1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * ceph_fs.h - Ceph constants and data types to share between kernel and 4 * user space. 5 * 6 * Most types in this file are defined as little-endian, and are 7 * primarily intended to describe data structures that pass over the 8 * wire or that are stored on disk. 9 * 10 * LGPL2 11 */ 12 13 #ifndef CEPH_FS_H 14 #define CEPH_FS_H 15 16 #include <linux/ceph/msgr.h> 17 #include <linux/ceph/rados.h> 18 19 /* 20 * subprotocol versions. when specific messages types or high-level 21 * protocols change, bump the affected components. we keep rev 22 * internal cluster protocols separately from the public, 23 * client-facing protocol. 24 */ 25 #define CEPH_OSDC_PROTOCOL 24 /* server/client */ 26 #define CEPH_MDSC_PROTOCOL 32 /* server/client */ 27 #define CEPH_MONC_PROTOCOL 15 /* server/client */ 28 29 30 #define CEPH_INO_ROOT 1 31 #define CEPH_INO_CEPH 2 /* hidden .ceph dir */ 32 #define CEPH_INO_DOTDOT 3 /* used by ceph fuse for parent (..) */ 33 34 /* arbitrary limit on max # of monitors (cluster of 3 is typical) */ 35 #define CEPH_MAX_MON 31 36 37 /* 38 * legacy ceph_file_layoute 39 */ 40 struct ceph_file_layout_legacy { 41 /* file -> object mapping */ 42 __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple 43 of page size. */ 44 __le32 fl_stripe_count; /* over this many objects */ 45 __le32 fl_object_size; /* until objects are this big, then move to 46 new objects */ 47 __le32 fl_cas_hash; /* UNUSED. 0 = none; 1 = sha256 */ 48 49 /* pg -> disk layout */ 50 __le32 fl_object_stripe_unit; /* UNUSED. for per-object parity, if any */ 51 52 /* object -> pg layout */ 53 __le32 fl_unused; /* unused; used to be preferred primary for pg (-1 for none) */ 54 __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */ 55 } __attribute__ ((packed)); 56 57 struct ceph_string; 58 /* 59 * ceph_file_layout - describe data layout for a file/inode 60 */ 61 struct ceph_file_layout { 62 /* file -> object mapping */ 63 u32 stripe_unit; /* stripe unit, in bytes */ 64 u32 stripe_count; /* over this many objects */ 65 u32 object_size; /* until objects are this big */ 66 s64 pool_id; /* rados pool id */ 67 struct ceph_string __rcu *pool_ns; /* rados pool namespace */ 68 }; 69 70 extern int ceph_file_layout_is_valid(const struct ceph_file_layout *layout); 71 extern void ceph_file_layout_from_legacy(struct ceph_file_layout *fl, 72 struct ceph_file_layout_legacy *legacy); 73 extern void ceph_file_layout_to_legacy(struct ceph_file_layout *fl, 74 struct ceph_file_layout_legacy *legacy); 75 76 #define CEPH_MIN_STRIPE_UNIT 65536 77 78 struct ceph_dir_layout { 79 __u8 dl_dir_hash; /* see ceph_hash.h for ids */ 80 __u8 dl_unused1; 81 __u16 dl_unused2; 82 __u32 dl_unused3; 83 } __attribute__ ((packed)); 84 85 /* crypto algorithms */ 86 #define CEPH_CRYPTO_NONE 0x0 87 #define CEPH_CRYPTO_AES 0x1 88 89 #define CEPH_AES_IV "cephsageyudagreg" 90 91 /* security/authentication protocols */ 92 #define CEPH_AUTH_UNKNOWN 0x0 93 #define CEPH_AUTH_NONE 0x1 94 #define CEPH_AUTH_CEPHX 0x2 95 96 #define CEPH_AUTH_MODE_NONE 0 97 #define CEPH_AUTH_MODE_AUTHORIZER 1 98 #define CEPH_AUTH_MODE_MON 10 99 100 /* msgr2 protocol modes */ 101 #define CEPH_CON_MODE_UNKNOWN 0x0 102 #define CEPH_CON_MODE_CRC 0x1 103 #define CEPH_CON_MODE_SECURE 0x2 104 105 #define CEPH_AUTH_UID_DEFAULT ((__u64) -1) 106 107 const char *ceph_auth_proto_name(int proto); 108 const char *ceph_con_mode_name(int mode); 109 110 /********************************************* 111 * message layer 112 */ 113 114 /* 115 * message types 116 */ 117 118 /* misc */ 119 #define CEPH_MSG_SHUTDOWN 1 120 #define CEPH_MSG_PING 2 121 122 /* client <-> monitor */ 123 #define CEPH_MSG_MON_MAP 4 124 #define CEPH_MSG_MON_GET_MAP 5 125 #define CEPH_MSG_STATFS 13 126 #define CEPH_MSG_STATFS_REPLY 14 127 #define CEPH_MSG_MON_SUBSCRIBE 15 128 #define CEPH_MSG_MON_SUBSCRIBE_ACK 16 129 #define CEPH_MSG_AUTH 17 130 #define CEPH_MSG_AUTH_REPLY 18 131 #define CEPH_MSG_MON_GET_VERSION 19 132 #define CEPH_MSG_MON_GET_VERSION_REPLY 20 133 134 /* client <-> mds */ 135 #define CEPH_MSG_MDS_MAP 21 136 #define CEPH_MSG_FS_MAP_USER 103 137 138 #define CEPH_MSG_CLIENT_SESSION 22 139 #define CEPH_MSG_CLIENT_RECONNECT 23 140 141 #define CEPH_MSG_CLIENT_REQUEST 24 142 #define CEPH_MSG_CLIENT_REQUEST_FORWARD 25 143 #define CEPH_MSG_CLIENT_REPLY 26 144 #define CEPH_MSG_CLIENT_METRICS 29 145 #define CEPH_MSG_CLIENT_CAPS 0x310 146 #define CEPH_MSG_CLIENT_LEASE 0x311 147 #define CEPH_MSG_CLIENT_SNAP 0x312 148 #define CEPH_MSG_CLIENT_CAPRELEASE 0x313 149 #define CEPH_MSG_CLIENT_QUOTA 0x314 150 151 /* pool ops */ 152 #define CEPH_MSG_POOLOP_REPLY 48 153 #define CEPH_MSG_POOLOP 49 154 155 /* mon commands */ 156 #define CEPH_MSG_MON_COMMAND 50 157 #define CEPH_MSG_MON_COMMAND_ACK 51 158 159 /* osd */ 160 #define CEPH_MSG_OSD_MAP 41 161 #define CEPH_MSG_OSD_OP 42 162 #define CEPH_MSG_OSD_OPREPLY 43 163 #define CEPH_MSG_WATCH_NOTIFY 44 164 #define CEPH_MSG_OSD_BACKOFF 61 165 166 167 /* watch-notify operations */ 168 enum { 169 CEPH_WATCH_EVENT_NOTIFY = 1, /* notifying watcher */ 170 CEPH_WATCH_EVENT_NOTIFY_COMPLETE = 2, /* notifier notified when done */ 171 CEPH_WATCH_EVENT_DISCONNECT = 3, /* we were disconnected */ 172 }; 173 174 175 struct ceph_mon_request_header { 176 __le64 have_version; 177 __le16 session_mon; 178 __le64 session_mon_tid; 179 } __attribute__ ((packed)); 180 181 struct ceph_mon_statfs { 182 struct ceph_mon_request_header monhdr; 183 struct ceph_fsid fsid; 184 __u8 contains_data_pool; 185 __le64 data_pool; 186 } __attribute__ ((packed)); 187 188 struct ceph_statfs { 189 __le64 kb, kb_used, kb_avail; 190 __le64 num_objects; 191 } __attribute__ ((packed)); 192 193 struct ceph_mon_statfs_reply { 194 struct ceph_fsid fsid; 195 __le64 version; 196 struct ceph_statfs st; 197 } __attribute__ ((packed)); 198 199 struct ceph_mon_command { 200 struct ceph_mon_request_header monhdr; 201 struct ceph_fsid fsid; 202 __le32 num_strs; /* always 1 */ 203 __le32 str_len; 204 char str[]; 205 } __attribute__ ((packed)); 206 207 struct ceph_osd_getmap { 208 struct ceph_mon_request_header monhdr; 209 struct ceph_fsid fsid; 210 __le32 start; 211 } __attribute__ ((packed)); 212 213 struct ceph_mds_getmap { 214 struct ceph_mon_request_header monhdr; 215 struct ceph_fsid fsid; 216 } __attribute__ ((packed)); 217 218 struct ceph_client_mount { 219 struct ceph_mon_request_header monhdr; 220 } __attribute__ ((packed)); 221 222 #define CEPH_SUBSCRIBE_ONETIME 1 /* i want only 1 update after have */ 223 224 struct ceph_mon_subscribe_item { 225 __le64 start; 226 __u8 flags; 227 } __attribute__ ((packed)); 228 229 struct ceph_mon_subscribe_ack { 230 __le32 duration; /* seconds */ 231 struct ceph_fsid fsid; 232 } __attribute__ ((packed)); 233 234 #define CEPH_FS_CLUSTER_ID_NONE -1 235 236 /* 237 * mdsmap flags 238 */ 239 #define CEPH_MDSMAP_DOWN (1<<0) /* cluster deliberately down */ 240 241 /* 242 * mds states 243 * > 0 -> in 244 * <= 0 -> out 245 */ 246 #define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */ 247 #define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees. 248 empty log. */ 249 #define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */ 250 #define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */ 251 #define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */ 252 #define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */ 253 #define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */ 254 #define CEPH_MDS_STATE_REPLAYONCE -9 /* up, replaying an active node's journal */ 255 256 #define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */ 257 #define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed 258 operations (import, rename, etc.) */ 259 #define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */ 260 #define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */ 261 #define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */ 262 #define CEPH_MDS_STATE_ACTIVE 13 /* up, active */ 263 #define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */ 264 265 extern const char *ceph_mds_state_name(int s); 266 267 268 /* 269 * metadata lock types. 270 * - these are bitmasks.. we can compose them 271 * - they also define the lock ordering by the MDS 272 * - a few of these are internal to the mds 273 */ 274 #define CEPH_LOCK_DVERSION 1 275 #define CEPH_LOCK_DN 2 276 #define CEPH_LOCK_ISNAP 16 277 #define CEPH_LOCK_IVERSION 32 /* mds internal */ 278 #define CEPH_LOCK_IFILE 64 279 #define CEPH_LOCK_IAUTH 128 280 #define CEPH_LOCK_ILINK 256 281 #define CEPH_LOCK_IDFT 512 /* dir frag tree */ 282 #define CEPH_LOCK_INEST 1024 /* mds internal */ 283 #define CEPH_LOCK_IXATTR 2048 284 #define CEPH_LOCK_IFLOCK 4096 /* advisory file locks */ 285 #define CEPH_LOCK_INO 8192 /* immutable inode bits; not a lock */ 286 #define CEPH_LOCK_IPOLICY 16384 /* policy lock on dirs. MDS internal */ 287 288 /* client_session ops */ 289 enum { 290 CEPH_SESSION_REQUEST_OPEN, 291 CEPH_SESSION_OPEN, 292 CEPH_SESSION_REQUEST_CLOSE, 293 CEPH_SESSION_CLOSE, 294 CEPH_SESSION_REQUEST_RENEWCAPS, 295 CEPH_SESSION_RENEWCAPS, 296 CEPH_SESSION_STALE, 297 CEPH_SESSION_RECALL_STATE, 298 CEPH_SESSION_FLUSHMSG, 299 CEPH_SESSION_FLUSHMSG_ACK, 300 CEPH_SESSION_FORCE_RO, 301 CEPH_SESSION_REJECT, 302 CEPH_SESSION_REQUEST_FLUSH_MDLOG, 303 }; 304 305 extern const char *ceph_session_op_name(int op); 306 307 struct ceph_mds_session_head { 308 __le32 op; 309 __le64 seq; 310 struct ceph_timespec stamp; 311 __le32 max_caps, max_leases; 312 } __attribute__ ((packed)); 313 314 /* client_request */ 315 /* 316 * metadata ops. 317 * & 0x001000 -> write op 318 * & 0x010000 -> follow symlink (e.g. stat(), not lstat()). 319 & & 0x100000 -> use weird ino/path trace 320 */ 321 #define CEPH_MDS_OP_WRITE 0x001000 322 enum { 323 CEPH_MDS_OP_LOOKUP = 0x00100, 324 CEPH_MDS_OP_GETATTR = 0x00101, 325 CEPH_MDS_OP_LOOKUPHASH = 0x00102, 326 CEPH_MDS_OP_LOOKUPPARENT = 0x00103, 327 CEPH_MDS_OP_LOOKUPINO = 0x00104, 328 CEPH_MDS_OP_LOOKUPNAME = 0x00105, 329 330 CEPH_MDS_OP_SETXATTR = 0x01105, 331 CEPH_MDS_OP_RMXATTR = 0x01106, 332 CEPH_MDS_OP_SETLAYOUT = 0x01107, 333 CEPH_MDS_OP_SETATTR = 0x01108, 334 CEPH_MDS_OP_SETFILELOCK= 0x01109, 335 CEPH_MDS_OP_GETFILELOCK= 0x00110, 336 CEPH_MDS_OP_SETDIRLAYOUT=0x0110a, 337 338 CEPH_MDS_OP_MKNOD = 0x01201, 339 CEPH_MDS_OP_LINK = 0x01202, 340 CEPH_MDS_OP_UNLINK = 0x01203, 341 CEPH_MDS_OP_RENAME = 0x01204, 342 CEPH_MDS_OP_MKDIR = 0x01220, 343 CEPH_MDS_OP_RMDIR = 0x01221, 344 CEPH_MDS_OP_SYMLINK = 0x01222, 345 346 CEPH_MDS_OP_CREATE = 0x01301, 347 CEPH_MDS_OP_OPEN = 0x00302, 348 CEPH_MDS_OP_READDIR = 0x00305, 349 350 CEPH_MDS_OP_LOOKUPSNAP = 0x00400, 351 CEPH_MDS_OP_MKSNAP = 0x01400, 352 CEPH_MDS_OP_RMSNAP = 0x01401, 353 CEPH_MDS_OP_LSSNAP = 0x00402, 354 CEPH_MDS_OP_RENAMESNAP = 0x01403, 355 }; 356 357 extern const char *ceph_mds_op_name(int op); 358 359 360 #define CEPH_SETATTR_MODE 1 361 #define CEPH_SETATTR_UID 2 362 #define CEPH_SETATTR_GID 4 363 #define CEPH_SETATTR_MTIME 8 364 #define CEPH_SETATTR_ATIME 16 365 #define CEPH_SETATTR_SIZE 32 366 #define CEPH_SETATTR_CTIME 64 367 368 /* 369 * Ceph setxattr request flags. 370 */ 371 #define CEPH_XATTR_CREATE (1 << 0) 372 #define CEPH_XATTR_REPLACE (1 << 1) 373 #define CEPH_XATTR_REMOVE (1 << 31) 374 375 /* 376 * readdir request flags; 377 */ 378 #define CEPH_READDIR_REPLY_BITFLAGS (1<<0) 379 380 /* 381 * readdir reply flags. 382 */ 383 #define CEPH_READDIR_FRAG_END (1<<0) 384 #define CEPH_READDIR_FRAG_COMPLETE (1<<8) 385 #define CEPH_READDIR_HASH_ORDER (1<<9) 386 #define CEPH_READDIR_OFFSET_HASH (1<<10) 387 388 /* 389 * open request flags 390 */ 391 #define CEPH_O_RDONLY 00000000 392 #define CEPH_O_WRONLY 00000001 393 #define CEPH_O_RDWR 00000002 394 #define CEPH_O_CREAT 00000100 395 #define CEPH_O_EXCL 00000200 396 #define CEPH_O_TRUNC 00001000 397 #define CEPH_O_DIRECTORY 00200000 398 #define CEPH_O_NOFOLLOW 00400000 399 400 union ceph_mds_request_args { 401 struct { 402 __le32 mask; /* CEPH_CAP_* */ 403 } __attribute__ ((packed)) getattr; 404 struct { 405 __le32 mode; 406 __le32 uid; 407 __le32 gid; 408 struct ceph_timespec mtime; 409 struct ceph_timespec atime; 410 __le64 size, old_size; /* old_size needed by truncate */ 411 __le32 mask; /* CEPH_SETATTR_* */ 412 } __attribute__ ((packed)) setattr; 413 struct { 414 __le32 frag; /* which dir fragment */ 415 __le32 max_entries; /* how many dentries to grab */ 416 __le32 max_bytes; 417 __le16 flags; 418 __le32 offset_hash; 419 } __attribute__ ((packed)) readdir; 420 struct { 421 __le32 mode; 422 __le32 rdev; 423 } __attribute__ ((packed)) mknod; 424 struct { 425 __le32 mode; 426 } __attribute__ ((packed)) mkdir; 427 struct { 428 __le32 flags; 429 __le32 mode; 430 __le32 stripe_unit; /* layout for newly created file */ 431 __le32 stripe_count; /* ... */ 432 __le32 object_size; 433 __le32 file_replication; 434 __le32 mask; /* CEPH_CAP_* */ 435 __le32 old_size; 436 } __attribute__ ((packed)) open; 437 struct { 438 __le32 flags; 439 __le32 osdmap_epoch; /* used for setting file/dir layouts */ 440 } __attribute__ ((packed)) setxattr; 441 struct { 442 struct ceph_file_layout_legacy layout; 443 } __attribute__ ((packed)) setlayout; 444 struct { 445 __u8 rule; /* currently fcntl or flock */ 446 __u8 type; /* shared, exclusive, remove*/ 447 __le64 owner; /* owner of the lock */ 448 __le64 pid; /* process id requesting the lock */ 449 __le64 start; /* initial location to lock */ 450 __le64 length; /* num bytes to lock from start */ 451 __u8 wait; /* will caller wait for lock to become available? */ 452 } __attribute__ ((packed)) filelock_change; 453 struct { 454 __le32 mask; /* CEPH_CAP_* */ 455 __le64 snapid; 456 __le64 parent; 457 __le32 hash; 458 } __attribute__ ((packed)) lookupino; 459 } __attribute__ ((packed)); 460 461 union ceph_mds_request_args_ext { 462 union ceph_mds_request_args old; 463 struct { 464 __le32 mode; 465 __le32 uid; 466 __le32 gid; 467 struct ceph_timespec mtime; 468 struct ceph_timespec atime; 469 __le64 size, old_size; /* old_size needed by truncate */ 470 __le32 mask; /* CEPH_SETATTR_* */ 471 struct ceph_timespec btime; 472 } __attribute__ ((packed)) setattr_ext; 473 }; 474 475 #define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */ 476 #define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */ 477 #define CEPH_MDS_FLAG_ASYNC 4 /* request is asynchronous */ 478 479 struct ceph_mds_request_head_old { 480 __le64 oldest_client_tid; 481 __le32 mdsmap_epoch; /* on client */ 482 __le32 flags; /* CEPH_MDS_FLAG_* */ 483 __u8 num_retry, num_fwd; /* count retry, fwd attempts */ 484 __le16 num_releases; /* # include cap/lease release records */ 485 __le32 op; /* mds op code */ 486 __le32 caller_uid, caller_gid; 487 __le64 ino; /* use this ino for openc, mkdir, mknod, 488 etc. (if replaying) */ 489 union ceph_mds_request_args args; 490 } __attribute__ ((packed)); 491 492 #define CEPH_MDS_REQUEST_HEAD_VERSION 1 493 494 struct ceph_mds_request_head { 495 __le16 version; /* struct version */ 496 __le64 oldest_client_tid; 497 __le32 mdsmap_epoch; /* on client */ 498 __le32 flags; /* CEPH_MDS_FLAG_* */ 499 __u8 num_retry, num_fwd; /* count retry, fwd attempts */ 500 __le16 num_releases; /* # include cap/lease release records */ 501 __le32 op; /* mds op code */ 502 __le32 caller_uid, caller_gid; 503 __le64 ino; /* use this ino for openc, mkdir, mknod, 504 etc. (if replaying) */ 505 union ceph_mds_request_args_ext args; 506 } __attribute__ ((packed)); 507 508 /* cap/lease release record */ 509 struct ceph_mds_request_release { 510 __le64 ino, cap_id; /* ino and unique cap id */ 511 __le32 caps, wanted; /* new issued, wanted */ 512 __le32 seq, issue_seq, mseq; 513 __le32 dname_seq; /* if releasing a dentry lease, a */ 514 __le32 dname_len; /* string follows. */ 515 } __attribute__ ((packed)); 516 517 /* client reply */ 518 struct ceph_mds_reply_head { 519 __le32 op; 520 __le32 result; 521 __le32 mdsmap_epoch; 522 __u8 safe; /* true if committed to disk */ 523 __u8 is_dentry, is_target; /* true if dentry, target inode records 524 are included with reply */ 525 } __attribute__ ((packed)); 526 527 /* one for each node split */ 528 struct ceph_frag_tree_split { 529 __le32 frag; /* this frag splits... */ 530 __le32 by; /* ...by this many bits */ 531 } __attribute__ ((packed)); 532 533 struct ceph_frag_tree_head { 534 __le32 nsplits; /* num ceph_frag_tree_split records */ 535 struct ceph_frag_tree_split splits[]; 536 } __attribute__ ((packed)); 537 538 /* capability issue, for bundling with mds reply */ 539 struct ceph_mds_reply_cap { 540 __le32 caps, wanted; /* caps issued, wanted */ 541 __le64 cap_id; 542 __le32 seq, mseq; 543 __le64 realm; /* snap realm */ 544 __u8 flags; /* CEPH_CAP_FLAG_* */ 545 } __attribute__ ((packed)); 546 547 #define CEPH_CAP_FLAG_AUTH (1 << 0) /* cap is issued by auth mds */ 548 #define CEPH_CAP_FLAG_RELEASE (1 << 1) /* release the cap */ 549 550 /* inode record, for bundling with mds reply */ 551 struct ceph_mds_reply_inode { 552 __le64 ino; 553 __le64 snapid; 554 __le32 rdev; 555 __le64 version; /* inode version */ 556 __le64 xattr_version; /* version for xattr blob */ 557 struct ceph_mds_reply_cap cap; /* caps issued for this inode */ 558 struct ceph_file_layout_legacy layout; 559 struct ceph_timespec ctime, mtime, atime; 560 __le32 time_warp_seq; 561 __le64 size, max_size, truncate_size; 562 __le32 truncate_seq; 563 __le32 mode, uid, gid; 564 __le32 nlink; 565 __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */ 566 struct ceph_timespec rctime; 567 struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */ 568 } __attribute__ ((packed)); 569 /* followed by frag array, symlink string, dir layout, xattr blob */ 570 571 /* reply_lease follows dname, and reply_inode */ 572 struct ceph_mds_reply_lease { 573 __le16 mask; /* lease type(s) */ 574 __le32 duration_ms; /* lease duration */ 575 __le32 seq; 576 } __attribute__ ((packed)); 577 578 #define CEPH_LEASE_VALID (1 | 2) /* old and new bit values */ 579 #define CEPH_LEASE_PRIMARY_LINK 4 /* primary linkage */ 580 581 struct ceph_mds_reply_dirfrag { 582 __le32 frag; /* fragment */ 583 __le32 auth; /* auth mds, if this is a delegation point */ 584 __le32 ndist; /* number of mds' this is replicated on */ 585 __le32 dist[]; 586 } __attribute__ ((packed)); 587 588 #define CEPH_LOCK_FCNTL 1 589 #define CEPH_LOCK_FLOCK 2 590 #define CEPH_LOCK_FCNTL_INTR 3 591 #define CEPH_LOCK_FLOCK_INTR 4 592 593 594 #define CEPH_LOCK_SHARED 1 595 #define CEPH_LOCK_EXCL 2 596 #define CEPH_LOCK_UNLOCK 4 597 598 struct ceph_filelock { 599 __le64 start;/* file offset to start lock at */ 600 __le64 length; /* num bytes to lock; 0 for all following start */ 601 __le64 client; /* which client holds the lock */ 602 __le64 owner; /* owner the lock */ 603 __le64 pid; /* process id holding the lock on the client */ 604 __u8 type; /* shared lock, exclusive lock, or unlock */ 605 } __attribute__ ((packed)); 606 607 608 /* file access modes */ 609 #define CEPH_FILE_MODE_PIN 0 610 #define CEPH_FILE_MODE_RD 1 611 #define CEPH_FILE_MODE_WR 2 612 #define CEPH_FILE_MODE_RDWR 3 /* RD | WR */ 613 #define CEPH_FILE_MODE_LAZY 4 /* lazy io */ 614 #define CEPH_FILE_MODE_BITS 4 615 #define CEPH_FILE_MODE_MASK ((1 << CEPH_FILE_MODE_BITS) - 1) 616 617 int ceph_flags_to_mode(int flags); 618 619 #define CEPH_INLINE_NONE ((__u64)-1) 620 621 /* capability bits */ 622 #define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */ 623 624 /* generic cap bits */ 625 #define CEPH_CAP_GSHARED 1 /* client can reads */ 626 #define CEPH_CAP_GEXCL 2 /* client can read and update */ 627 #define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */ 628 #define CEPH_CAP_GRD 8 /* (file) client can read */ 629 #define CEPH_CAP_GWR 16 /* (file) client can write */ 630 #define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */ 631 #define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */ 632 #define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */ 633 634 #define CEPH_CAP_SIMPLE_BITS 2 635 #define CEPH_CAP_FILE_BITS 8 636 637 /* per-lock shift */ 638 #define CEPH_CAP_SAUTH 2 639 #define CEPH_CAP_SLINK 4 640 #define CEPH_CAP_SXATTR 6 641 #define CEPH_CAP_SFILE 8 642 #define CEPH_CAP_SFLOCK 20 643 644 #define CEPH_CAP_BITS 22 645 646 /* composed values */ 647 #define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH) 648 #define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH) 649 #define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK) 650 #define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK) 651 #define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR) 652 #define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR) 653 #define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE) 654 #define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE) 655 #define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE) 656 #define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE) 657 #define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE) 658 #define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE) 659 #define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE) 660 #define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE) 661 #define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE) 662 #define CEPH_CAP_FLOCK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFLOCK) 663 #define CEPH_CAP_FLOCK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFLOCK) 664 665 666 /* cap masks (for getattr) */ 667 #define CEPH_STAT_CAP_INODE CEPH_CAP_PIN 668 #define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */ 669 #define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN 670 #define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED 671 #define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED 672 #define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED 673 #define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED 674 #define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED 675 #define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED 676 #define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED 677 #define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */ 678 #define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED 679 #define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \ 680 CEPH_CAP_AUTH_SHARED | \ 681 CEPH_CAP_LINK_SHARED | \ 682 CEPH_CAP_FILE_SHARED | \ 683 CEPH_CAP_XATTR_SHARED) 684 #define CEPH_STAT_CAP_INLINE_DATA (CEPH_CAP_FILE_SHARED | \ 685 CEPH_CAP_FILE_RD) 686 #define CEPH_STAT_RSTAT CEPH_CAP_FILE_WREXTEND 687 688 #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \ 689 CEPH_CAP_LINK_SHARED | \ 690 CEPH_CAP_XATTR_SHARED | \ 691 CEPH_CAP_FILE_SHARED) 692 #define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \ 693 CEPH_CAP_FILE_CACHE) 694 695 #define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \ 696 CEPH_CAP_LINK_EXCL | \ 697 CEPH_CAP_XATTR_EXCL | \ 698 CEPH_CAP_FILE_EXCL) 699 #define CEPH_CAP_ANY_FILE_RD (CEPH_CAP_FILE_RD | CEPH_CAP_FILE_CACHE | \ 700 CEPH_CAP_FILE_SHARED) 701 #define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \ 702 CEPH_CAP_FILE_EXCL) 703 #define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR) 704 #define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \ 705 CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \ 706 CEPH_CAP_PIN) 707 #define CEPH_CAP_ALL_FILE (CEPH_CAP_PIN | CEPH_CAP_ANY_SHARED | \ 708 CEPH_CAP_AUTH_EXCL | CEPH_CAP_XATTR_EXCL | \ 709 CEPH_CAP_ANY_FILE_RD | CEPH_CAP_ANY_FILE_WR) 710 711 #define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \ 712 CEPH_LOCK_IXATTR) 713 714 /* cap masks async dir operations */ 715 #define CEPH_CAP_DIR_CREATE CEPH_CAP_FILE_CACHE 716 #define CEPH_CAP_DIR_UNLINK CEPH_CAP_FILE_RD 717 #define CEPH_CAP_ANY_DIR_OPS (CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_RD | \ 718 CEPH_CAP_FILE_WREXTEND | CEPH_CAP_FILE_LAZYIO) 719 720 int ceph_caps_for_mode(int mode); 721 722 enum { 723 CEPH_CAP_OP_GRANT, /* mds->client grant */ 724 CEPH_CAP_OP_REVOKE, /* mds->client revoke */ 725 CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */ 726 CEPH_CAP_OP_EXPORT, /* mds has exported the cap */ 727 CEPH_CAP_OP_IMPORT, /* mds has imported the cap */ 728 CEPH_CAP_OP_UPDATE, /* client->mds update */ 729 CEPH_CAP_OP_DROP, /* client->mds drop cap bits */ 730 CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */ 731 CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */ 732 CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */ 733 CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */ 734 CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */ 735 CEPH_CAP_OP_RENEW, /* client->mds renewal request */ 736 }; 737 738 extern const char *ceph_cap_op_name(int op); 739 740 /* flags field in client cap messages (version >= 10) */ 741 #define CEPH_CLIENT_CAPS_SYNC (1<<0) 742 #define CEPH_CLIENT_CAPS_NO_CAPSNAP (1<<1) 743 #define CEPH_CLIENT_CAPS_PENDING_CAPSNAP (1<<2) 744 745 /* 746 * caps message, used for capability callbacks, acks, requests, etc. 747 */ 748 struct ceph_mds_caps { 749 __le32 op; /* CEPH_CAP_OP_* */ 750 __le64 ino, realm; 751 __le64 cap_id; 752 __le32 seq, issue_seq; 753 __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */ 754 __le32 migrate_seq; 755 __le64 snap_follows; 756 __le32 snap_trace_len; 757 758 /* authlock */ 759 __le32 uid, gid, mode; 760 761 /* linklock */ 762 __le32 nlink; 763 764 /* xattrlock */ 765 __le32 xattr_len; 766 __le64 xattr_version; 767 768 /* filelock */ 769 __le64 size, max_size, truncate_size; 770 __le32 truncate_seq; 771 struct ceph_timespec mtime, atime, ctime; 772 struct ceph_file_layout_legacy layout; 773 __le32 time_warp_seq; 774 } __attribute__ ((packed)); 775 776 struct ceph_mds_cap_peer { 777 __le64 cap_id; 778 __le32 seq; 779 __le32 mseq; 780 __le32 mds; 781 __u8 flags; 782 } __attribute__ ((packed)); 783 784 /* cap release msg head */ 785 struct ceph_mds_cap_release { 786 __le32 num; /* number of cap_items that follow */ 787 } __attribute__ ((packed)); 788 789 struct ceph_mds_cap_item { 790 __le64 ino; 791 __le64 cap_id; 792 __le32 migrate_seq, seq; 793 } __attribute__ ((packed)); 794 795 #define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */ 796 #define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */ 797 #define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */ 798 #define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */ 799 800 extern const char *ceph_lease_op_name(int o); 801 802 /* lease msg header */ 803 struct ceph_mds_lease { 804 __u8 action; /* CEPH_MDS_LEASE_* */ 805 __le16 mask; /* which lease */ 806 __le64 ino; 807 __le64 first, last; /* snap range */ 808 __le32 seq; 809 __le32 duration_ms; /* duration of renewal */ 810 } __attribute__ ((packed)); 811 /* followed by a __le32+string for dname */ 812 813 /* client reconnect */ 814 struct ceph_mds_cap_reconnect { 815 __le64 cap_id; 816 __le32 wanted; 817 __le32 issued; 818 __le64 snaprealm; 819 __le64 pathbase; /* base ino for our path to this ino */ 820 __le32 flock_len; /* size of flock state blob, if any */ 821 } __attribute__ ((packed)); 822 /* followed by flock blob */ 823 824 struct ceph_mds_cap_reconnect_v1 { 825 __le64 cap_id; 826 __le32 wanted; 827 __le32 issued; 828 __le64 size; 829 struct ceph_timespec mtime, atime; 830 __le64 snaprealm; 831 __le64 pathbase; /* base ino for our path to this ino */ 832 } __attribute__ ((packed)); 833 834 struct ceph_mds_snaprealm_reconnect { 835 __le64 ino; /* snap realm base */ 836 __le64 seq; /* snap seq for this snap realm */ 837 __le64 parent; /* parent realm */ 838 } __attribute__ ((packed)); 839 840 /* 841 * snaps 842 */ 843 enum { 844 CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */ 845 CEPH_SNAP_OP_CREATE, 846 CEPH_SNAP_OP_DESTROY, 847 CEPH_SNAP_OP_SPLIT, 848 }; 849 850 extern const char *ceph_snap_op_name(int o); 851 852 /* snap msg header */ 853 struct ceph_mds_snap_head { 854 __le32 op; /* CEPH_SNAP_OP_* */ 855 __le64 split; /* ino to split off, if any */ 856 __le32 num_split_inos; /* # inos belonging to new child realm */ 857 __le32 num_split_realms; /* # child realms udner new child realm */ 858 __le32 trace_len; /* size of snap trace blob */ 859 } __attribute__ ((packed)); 860 /* followed by split ino list, then split realms, then the trace blob */ 861 862 /* 863 * encode info about a snaprealm, as viewed by a client 864 */ 865 struct ceph_mds_snap_realm { 866 __le64 ino; /* ino */ 867 __le64 created; /* snap: when created */ 868 __le64 parent; /* ino: parent realm */ 869 __le64 parent_since; /* snap: same parent since */ 870 __le64 seq; /* snap: version */ 871 __le32 num_snaps; 872 __le32 num_prior_parent_snaps; 873 } __attribute__ ((packed)); 874 /* followed by my snap list, then prior parent snap list */ 875 876 /* 877 * quotas 878 */ 879 struct ceph_mds_quota { 880 __le64 ino; /* ino */ 881 struct ceph_timespec rctime; 882 __le64 rbytes; /* dir stats */ 883 __le64 rfiles; 884 __le64 rsubdirs; 885 __u8 struct_v; /* compat */ 886 __u8 struct_compat; 887 __le32 struct_len; 888 __le64 max_bytes; /* quota max. bytes */ 889 __le64 max_files; /* quota max. files */ 890 } __attribute__ ((packed)); 891 892 #endif 893
