arbitrary limit on max # of monitors (cluster of 3 is typical) */
  50 #define CEPH_MAX_MON   31
  51
  52
  53 /*
  54  * feature bits
  55  */
  56 #define CEPH_FEATURE_UID        1
  57 #define CEPH_FEATURE_NOSRCADDR  2
  58 #define CEPH_FEATURE_FLOCK      4
  59
  60 #define CEPH_FEATURE_SUPPORTED_MON  CEPH_FEATURE_UID|CEPH_FEATURE_NOSRCADDR
  61 #define CEPH_FEATURE_REQUIRED_MON   CEPH_FEATURE_UID
  62 #define CEPH_FEATURE_SUPPORTED_MDS  CEPH_FEATURE_UID|CEPH_FEATURE_NOSRCADDR|CEPH_FEATURE_FLOCK
  63 #define CEPH_FEATURE_REQUIRED_MDS   CEPH_FEATURE_UID
  64 #define CEPH_FEATURE_SUPPORTED_OSD  CEPH_FEATURE_UID|CEPH_FEATURE_NOSRCADDR
  65 #define CEPH_FEATURE_REQUIRED_OSD   CEPH_FEATURE_UID
  66 #define CEPH_FEATURE_SUPPORTED_CLIENT CEPH_FEATURE_NOSRCADDR
  67 #define CEPH_FEATURE_REQUIRED_CLIENT CEPH_FEATURE_NOSRCADDR
  68
  69
  70 /*
  71  * ceph_file_layout - describe data layout for a file/inode
  72  */
  73 struct ceph_file_layout {
  74         /* file -> object mapping */
  75         __le32 fl_stripe_unit;     /* stripe unit, in bytes.  must be multiple
  76                                       of page size. */
  77         __le32 fl_stripe_count;    /* over this many objects */
  78         __le32 fl_object_size;     /* until objects are this big, then move to
  79                                       new objects */
  80         __le32 fl_cas_hash;        /* 0 = none; 1 = sha256 */
  81
  82         /* pg -> disk layout */
  83         __le32 fl_object_stripe_unit;  /* for per-object parity, if any */
  84
  85         /* object -> pg layout */
  86         __le32 fl_pg_preferred; /* preferred primary for pg (-1 for none) */
  87         __le32 fl_pg_pool;      /* namespace, crush ruleset, rep level */
  88 } __attribute__ ((packed));
  89
  90 #define CEPH_MIN_STRIPE_UNIT 65536
  91
  92 int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
  93
  94
  95 /* crypto algorithms */
  96 #define CEPH_CRYPTO_NONE 0x0
  97 #define CEPH_CRYPTO_AES  0x1
  98
  99 /* security/authentication protocols */
 100 #define CEPH_AUTH_UNKNOWN       0x0
 101 #define CEPH_AUTH_NONE          0x1
 102 #define CEPH_AUTH_CEPHX         0x2
 103
 104 #define CEPH_AUTH_UID_DEFAULT ((__u64) -1)
 105
 106
 107 /*********************************************
 108  * message layer
 109  */
 110
 111 /*
 112  * message types
 113  */
 114
 115 /* misc */
 116 #define CEPH_MSG_SHUTDOWN               1
 117 #define CEPH_MSG_PING                   2
 118
 119 /* client <-> monitor */
 120 #define CEPH_MSG_MON_MAP                4
 121 #define CEPH_MSG_MON_GET_MAP            5
 122 #define CEPH_MSG_STATFS                 13
 123 #define CEPH_MSG_STATFS_REPLY           14
 124 #define CEPH_MSG_MON_SUBSCRIBE          15
 125 #define CEPH_MSG_MON_SUBSCRIBE_ACK      16
 126 #define CEPH_MSG_AUTH                   17
 127 #define CEPH_MSG_AUTH_REPLY             18
 128
 129 /* client <-> mds */
 130 #define CEPH_MSG_MDS_MAP                21
 131
 132 #define CEPH_MSG_CLIENT_SESSION         22
 133 #define CEPH_MSG_CLIENT_RECONNECT       23
 134
 135 #define CEPH_MSG_CLIENT_REQUEST         24
 136 #define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
 137 #define CEPH_MSG_CLIENT_REPLY           26
 138 #define CEPH_MSG_CLIENT_CAPS            0x310
 139 #define CEPH_MSG_CLIENT_LEASE           0x311
 140 #define CEPH_MSG_CLIENT_SNAP            0x312
 141 #define CEPH_MSG_CLIENT_CAPRELEASE      0x313
 142
 143 /* pool ops */
 144 #define CEPH_MSG_POOLOP_REPLY           48
 145 #define CEPH_MSG_POOLOP                 49
 146
 147
 148 /* osd */
 149 #define CEPH_MSG_OSD_MAP          41
 150 #define CEPH_MSG_OSD_OP           42
 151 #define CEPH_MSG_OSD_OPREPLY      43
 152
 153 /* pool operations */
 154 enum {
 155   POOL_OP_CREATE                        = 0x01,
 156   POOL_OP_DELETE                        = 0x02,
 157   POOL_OP_AUID_CHANGE                   = 0x03,
 158   POOL_OP_CREATE_SNAP                   = 0x11,
 159   POOL_OP_DELETE_SNAP                   = 0x12,
 160   POOL_OP_CREATE_UNMANAGED_SNAP         = 0x21,
 161   POOL_OP_DELETE_UNMANAGED_SNAP         = 0x22,
 162 };
 163
 164 struct ceph_mon_request_header {
 165         __le64 have_version;
 166         __le16 session_mon;
 167         __le64 session_mon_tid;
 168 } __attribute__ ((packed));
 169
 170 struct ceph_mon_statfs {
 171         struct ceph_mon_request_header monhdr;
 172         struct ceph_fsid fsid;
 173 } __attribute__ ((packed));
 174
 175 struct ceph_statfs {
 176         __le64 kb, kb_used, kb_avail;
 177         __le64 num_objects;
 178 } __attribute__ ((packed));
 179
 180 struct ceph_mon_statfs_reply {
 181         struct ceph_fsid fsid;
 182         __le64 version;
 183         struct ceph_statfs st;
 184 } __attribute__ ((packed));
 185
 186 const char *ceph_pool_op_name(int op);
 187
 188 struct ceph_mon_poolop {
 189         struct ceph_mon_request_header monhdr;
 190         struct ceph_fsid fsid;
 191         __le32 pool;
 192         __le32 op;
 193         __le64 auid;
 194         __le64 snapid;
 195         __le32 name_len;
 196 } __attribute__ ((packed));
 197
 198 struct ceph_mon_poolop_reply {
 199         struct ceph_mon_request_header monhdr;
 200         struct ceph_fsid fsid;
 201         __le32 reply_code;
 202         __le32 epoch;
 203         char has_data;
 204         char data[0];
 205 } __attribute__ ((packed));
 206
 207 struct ceph_mon_unmanaged_snap {
 208         __le64 snapid;
 209 } __attribute__ ((packed));
 210
 211 struct ceph_osd_getmap {
 212         struct ceph_mon_request_header monhdr;
 213         struct ceph_fsid fsid;
 214         __le32 start;
 215 } __attribute__ ((packed));
 216
 217 struct ceph_mds_getmap {
 218         struct ceph_mon_request_header monhdr;
 219         struct ceph_fsid fsid;
 220 } __attribute__ ((packed));
 221
 222 struct ceph_client_mount {
 223         struct ceph_mon_request_header monhdr;
 224 } __attribute__ ((packed));
 225
 226 struct ceph_mon_subscribe_item {
 227         __le64 have_version;    __le64 have;
 228         __u8 onetime;
 229 } __attribute__ ((packed));
 230
 231 struct ceph_mon_subscribe_ack {
 232         __le32 duration;         /* seconds */
 233         struct ceph_fsid fsid;
 234 } __attribute__ ((packed));
 235
 236 /*
 237  * mds states
 238  *   > 0 -> in
 239  *  <= 0 -> out
 240  */
 241 #define CEPH_MDS_STATE_DNE          0  /* down, does not exist. */
 242 #define CEPH_MDS_STATE_STOPPED     -1  /* down, once existed, but no subtrees.
 243                                           empty log. */
 244 #define CEPH_MDS_STATE_BOOT        -4  /* up, boot announcement. */
 245 #define CEPH_MDS_STATE_STANDBY     -5  /* up, idle.  waiting for assignment. */
 246 #define CEPH_MDS_STATE_CREATING    -6  /* up, creating MDS instance. */
 247 #define CEPH_MDS_STATE_STARTING    -7  /* up, starting previously stopped mds */
 248 #define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
 249
 250 #define CEPH_MDS_STATE_REPLAY       8  /* up, replaying journal. */
 251 #define CEPH_MDS_STATE_RESOLVE      9  /* up, disambiguating distributed
 252                                           operations (import, rename, etc.) */
 253 #define CEPH_MDS_STATE_RECONNECT    10 /* up, reconnect to clients */
 254 #define CEPH_MDS_STATE_REJOIN       11 /* up, rejoining distributed cache */
 255 #define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
 256 #define CEPH_MDS_STATE_ACTIVE       13 /* up, active */
 257 #define CEPH_MDS_STATE_STOPPING     14 /* up, but exporting metadata */
 258
 259 extern const char *ceph_mds_state_name(int s);
 260
 261
 262 /*
 263  * metadata lock types.
 264  *  - these are bitmasks.. we can compose them
 265  *  - they also define the lock ordering by the MDS
 266  *  - a few of these are internal to the mds
 267  */
 268 #define CEPH_LOCK_DVERSION    1
 269 #define CEPH_LOCK_DN          2
 270 #define CEPH_LOCK_ISNAP       16
 271 #define CEPH_LOCK_IVERSION    32    /* mds internal */
 272 #define CEPH_LOCK_IFILE       64
 273 #define CEPH_LOCK_IAUTH       128
 274 #define CEPH_LOCK_ILINK       256
 275 #define CEPH_LOCK_IDFT        512   /* dir frag tree */
 276 #define CEPH_LOCK_INEST       1024  /* mds internal */
 277 #define CEPH_LOCK_IXATTR      2048
 278 #define CEPH_LOCK_INO         8192  /* immutable inode bits; not a lock */
 279
 280 /* client_session ops */
 281 enum {
 282         CEPH_SESSION_REQUEST_OPEN,
 283         CEPH_SESSION_OPEN,
 284         CEPH_SESSION_REQUEST_CLOSE,
 285         CEPH_SESSION_CLOSE,
 286         CEPH_SESSION_REQUEST_RENEWCAPS,
 287         CEPH_SESSION_RENEWCAPS,
 288         CEPH_SESSION_STALE,
 289         CEPH_SESSION_RECALL_STATE,
 290 };
 291
 292 extern const char *ceph_session_op_name(int op);
 293
 294 struct ceph_mds_session_head {
 295         __le32 op;
 296         __le64 seq;
 297         struct ceph_timespec stamp;
 298         __le32 max_caps, max_leases;
 299 } __attribute__ ((packed));
 300
 301 /* client_request */
 302 /*
 303  * metadata ops.
 304  *  & 0x001000 -> write op
 305  *  & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
 306  &  & 0x100000 -> use weird ino/path trace
 307  */
 308 #define CEPH_MDS_OP_WRITE        0x001000
 309 enum {
 310         CEPH_MDS_OP_LOOKUP     = 0x00100,
 311         CEPH_MDS_OP_GETATTR    = 0x00101,
 312         CEPH_MDS_OP_LOOKUPHASH = 0x00102,
 313         CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
 314
 315         CEPH_MDS_OP_SETXATTR   = 0x01105,
 316         CEPH_MDS_OP_RMXATTR    = 0x01106,
 317         CEPH_MDS_OP_SETLAYOUT  = 0x01107,
 318         CEPH_MDS_OP_SETATTR    = 0x01108,
 319
 320         CEPH_MDS_OP_MKNOD      = 0x01201,
 321         CEPH_MDS_OP_LINK       = 0x01202,
 322         CEPH_MDS_OP_UNLINK     = 0x01203,
 323         CEPH_MDS_OP_RENAME     = 0x01204,
 324         CEPH_MDS_OP_MKDIR      = 0x01220,
 325         CEPH_MDS_OP_RMDIR      = 0x01221,
 326         CEPH_MDS_OP_SYMLINK    = 0x01222,
 327
 328         CEPH_MDS_OP_CREATE     = 0x01301,
 329         CEPH_MDS_OP_OPEN       = 0x00302,
 330         CEPH_MDS_OP_READDIR    = 0x00305,
 331
 332         CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
 333         CEPH_MDS_OP_MKSNAP     = 0x01400,
 334         CEPH_MDS_OP_RMSNAP     = 0x01401,
 335         CEPH_MDS_OP_LSSNAP     = 0x00402,
 336 };
 337
 338 extern const char *ceph_mds_op_name(int op);
 339
 340
 341 #define CEPH_SETATTR_MODE   1
 342 #define CEPH_SETATTR_UID    2
 343 #define CEPH_SETATTR_GID    4
 344 #define CEPH_SETATTR_MTIME  8
 345 #define CEPH_SETATTR_ATIME 16
 346 #define CEPH_SETATTR_SIZE  32
 347 #define CEPH_SETATTR_CTIME 64
 348
 349 union ceph_mds_request_args {
 350         struct {
 351                 __le32 mask;                 /* CEPH_CAP_* */
 352         } __attribute__ ((packed)) getattr;
 353         struct {
 354                 __le32 mode;
 355                 __le32 uid;
 356                 __le32 gid;
 357                 struct ceph_timespec mtime;
 358                 struct ceph_timespec atime;
 359                 __le64 size, old_size;       /* old_size needed by truncate */
 360                 __le32 mask;                 /* CEPH_SETATTR_* */
 361         } __attribute__ ((packed)) setattr;
 362         struct {
 363                 __le32 frag;                 /* which dir fragment */
 364                 __le32 max_entries;          /* how many dentries to grab */
 365                 __le32 max_bytes;
 366         } __attribute__ ((packed)) readdir;
 367         struct {
 368                 __le32 mode;
 369                 __le32 rdev;
 370         } __attribute__ ((packed)) mknod;
 371         struct {
 372                 __le32 mode;
 373         } __attribute__ ((packed)) mkdir;
 374         struct {
 375                 __le32 flags;
 376                 __le32 mode;
 377                 __le32 stripe_unit;          /* layout for newly created file */
 378                 __le32 stripe_count;         /* ... */
 379                 __le32 object_size;
 380                 __le32 file_replication;
 381                 __le32 preferred;
 382         } __attribute__ ((packed)) open;
 383         struct {
 384                 __le32 flags;
 385         } __attribute__ ((packed)) setxattr;
 386         struct {
 387                 struct ceph_file_layout layout;
 388         } __attribute__ ((packed)) setlayout;
 389 } __attribute__ ((packed));
 390
 391 #define CEPH_MDS_FLAG_REPLAY        1  /* this is a replayed op */
 392 #define CEPH_MDS_FLAG_WANT_DENTRY   2  /* want dentry in reply */
 393
 394 struct ceph_mds_request_head {
 395         __le64 oldest_client_tid;
 396         __le32 mdsmap_epoch;           /* on client */
 397         __le32 flags;                  /* CEPH_MDS_FLAG_* */
 398         __u8 num_retry, num_fwd;       /* count retry, fwd attempts */
 399         __le16 num_releases;           /* # include cap/lease release records */
 400         __le32 op;                     /* mds op code */
 401         __le32 caller_uid, caller_gid;
 402         __le64 ino;                    /* use this ino for openc, mkdir, mknod,
 403                                           etc. (if replaying) */
 404         union ceph_mds_request_args args;
 405 } __attribute__ ((packed));
 406
 407 /* cap/lease release record */
 408 struct ceph_mds_request_release {
 409         __le64 ino, cap_id;            /* ino and unique cap id */
 410         __le32 caps, wanted;           /* new issued, wanted */
 411         __le32 seq, issue_seq, mseq;
 412         __le32 dname_seq;              /* if releasing a dentry lease, a */
 413         __le32 dname_len;              /* string follows. */
 414 } __attribute__ ((packed));
 415
 416 /* client reply */
 417 struct ceph_mds_reply_head {
 418         __le32 op;
 419         __le32 result;
 420         __le32 mdsmap_epoch;
 421         __u8 safe;                     /* true if committed to disk */
 422         __u8 is_dentry, is_target;     /* true if dentry, target inode records
 423                                           are included with reply */
 424 } __attribute__ ((packed));
 425
 426 /* one for each node split */
 427 struct ceph_frag_tree_split {
 428         __le32 frag;                   /* this frag splits... */
 429         __le32 by;                     /* ...by this many bits */
 430 } __attribute__ ((packed));
 431
 432 struct ceph_frag_tree_head {
 433         __le32 nsplits;                /* num ceph_frag_tree_split records */
 434         struct ceph_frag_tree_split splits[];
 435 } __attribute__ ((packed));
 436
 437 /* capability issue, for bundling with mds reply */
 438 struct ceph_mds_reply_cap {
 439         __le32 caps, wanted;           /* caps issued, wanted */
 440         __le64 cap_id;
 441         __le32 seq, mseq;
 442         __le64 realm;                  /* snap realm */
 443         __u8 flags;                    /* CEPH_CAP_FLAG_* */
 444 } __attribute__ ((packed));
 445
 446 #define CEPH_CAP_FLAG_AUTH  1          /* cap is issued by auth mds */
 447
 448 /* inode record, for bundling with mds reply */
 449 struct ceph_mds_reply_inode {
 450         __le64 ino;
 451         __le64 snapid;
 452         __le32 rdev;
 453         __le64 version;                /* inode version */
 454         __le64 xattr_version;          /* version for xattr blob */
 455         struct ceph_mds_reply_cap cap; /* caps issued for this inode */
 456         struct ceph_file_layout layout;
 457         struct ceph_timespec ctime, mtime, atime;
 458         __le32 time_warp_seq;
 459         __le64 size, max_size, truncate_size;
 460         __le32 truncate_seq;
 461         __le32 mode, uid, gid;
 462         __le32 nlink;
 463         __le64 files, subdirs, rbytes, rfiles, rsubdirs;  /* dir stats */
 464         struct ceph_timespec rctime;
 465         struct ceph_frag_tree_head fragtree;  /* (must be at end of struct) */
 466 } __attribute__ ((packed));
 467 /* followed by frag array, then symlink string, then xattr blob */
 468
 469 /* reply_lease follows dname, and reply_inode */
 470 struct ceph_mds_reply_lease {
 471         __le16 mask;            /* lease type(s) */
 472         __le32 duration_ms;     /* lease duration */
 473         __le32 seq;
 474 } __attribute__ ((packed));
 475
 476 struct ceph_mds_reply_dirfrag {
 477         __le32 frag;            /* fragment */
 478         __le32 auth;            /* auth mds, if this is a delegation point */
 479         __le32 ndist;           /* number of mds' this is replicated on */
 480         __le32 dist[];
 481 } __attribute__ ((packed));
 482
 483 /* file access modes */
 484 #define CEPH_FILE_MODE_PIN        0
 485 #define CEPH_FILE_MODE_RD         1
 486 #define CEPH_FILE_MODE_WR         2
 487 #define CEPH_FILE_MODE_RDWR       3  /* RD | WR */
 488 #define CEPH_FILE_MODE_LAZY       4  /* lazy io */
 489 #define CEPH_FILE_MODE_NUM        8  /* bc these are bit fields.. mostly */
 490
 491 int ceph_flags_to_mode(int flags);
 492
 493
 494 /* capability bits */
 495 #define CEPH_CAP_PIN         1  /* no specific capabilities beyond the pin */
 496
 497 /* generic cap bits */
 498 #define CEPH_CAP_GSHARED     1  /* client can reads */
 499 #define CEPH_CAP_GEXCL       2  /* client can read and update */
 500 #define CEPH_CAP_GCACHE      4  /* (file) client can cache reads */
 501 #define CEPH_CAP_GRD         8  /* (file) client can read */
 502 #define CEPH_CAP_GWR        16  /* (file) client can write */
 503 #define CEPH_CAP_GBUFFER    32  /* (file) client can buffer writes */
 504 #define CEPH_CAP_GWREXTEND  64  /* (file) client can extend EOF */
 505 #define CEPH_CAP_GLAZYIO   128  /* (file) client can perform lazy io */
 506
 507 /* per-lock shift */
 508 #define CEPH_CAP_SAUTH      2
 509 #define CEPH_CAP_SLINK      4
 510 #define CEPH_CAP_SXATTR     6
 511 #define CEPH_CAP_SFILE      8   /* goes at the end (uses >2 cap bits) */
 512
 513 #define CEPH_CAP_BITS       16
 514
 515 /* composed values */
 516 #define CEPH_CAP_AUTH_SHARED  (CEPH_CAP_GSHARED  << CEPH_CAP_SAUTH)
 517 #define CEPH_CAP_AUTH_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SAUTH)
 518 #define CEPH_CAP_LINK_SHARED  (CEPH_CAP_GSHARED  << CEPH_CAP_SLINK)
 519 #define CEPH_CAP_LINK_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SLINK)
 520 #define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED  << CEPH_CAP_SXATTR)
 521 #define CEPH_CAP_XATTR_EXCL    (CEPH_CAP_GEXCL     << CEPH_CAP_SXATTR)
 522 #define CEPH_CAP_FILE(x)    (x << CEPH_CAP_SFILE)
 523 #define CEPH_CAP_FILE_SHARED   (CEPH_CAP_GSHARED   << CEPH_CAP_SFILE)
 524 #define CEPH_CAP_FILE_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SFILE)
 525 #define CEPH_CAP_FILE_CACHE    (CEPH_CAP_GCACHE    << CEPH_CAP_SFILE)
 526 #define CEPH_CAP_FILE_RD       (CEPH_CAP_GRD       << CEPH_CAP_SFILE)
 527 #define CEPH_CAP_FILE_WR       (CEPH_CAP_GWR       << CEPH_CAP_SFILE)
 528 #define CEPH_CAP_FILE_BUFFER   (CEPH_CAP_GBUFFER   << CEPH_CAP_SFILE)
 529 #define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
 530 #define CEPH_CAP_FILE_LAZYIO   (CEPH_CAP_GLAZYIO   << CEPH_CAP_SFILE)
 531
 532 /* cap masks (for getattr) */
 533 #define CEPH_STAT_CAP_INODE    CEPH_CAP_PIN
 534 #define CEPH_STAT_CAP_TYPE     CEPH_CAP_PIN  /* mode >> 12 */
 535 #define CEPH_STAT_CAP_SYMLINK  CEPH_CAP_PIN
 536 #define CEPH_STAT_CAP_UID      CEPH_CAP_AUTH_SHARED
 537 #define CEPH_STAT_CAP_GID      CEPH_CAP_AUTH_SHARED
 538 #define CEPH_STAT_CAP_MODE     CEPH_CAP_AUTH_SHARED
 539 #define CEPH_STAT_CAP_NLINK    CEPH_CAP_LINK_SHARED
 540 #define CEPH_STAT_CAP_LAYOUT   CEPH_CAP_FILE_SHARED
 541 #define CEPH_STAT_CAP_MTIME    CEPH_CAP_FILE_SHARED
 542 #define CEPH_STAT_CAP_SIZE     CEPH_CAP_FILE_SHARED
 543 #define CEPH_STAT_CAP_ATIME    CEPH_CAP_FILE_SHARED  /* fixme */
 544 #define CEPH_STAT_CAP_XATTR    CEPH_CAP_XATTR_SHARED
 545 #define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN |                 \
 546                                  CEPH_CAP_AUTH_SHARED | \
 547                                  CEPH_CAP_LINK_SHARED | \
 548                                  CEPH_CAP_FILE_SHARED | \
 549                                  CEPH_CAP_XATTR_SHARED)
 550
 551 #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED |                     \
 552                               CEPH_CAP_LINK_SHARED |                    \
 553                               CEPH_CAP_XATTR_SHARED |                   \
 554                               CEPH_CAP_FILE_SHARED)
 555 #define CEPH_CAP_ANY_RD   (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD |     \
 556                            CEPH_CAP_FILE_CACHE)
 557
 558 #define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL |         \
 559                            CEPH_CAP_LINK_EXCL |         \
 560                            CEPH_CAP_XATTR_EXCL |        \
 561                            CEPH_CAP_FILE_EXCL)
 562 #define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \
 563                               CEPH_CAP_FILE_EXCL)
 564 #define CEPH_CAP_ANY_WR   (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
 565 #define CEPH_CAP_ANY      (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
 566                            CEPH_CAP_ANY_FILE_WR | CEPH_CAP_PIN)
 567
 568 #define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
 569                         CEPH_LOCK_IXATTR)
 570
 571 int ceph_caps_for_mode(int mode);
 572
 573 enum {
 574         CEPH_CAP_OP_GRANT,         /* mds->client grant */
 575         CEPH_CAP_OP_REVOKE,        /* mds->client revoke */
 576         CEPH_CAP_OP_TRUNC,         /* mds->client trunc notify */
 577         CEPH_CAP_OP_EXPORT,        /* mds has exported the cap */
 578         CEPH_CAP_OP_IMPORT,        /* mds has imported the cap */
 579         CEPH_CAP_OP_UPDATE,        /* client->mds update */
 580         CEPH_CAP_OP_DROP,          /* client->mds drop cap bits */
 581         CEPH_CAP_OP_FLUSH,         /* client->mds cap writeback */
 582         CEPH_CAP_OP_FLUSH_ACK,     /* mds->client flushed */
 583         CEPH_CAP_OP_FLUSHSNAP,     /* client->mds flush snapped metadata */
 584         CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
 585         CEPH_CAP_OP_RELEASE,       /* client->mds release (clean) cap */
 586         CEPH_CAP_OP_RENEW,         /* client->mds renewal request */
 587 };
 588
 589 extern const char *ceph_cap_op_name(int op);
 590
 591 /*
 592  * caps message, used for capability callbacks, acks, requests, etc.
 593  */
 594 struct ceph_mds_caps {
 595         __le32 op;                  /* CEPH_CAP_OP_* */
 596         __le64 ino, realm;
 597         __le64 cap_id;
 598         __le32 seq, issue_seq;
 599         __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
 600         __le32 migrate_seq;
 601         __le64 snap_follows;
 602         __le32 snap_trace_len;
 603
 604         /* authlock */
 605         __le32 uid, gid, mode;
 606
 607         /* linklock */
 608         __le32 nlink;
 609
 610         /* xattrlock */
 611         __le32 xattr_len;
 612         __le64 xattr_version;
 613
 614         /* filelock */
 615         __le64 size, max_size, truncate_size;
 616         __le32 truncate_seq;
 617         struct ceph_timespec mtime, atime, ctime;
 618         struct ceph_file_layout layout;
 619         __le32 time_warp_seq;
 620 } __attribute__ ((packed));
 621
 622 /* cap release msg head */
 623 struct ceph_mds_cap_release {
 624         __le32 num;                /* number of cap_items that follow */
 625 } __attribute__ ((packed));
 626
 627 struct ceph_mds_cap_item {
 628         __le64 ino;
 629         __le64 cap_id;
 630         __le32 migrate_seq, seq;
 631 } __attribute__ ((packed));
 632
 633 #define CEPH_MDS_LEASE_REVOKE           1  /*    mds  -> client */
 634 #define CEPH_MDS_LEASE_RELEASE          2  /* client  -> mds    */
 635 #define CEPH_MDS_LEASE_RENEW            3  /* client <-> mds    */
 636 #define CEPH_MDS_LEASE_REVOKE_ACK       4  /* client  -> mds    */
 637
 638 extern const char *ceph_lease_op_name(int o);
 639
 640 /* lease msg header */
 641 struct ceph_mds_lease {
 642         __u8 action;            /* CEPH_MDS_LEASE_* */
 643         __le16 mask;            /* which lease */
 644         __le64 ino;
 645         __le64 first, last;     /* snap range */
 646         __le32 seq;
 647         __le32 duration_ms;     /* duration of renewal */
 648 } __attribute__ ((packed));
 649 /* followed by a __le32+string for dname */
 650
 651 /* client reconnect */
 652 struct ceph_mds_cap_reconnect {
 653         __le64 cap_id;
 654         __le32 wanted;
 655         __le32 issued;
 656         __le64 size;
 657         struct ceph_timespec mtime, atime;
 658         __le64 snaprealm;
 659         __le64 pathbase;        /* base ino for our path to this ino */
 660 } __attribute__ ((packed));
 661 /* followed by encoded string */
 662
 663 struct ceph_mds_snaprealm_reconnect {
 664         __le64 ino;     /* snap realm base */
 665         __le64 seq;     /* snap seq for this snap realm */
 666         __le64 parent;  /* parent realm */
 667 } __attribute__ ((packed));
 668
 669 /*
 670  * snaps
 671  */
 672 enum {
 673         CEPH_SNAP_OP_UPDATE,  /* CREATE or DESTROY */
 674         CEPH_SNAP_OP_CREATE,
 675         CEPH_SNAP_OP_DESTROY,
 676         CEPH_SNAP_OP_SPLIT,
 677 };
 678
 679 extern const char *ceph_snap_op_name(int o);
 680
 681 /* snap msg header */
 682 struct ceph_mds_snap_head {
 683         __le32 op;                /* CEPH_SNAP_OP_* */
 684         __le64 split;             /* ino to split off, if any */
 685         __le32 num_split_inos;    /* # inos belonging to new child realm */
 686         __le32 num_split_realms;  /* # child realms udner new child realm */
 687         __le32 trace_len;         /* size of snap trace blob */
 688 } __attribute__ ((packed));
 689 /* followed by split ino list, then split realms, then the trace blob */
 690
 691 /*
 692  * encode info about a snaprealm, as viewed by a client
 693  */
 694 struct ceph_mds_snap_realm {
 695         __le64 ino;           /* ino */
 696         __le64 created;       /* snap: when created */
 697         __le64 parent;        /* ino: parent realm */
 698         __le64 parent_since;  /* snap: same parent since */
 699         __le64 seq;           /* snap: version */
 700         __le32 num_snaps;
 701         __le32 num_prior_parent_snaps;
 702 } __attribute__ ((packed));
 703 /* followed by my snap list, then prior parent snap list */
 704
 705 #endif