xref: /linux-6.15/include/linux/ceph/ceph_fs.h (revision f96cffd7)
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 CEPH_FS_H
13 #define CEPH_FS_H
14 
15 #include <linux/ceph/msgr.h>
16 #include <linux/ceph/rados.h>
17 
18 /*
19  * subprotocol versions.  when specific messages types or high-level
20  * protocols change, bump the affected components.  we keep rev
21  * internal cluster protocols separately from the public,
22  * client-facing protocol.
23  */
24 #define CEPH_OSDC_PROTOCOL   24 /* server/client */
25 #define CEPH_MDSC_PROTOCOL   32 /* server/client */
26 #define CEPH_MONC_PROTOCOL   15 /* server/client */
27 
28 
29 #define CEPH_INO_ROOT   1
30 #define CEPH_INO_CEPH   2       /* hidden .ceph dir */
31 #define CEPH_INO_DOTDOT 3	/* used by ceph fuse for parent (..) */
32 
33 /* arbitrary limit on max # of monitors (cluster of 3 is typical) */
34 #define CEPH_MAX_MON   31
35 
36 /*
37  * legacy ceph_file_layoute
38  */
39 struct ceph_file_layout_legacy {
40 	/* file -> object mapping */
41 	__le32 fl_stripe_unit;     /* stripe unit, in bytes.  must be multiple
42 				      of page size. */
43 	__le32 fl_stripe_count;    /* over this many objects */
44 	__le32 fl_object_size;     /* until objects are this big, then move to
45 				      new objects */
46 	__le32 fl_cas_hash;        /* UNUSED.  0 = none; 1 = sha256 */
47 
48 	/* pg -> disk layout */
49 	__le32 fl_object_stripe_unit;  /* UNUSED.  for per-object parity, if any */
50 
51 	/* object -> pg layout */
52 	__le32 fl_unused;       /* unused; used to be preferred primary for pg (-1 for none) */
53 	__le32 fl_pg_pool;      /* namespace, crush ruleset, rep level */
54 } __attribute__ ((packed));
55 
56 struct ceph_string;
57 /*
58  * ceph_file_layout - describe data layout for a file/inode
59  */
60 struct ceph_file_layout {
61 	/* file -> object mapping */
62 	u32 stripe_unit;   /* stripe unit, in bytes */
63 	u32 stripe_count;  /* over this many objects */
64 	u32 object_size;   /* until objects are this big */
65 	s64 pool_id;        /* rados pool id */
66 	struct ceph_string __rcu *pool_ns; /* rados pool namespace */
67 };
68 
69 extern int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
70 extern void ceph_file_layout_from_legacy(struct ceph_file_layout *fl,
71 				struct ceph_file_layout_legacy *legacy);
72 extern void ceph_file_layout_to_legacy(struct ceph_file_layout *fl,
73 				struct ceph_file_layout_legacy *legacy);
74 
75 #define CEPH_MIN_STRIPE_UNIT 65536
76 
77 struct ceph_dir_layout {
78 	__u8   dl_dir_hash;   /* see ceph_hash.h for ids */
79 	__u8   dl_unused1;
80 	__u16  dl_unused2;
81 	__u32  dl_unused3;
82 } __attribute__ ((packed));
83 
84 /* crypto algorithms */
85 #define CEPH_CRYPTO_NONE 0x0
86 #define CEPH_CRYPTO_AES  0x1
87 
88 #define CEPH_AES_IV "cephsageyudagreg"
89 
90 /* security/authentication protocols */
91 #define CEPH_AUTH_UNKNOWN	0x0
92 #define CEPH_AUTH_NONE	 	0x1
93 #define CEPH_AUTH_CEPHX	 	0x2
94 
95 #define CEPH_AUTH_UID_DEFAULT ((__u64) -1)
96 
97 
98 /*********************************************
99  * message layer
100  */
101 
102 /*
103  * message types
104  */
105 
106 /* misc */
107 #define CEPH_MSG_SHUTDOWN               1
108 #define CEPH_MSG_PING                   2
109 
110 /* client <-> monitor */
111 #define CEPH_MSG_MON_MAP                4
112 #define CEPH_MSG_MON_GET_MAP            5
113 #define CEPH_MSG_STATFS                 13
114 #define CEPH_MSG_STATFS_REPLY           14
115 #define CEPH_MSG_MON_SUBSCRIBE          15
116 #define CEPH_MSG_MON_SUBSCRIBE_ACK      16
117 #define CEPH_MSG_AUTH			17
118 #define CEPH_MSG_AUTH_REPLY		18
119 #define CEPH_MSG_MON_GET_VERSION        19
120 #define CEPH_MSG_MON_GET_VERSION_REPLY  20
121 
122 /* client <-> mds */
123 #define CEPH_MSG_MDS_MAP                21
124 #define CEPH_MSG_FS_MAP_USER            103
125 
126 #define CEPH_MSG_CLIENT_SESSION         22
127 #define CEPH_MSG_CLIENT_RECONNECT       23
128 
129 #define CEPH_MSG_CLIENT_REQUEST         24
130 #define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
131 #define CEPH_MSG_CLIENT_REPLY           26
132 #define CEPH_MSG_CLIENT_CAPS            0x310
133 #define CEPH_MSG_CLIENT_LEASE           0x311
134 #define CEPH_MSG_CLIENT_SNAP            0x312
135 #define CEPH_MSG_CLIENT_CAPRELEASE      0x313
136 
137 /* pool ops */
138 #define CEPH_MSG_POOLOP_REPLY           48
139 #define CEPH_MSG_POOLOP                 49
140 
141 /* mon commands */
142 #define CEPH_MSG_MON_COMMAND            50
143 #define CEPH_MSG_MON_COMMAND_ACK        51
144 
145 /* osd */
146 #define CEPH_MSG_OSD_MAP                41
147 #define CEPH_MSG_OSD_OP                 42
148 #define CEPH_MSG_OSD_OPREPLY            43
149 #define CEPH_MSG_WATCH_NOTIFY           44
150 
151 
152 /* watch-notify operations */
153 enum {
154 	CEPH_WATCH_EVENT_NOTIFY		  = 1, /* notifying watcher */
155 	CEPH_WATCH_EVENT_NOTIFY_COMPLETE  = 2, /* notifier notified when done */
156 	CEPH_WATCH_EVENT_DISCONNECT       = 3, /* we were disconnected */
157 };
158 
159 
160 struct ceph_mon_request_header {
161 	__le64 have_version;
162 	__le16 session_mon;
163 	__le64 session_mon_tid;
164 } __attribute__ ((packed));
165 
166 struct ceph_mon_statfs {
167 	struct ceph_mon_request_header monhdr;
168 	struct ceph_fsid fsid;
169 } __attribute__ ((packed));
170 
171 struct ceph_statfs {
172 	__le64 kb, kb_used, kb_avail;
173 	__le64 num_objects;
174 } __attribute__ ((packed));
175 
176 struct ceph_mon_statfs_reply {
177 	struct ceph_fsid fsid;
178 	__le64 version;
179 	struct ceph_statfs st;
180 } __attribute__ ((packed));
181 
182 struct ceph_mon_command {
183 	struct ceph_mon_request_header monhdr;
184 	struct ceph_fsid fsid;
185 	__le32 num_strs;         /* always 1 */
186 	__le32 str_len;
187 	char str[];
188 } __attribute__ ((packed));
189 
190 struct ceph_osd_getmap {
191 	struct ceph_mon_request_header monhdr;
192 	struct ceph_fsid fsid;
193 	__le32 start;
194 } __attribute__ ((packed));
195 
196 struct ceph_mds_getmap {
197 	struct ceph_mon_request_header monhdr;
198 	struct ceph_fsid fsid;
199 } __attribute__ ((packed));
200 
201 struct ceph_client_mount {
202 	struct ceph_mon_request_header monhdr;
203 } __attribute__ ((packed));
204 
205 #define CEPH_SUBSCRIBE_ONETIME    1  /* i want only 1 update after have */
206 
207 struct ceph_mon_subscribe_item {
208 	__le64 start;
209 	__u8 flags;
210 } __attribute__ ((packed));
211 
212 struct ceph_mon_subscribe_ack {
213 	__le32 duration;         /* seconds */
214 	struct ceph_fsid fsid;
215 } __attribute__ ((packed));
216 
217 #define CEPH_FS_CLUSTER_ID_NONE  -1
218 
219 /*
220  * mdsmap flags
221  */
222 #define CEPH_MDSMAP_DOWN    (1<<0)  /* cluster deliberately down */
223 
224 /*
225  * mds states
226  *   > 0 -> in
227  *  <= 0 -> out
228  */
229 #define CEPH_MDS_STATE_DNE          0  /* down, does not exist. */
230 #define CEPH_MDS_STATE_STOPPED     -1  /* down, once existed, but no subtrees.
231 					  empty log. */
232 #define CEPH_MDS_STATE_BOOT        -4  /* up, boot announcement. */
233 #define CEPH_MDS_STATE_STANDBY     -5  /* up, idle.  waiting for assignment. */
234 #define CEPH_MDS_STATE_CREATING    -6  /* up, creating MDS instance. */
235 #define CEPH_MDS_STATE_STARTING    -7  /* up, starting previously stopped mds */
236 #define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
237 #define CEPH_MDS_STATE_REPLAYONCE   -9 /* up, replaying an active node's journal */
238 
239 #define CEPH_MDS_STATE_REPLAY       8  /* up, replaying journal. */
240 #define CEPH_MDS_STATE_RESOLVE      9  /* up, disambiguating distributed
241 					  operations (import, rename, etc.) */
242 #define CEPH_MDS_STATE_RECONNECT    10 /* up, reconnect to clients */
243 #define CEPH_MDS_STATE_REJOIN       11 /* up, rejoining distributed cache */
244 #define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
245 #define CEPH_MDS_STATE_ACTIVE       13 /* up, active */
246 #define CEPH_MDS_STATE_STOPPING     14 /* up, but exporting metadata */
247 
248 extern const char *ceph_mds_state_name(int s);
249 
250 
251 /*
252  * metadata lock types.
253  *  - these are bitmasks.. we can compose them
254  *  - they also define the lock ordering by the MDS
255  *  - a few of these are internal to the mds
256  */
257 #define CEPH_LOCK_DVERSION    1
258 #define CEPH_LOCK_DN          2
259 #define CEPH_LOCK_ISNAP       16
260 #define CEPH_LOCK_IVERSION    32    /* mds internal */
261 #define CEPH_LOCK_IFILE       64
262 #define CEPH_LOCK_IAUTH       128
263 #define CEPH_LOCK_ILINK       256
264 #define CEPH_LOCK_IDFT        512   /* dir frag tree */
265 #define CEPH_LOCK_INEST       1024  /* mds internal */
266 #define CEPH_LOCK_IXATTR      2048
267 #define CEPH_LOCK_IFLOCK      4096  /* advisory file locks */
268 #define CEPH_LOCK_INO         8192  /* immutable inode bits; not a lock */
269 #define CEPH_LOCK_IPOLICY     16384 /* policy lock on dirs. MDS internal */
270 
271 /* client_session ops */
272 enum {
273 	CEPH_SESSION_REQUEST_OPEN,
274 	CEPH_SESSION_OPEN,
275 	CEPH_SESSION_REQUEST_CLOSE,
276 	CEPH_SESSION_CLOSE,
277 	CEPH_SESSION_REQUEST_RENEWCAPS,
278 	CEPH_SESSION_RENEWCAPS,
279 	CEPH_SESSION_STALE,
280 	CEPH_SESSION_RECALL_STATE,
281 	CEPH_SESSION_FLUSHMSG,
282 	CEPH_SESSION_FLUSHMSG_ACK,
283 	CEPH_SESSION_FORCE_RO,
284 	CEPH_SESSION_REJECT,
285 };
286 
287 extern const char *ceph_session_op_name(int op);
288 
289 struct ceph_mds_session_head {
290 	__le32 op;
291 	__le64 seq;
292 	struct ceph_timespec stamp;
293 	__le32 max_caps, max_leases;
294 } __attribute__ ((packed));
295 
296 /* client_request */
297 /*
298  * metadata ops.
299  *  & 0x001000 -> write op
300  *  & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
301  &  & 0x100000 -> use weird ino/path trace
302  */
303 #define CEPH_MDS_OP_WRITE        0x001000
304 enum {
305 	CEPH_MDS_OP_LOOKUP     = 0x00100,
306 	CEPH_MDS_OP_GETATTR    = 0x00101,
307 	CEPH_MDS_OP_LOOKUPHASH = 0x00102,
308 	CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
309 	CEPH_MDS_OP_LOOKUPINO  = 0x00104,
310 	CEPH_MDS_OP_LOOKUPNAME = 0x00105,
311 
312 	CEPH_MDS_OP_SETXATTR   = 0x01105,
313 	CEPH_MDS_OP_RMXATTR    = 0x01106,
314 	CEPH_MDS_OP_SETLAYOUT  = 0x01107,
315 	CEPH_MDS_OP_SETATTR    = 0x01108,
316 	CEPH_MDS_OP_SETFILELOCK= 0x01109,
317 	CEPH_MDS_OP_GETFILELOCK= 0x00110,
318 	CEPH_MDS_OP_SETDIRLAYOUT=0x0110a,
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 	CEPH_MDS_OP_RENAMESNAP = 0x01403,
337 };
338 
339 extern const char *ceph_mds_op_name(int op);
340 
341 
342 #define CEPH_SETATTR_MODE   1
343 #define CEPH_SETATTR_UID    2
344 #define CEPH_SETATTR_GID    4
345 #define CEPH_SETATTR_MTIME  8
346 #define CEPH_SETATTR_ATIME 16
347 #define CEPH_SETATTR_SIZE  32
348 #define CEPH_SETATTR_CTIME 64
349 
350 /*
351  * Ceph setxattr request flags.
352  */
353 #define CEPH_XATTR_CREATE  (1 << 0)
354 #define CEPH_XATTR_REPLACE (1 << 1)
355 #define CEPH_XATTR_REMOVE  (1 << 31)
356 
357 /*
358  * readdir request flags;
359  */
360 #define CEPH_READDIR_REPLY_BITFLAGS	(1<<0)
361 
362 /*
363  * readdir reply flags.
364  */
365 #define CEPH_READDIR_FRAG_END		(1<<0)
366 #define CEPH_READDIR_FRAG_COMPLETE	(1<<8)
367 #define CEPH_READDIR_HASH_ORDER		(1<<9)
368 
369 union ceph_mds_request_args {
370 	struct {
371 		__le32 mask;                 /* CEPH_CAP_* */
372 	} __attribute__ ((packed)) getattr;
373 	struct {
374 		__le32 mode;
375 		__le32 uid;
376 		__le32 gid;
377 		struct ceph_timespec mtime;
378 		struct ceph_timespec atime;
379 		__le64 size, old_size;       /* old_size needed by truncate */
380 		__le32 mask;                 /* CEPH_SETATTR_* */
381 	} __attribute__ ((packed)) setattr;
382 	struct {
383 		__le32 frag;                 /* which dir fragment */
384 		__le32 max_entries;          /* how many dentries to grab */
385 		__le32 max_bytes;
386 		__le16 flags;
387 	} __attribute__ ((packed)) readdir;
388 	struct {
389 		__le32 mode;
390 		__le32 rdev;
391 	} __attribute__ ((packed)) mknod;
392 	struct {
393 		__le32 mode;
394 	} __attribute__ ((packed)) mkdir;
395 	struct {
396 		__le32 flags;
397 		__le32 mode;
398 		__le32 stripe_unit;          /* layout for newly created file */
399 		__le32 stripe_count;         /* ... */
400 		__le32 object_size;
401 		__le32 file_replication;
402                __le32 mask;                 /* CEPH_CAP_* */
403                __le32 old_size;
404 	} __attribute__ ((packed)) open;
405 	struct {
406 		__le32 flags;
407 	} __attribute__ ((packed)) setxattr;
408 	struct {
409 		struct ceph_file_layout_legacy layout;
410 	} __attribute__ ((packed)) setlayout;
411 	struct {
412 		__u8 rule; /* currently fcntl or flock */
413 		__u8 type; /* shared, exclusive, remove*/
414 		__le64 owner; /* owner of the lock */
415 		__le64 pid; /* process id requesting the lock */
416 		__le64 start; /* initial location to lock */
417 		__le64 length; /* num bytes to lock from start */
418 		__u8 wait; /* will caller wait for lock to become available? */
419 	} __attribute__ ((packed)) filelock_change;
420 } __attribute__ ((packed));
421 
422 #define CEPH_MDS_FLAG_REPLAY        1  /* this is a replayed op */
423 #define CEPH_MDS_FLAG_WANT_DENTRY   2  /* want dentry in reply */
424 
425 struct ceph_mds_request_head {
426 	__le64 oldest_client_tid;
427 	__le32 mdsmap_epoch;           /* on client */
428 	__le32 flags;                  /* CEPH_MDS_FLAG_* */
429 	__u8 num_retry, num_fwd;       /* count retry, fwd attempts */
430 	__le16 num_releases;           /* # include cap/lease release records */
431 	__le32 op;                     /* mds op code */
432 	__le32 caller_uid, caller_gid;
433 	__le64 ino;                    /* use this ino for openc, mkdir, mknod,
434 					  etc. (if replaying) */
435 	union ceph_mds_request_args args;
436 } __attribute__ ((packed));
437 
438 /* cap/lease release record */
439 struct ceph_mds_request_release {
440 	__le64 ino, cap_id;            /* ino and unique cap id */
441 	__le32 caps, wanted;           /* new issued, wanted */
442 	__le32 seq, issue_seq, mseq;
443 	__le32 dname_seq;              /* if releasing a dentry lease, a */
444 	__le32 dname_len;              /* string follows. */
445 } __attribute__ ((packed));
446 
447 /* client reply */
448 struct ceph_mds_reply_head {
449 	__le32 op;
450 	__le32 result;
451 	__le32 mdsmap_epoch;
452 	__u8 safe;                     /* true if committed to disk */
453 	__u8 is_dentry, is_target;     /* true if dentry, target inode records
454 					  are included with reply */
455 } __attribute__ ((packed));
456 
457 /* one for each node split */
458 struct ceph_frag_tree_split {
459 	__le32 frag;                   /* this frag splits... */
460 	__le32 by;                     /* ...by this many bits */
461 } __attribute__ ((packed));
462 
463 struct ceph_frag_tree_head {
464 	__le32 nsplits;                /* num ceph_frag_tree_split records */
465 	struct ceph_frag_tree_split splits[];
466 } __attribute__ ((packed));
467 
468 /* capability issue, for bundling with mds reply */
469 struct ceph_mds_reply_cap {
470 	__le32 caps, wanted;           /* caps issued, wanted */
471 	__le64 cap_id;
472 	__le32 seq, mseq;
473 	__le64 realm;                  /* snap realm */
474 	__u8 flags;                    /* CEPH_CAP_FLAG_* */
475 } __attribute__ ((packed));
476 
477 #define CEPH_CAP_FLAG_AUTH	(1 << 0)  /* cap is issued by auth mds */
478 #define CEPH_CAP_FLAG_RELEASE	(1 << 1)  /* release the cap */
479 
480 /* inode record, for bundling with mds reply */
481 struct ceph_mds_reply_inode {
482 	__le64 ino;
483 	__le64 snapid;
484 	__le32 rdev;
485 	__le64 version;                /* inode version */
486 	__le64 xattr_version;          /* version for xattr blob */
487 	struct ceph_mds_reply_cap cap; /* caps issued for this inode */
488 	struct ceph_file_layout_legacy layout;
489 	struct ceph_timespec ctime, mtime, atime;
490 	__le32 time_warp_seq;
491 	__le64 size, max_size, truncate_size;
492 	__le32 truncate_seq;
493 	__le32 mode, uid, gid;
494 	__le32 nlink;
495 	__le64 files, subdirs, rbytes, rfiles, rsubdirs;  /* dir stats */
496 	struct ceph_timespec rctime;
497 	struct ceph_frag_tree_head fragtree;  /* (must be at end of struct) */
498 } __attribute__ ((packed));
499 /* followed by frag array, symlink string, dir layout, xattr blob */
500 
501 /* reply_lease follows dname, and reply_inode */
502 struct ceph_mds_reply_lease {
503 	__le16 mask;            /* lease type(s) */
504 	__le32 duration_ms;     /* lease duration */
505 	__le32 seq;
506 } __attribute__ ((packed));
507 
508 struct ceph_mds_reply_dirfrag {
509 	__le32 frag;            /* fragment */
510 	__le32 auth;            /* auth mds, if this is a delegation point */
511 	__le32 ndist;           /* number of mds' this is replicated on */
512 	__le32 dist[];
513 } __attribute__ ((packed));
514 
515 #define CEPH_LOCK_FCNTL		1
516 #define CEPH_LOCK_FLOCK		2
517 #define CEPH_LOCK_FCNTL_INTR    3
518 #define CEPH_LOCK_FLOCK_INTR    4
519 
520 
521 #define CEPH_LOCK_SHARED   1
522 #define CEPH_LOCK_EXCL     2
523 #define CEPH_LOCK_UNLOCK   4
524 
525 struct ceph_filelock {
526 	__le64 start;/* file offset to start lock at */
527 	__le64 length; /* num bytes to lock; 0 for all following start */
528 	__le64 client; /* which client holds the lock */
529 	__le64 owner; /* owner the lock */
530 	__le64 pid; /* process id holding the lock on the client */
531 	__u8 type; /* shared lock, exclusive lock, or unlock */
532 } __attribute__ ((packed));
533 
534 
535 /* file access modes */
536 #define CEPH_FILE_MODE_PIN        0
537 #define CEPH_FILE_MODE_RD         1
538 #define CEPH_FILE_MODE_WR         2
539 #define CEPH_FILE_MODE_RDWR       3  /* RD | WR */
540 #define CEPH_FILE_MODE_LAZY       4  /* lazy io */
541 #define CEPH_FILE_MODE_BITS       4
542 
543 int ceph_flags_to_mode(int flags);
544 
545 #define CEPH_INLINE_NONE	((__u64)-1)
546 
547 /* capability bits */
548 #define CEPH_CAP_PIN         1  /* no specific capabilities beyond the pin */
549 
550 /* generic cap bits */
551 #define CEPH_CAP_GSHARED     1  /* client can reads */
552 #define CEPH_CAP_GEXCL       2  /* client can read and update */
553 #define CEPH_CAP_GCACHE      4  /* (file) client can cache reads */
554 #define CEPH_CAP_GRD         8  /* (file) client can read */
555 #define CEPH_CAP_GWR        16  /* (file) client can write */
556 #define CEPH_CAP_GBUFFER    32  /* (file) client can buffer writes */
557 #define CEPH_CAP_GWREXTEND  64  /* (file) client can extend EOF */
558 #define CEPH_CAP_GLAZYIO   128  /* (file) client can perform lazy io */
559 
560 #define CEPH_CAP_SIMPLE_BITS  2
561 #define CEPH_CAP_FILE_BITS    8
562 
563 /* per-lock shift */
564 #define CEPH_CAP_SAUTH      2
565 #define CEPH_CAP_SLINK      4
566 #define CEPH_CAP_SXATTR     6
567 #define CEPH_CAP_SFILE      8
568 #define CEPH_CAP_SFLOCK    20
569 
570 #define CEPH_CAP_BITS      22
571 
572 /* composed values */
573 #define CEPH_CAP_AUTH_SHARED  (CEPH_CAP_GSHARED  << CEPH_CAP_SAUTH)
574 #define CEPH_CAP_AUTH_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SAUTH)
575 #define CEPH_CAP_LINK_SHARED  (CEPH_CAP_GSHARED  << CEPH_CAP_SLINK)
576 #define CEPH_CAP_LINK_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SLINK)
577 #define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED  << CEPH_CAP_SXATTR)
578 #define CEPH_CAP_XATTR_EXCL    (CEPH_CAP_GEXCL     << CEPH_CAP_SXATTR)
579 #define CEPH_CAP_FILE(x)    (x << CEPH_CAP_SFILE)
580 #define CEPH_CAP_FILE_SHARED   (CEPH_CAP_GSHARED   << CEPH_CAP_SFILE)
581 #define CEPH_CAP_FILE_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SFILE)
582 #define CEPH_CAP_FILE_CACHE    (CEPH_CAP_GCACHE    << CEPH_CAP_SFILE)
583 #define CEPH_CAP_FILE_RD       (CEPH_CAP_GRD       << CEPH_CAP_SFILE)
584 #define CEPH_CAP_FILE_WR       (CEPH_CAP_GWR       << CEPH_CAP_SFILE)
585 #define CEPH_CAP_FILE_BUFFER   (CEPH_CAP_GBUFFER   << CEPH_CAP_SFILE)
586 #define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
587 #define CEPH_CAP_FILE_LAZYIO   (CEPH_CAP_GLAZYIO   << CEPH_CAP_SFILE)
588 #define CEPH_CAP_FLOCK_SHARED  (CEPH_CAP_GSHARED   << CEPH_CAP_SFLOCK)
589 #define CEPH_CAP_FLOCK_EXCL    (CEPH_CAP_GEXCL     << CEPH_CAP_SFLOCK)
590 
591 
592 /* cap masks (for getattr) */
593 #define CEPH_STAT_CAP_INODE    CEPH_CAP_PIN
594 #define CEPH_STAT_CAP_TYPE     CEPH_CAP_PIN  /* mode >> 12 */
595 #define CEPH_STAT_CAP_SYMLINK  CEPH_CAP_PIN
596 #define CEPH_STAT_CAP_UID      CEPH_CAP_AUTH_SHARED
597 #define CEPH_STAT_CAP_GID      CEPH_CAP_AUTH_SHARED
598 #define CEPH_STAT_CAP_MODE     CEPH_CAP_AUTH_SHARED
599 #define CEPH_STAT_CAP_NLINK    CEPH_CAP_LINK_SHARED
600 #define CEPH_STAT_CAP_LAYOUT   CEPH_CAP_FILE_SHARED
601 #define CEPH_STAT_CAP_MTIME    CEPH_CAP_FILE_SHARED
602 #define CEPH_STAT_CAP_SIZE     CEPH_CAP_FILE_SHARED
603 #define CEPH_STAT_CAP_ATIME    CEPH_CAP_FILE_SHARED  /* fixme */
604 #define CEPH_STAT_CAP_XATTR    CEPH_CAP_XATTR_SHARED
605 #define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN |			\
606 				 CEPH_CAP_AUTH_SHARED |	\
607 				 CEPH_CAP_LINK_SHARED |	\
608 				 CEPH_CAP_FILE_SHARED |	\
609 				 CEPH_CAP_XATTR_SHARED)
610 #define CEPH_STAT_CAP_INLINE_DATA (CEPH_CAP_FILE_SHARED | \
611 				   CEPH_CAP_FILE_RD)
612 
613 #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED |			\
614 			      CEPH_CAP_LINK_SHARED |			\
615 			      CEPH_CAP_XATTR_SHARED |			\
616 			      CEPH_CAP_FILE_SHARED)
617 #define CEPH_CAP_ANY_RD   (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD |	\
618 			   CEPH_CAP_FILE_CACHE)
619 
620 #define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL |		\
621 			   CEPH_CAP_LINK_EXCL |		\
622 			   CEPH_CAP_XATTR_EXCL |	\
623 			   CEPH_CAP_FILE_EXCL)
624 #define CEPH_CAP_ANY_FILE_RD (CEPH_CAP_FILE_RD | CEPH_CAP_FILE_CACHE | \
625 			      CEPH_CAP_FILE_SHARED)
626 #define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER |	\
627 			      CEPH_CAP_FILE_EXCL)
628 #define CEPH_CAP_ANY_WR   (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
629 #define CEPH_CAP_ANY      (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
630 			   CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \
631 			   CEPH_CAP_PIN)
632 
633 #define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
634 			CEPH_LOCK_IXATTR)
635 
636 int ceph_caps_for_mode(int mode);
637 
638 enum {
639 	CEPH_CAP_OP_GRANT,         /* mds->client grant */
640 	CEPH_CAP_OP_REVOKE,        /* mds->client revoke */
641 	CEPH_CAP_OP_TRUNC,         /* mds->client trunc notify */
642 	CEPH_CAP_OP_EXPORT,        /* mds has exported the cap */
643 	CEPH_CAP_OP_IMPORT,        /* mds has imported the cap */
644 	CEPH_CAP_OP_UPDATE,        /* client->mds update */
645 	CEPH_CAP_OP_DROP,          /* client->mds drop cap bits */
646 	CEPH_CAP_OP_FLUSH,         /* client->mds cap writeback */
647 	CEPH_CAP_OP_FLUSH_ACK,     /* mds->client flushed */
648 	CEPH_CAP_OP_FLUSHSNAP,     /* client->mds flush snapped metadata */
649 	CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
650 	CEPH_CAP_OP_RELEASE,       /* client->mds release (clean) cap */
651 	CEPH_CAP_OP_RENEW,         /* client->mds renewal request */
652 };
653 
654 extern const char *ceph_cap_op_name(int op);
655 
656 /*
657  * caps message, used for capability callbacks, acks, requests, etc.
658  */
659 struct ceph_mds_caps {
660 	__le32 op;                  /* CEPH_CAP_OP_* */
661 	__le64 ino, realm;
662 	__le64 cap_id;
663 	__le32 seq, issue_seq;
664 	__le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
665 	__le32 migrate_seq;
666 	__le64 snap_follows;
667 	__le32 snap_trace_len;
668 
669 	/* authlock */
670 	__le32 uid, gid, mode;
671 
672 	/* linklock */
673 	__le32 nlink;
674 
675 	/* xattrlock */
676 	__le32 xattr_len;
677 	__le64 xattr_version;
678 
679 	/* filelock */
680 	__le64 size, max_size, truncate_size;
681 	__le32 truncate_seq;
682 	struct ceph_timespec mtime, atime, ctime;
683 	struct ceph_file_layout_legacy layout;
684 	__le32 time_warp_seq;
685 } __attribute__ ((packed));
686 
687 struct ceph_mds_cap_peer {
688 	__le64 cap_id;
689 	__le32 seq;
690 	__le32 mseq;
691 	__le32 mds;
692 	__u8   flags;
693 } __attribute__ ((packed));
694 
695 /* cap release msg head */
696 struct ceph_mds_cap_release {
697 	__le32 num;                /* number of cap_items that follow */
698 } __attribute__ ((packed));
699 
700 struct ceph_mds_cap_item {
701 	__le64 ino;
702 	__le64 cap_id;
703 	__le32 migrate_seq, seq;
704 } __attribute__ ((packed));
705 
706 #define CEPH_MDS_LEASE_REVOKE           1  /*    mds  -> client */
707 #define CEPH_MDS_LEASE_RELEASE          2  /* client  -> mds    */
708 #define CEPH_MDS_LEASE_RENEW            3  /* client <-> mds    */
709 #define CEPH_MDS_LEASE_REVOKE_ACK       4  /* client  -> mds    */
710 
711 extern const char *ceph_lease_op_name(int o);
712 
713 /* lease msg header */
714 struct ceph_mds_lease {
715 	__u8 action;            /* CEPH_MDS_LEASE_* */
716 	__le16 mask;            /* which lease */
717 	__le64 ino;
718 	__le64 first, last;     /* snap range */
719 	__le32 seq;
720 	__le32 duration_ms;     /* duration of renewal */
721 } __attribute__ ((packed));
722 /* followed by a __le32+string for dname */
723 
724 /* client reconnect */
725 struct ceph_mds_cap_reconnect {
726 	__le64 cap_id;
727 	__le32 wanted;
728 	__le32 issued;
729 	__le64 snaprealm;
730 	__le64 pathbase;        /* base ino for our path to this ino */
731 	__le32 flock_len;       /* size of flock state blob, if any */
732 } __attribute__ ((packed));
733 /* followed by flock blob */
734 
735 struct ceph_mds_cap_reconnect_v1 {
736 	__le64 cap_id;
737 	__le32 wanted;
738 	__le32 issued;
739 	__le64 size;
740 	struct ceph_timespec mtime, atime;
741 	__le64 snaprealm;
742 	__le64 pathbase;        /* base ino for our path to this ino */
743 } __attribute__ ((packed));
744 
745 struct ceph_mds_snaprealm_reconnect {
746 	__le64 ino;     /* snap realm base */
747 	__le64 seq;     /* snap seq for this snap realm */
748 	__le64 parent;  /* parent realm */
749 } __attribute__ ((packed));
750 
751 /*
752  * snaps
753  */
754 enum {
755 	CEPH_SNAP_OP_UPDATE,  /* CREATE or DESTROY */
756 	CEPH_SNAP_OP_CREATE,
757 	CEPH_SNAP_OP_DESTROY,
758 	CEPH_SNAP_OP_SPLIT,
759 };
760 
761 extern const char *ceph_snap_op_name(int o);
762 
763 /* snap msg header */
764 struct ceph_mds_snap_head {
765 	__le32 op;                /* CEPH_SNAP_OP_* */
766 	__le64 split;             /* ino to split off, if any */
767 	__le32 num_split_inos;    /* # inos belonging to new child realm */
768 	__le32 num_split_realms;  /* # child realms udner new child realm */
769 	__le32 trace_len;         /* size of snap trace blob */
770 } __attribute__ ((packed));
771 /* followed by split ino list, then split realms, then the trace blob */
772 
773 /*
774  * encode info about a snaprealm, as viewed by a client
775  */
776 struct ceph_mds_snap_realm {
777 	__le64 ino;           /* ino */
778 	__le64 created;       /* snap: when created */
779 	__le64 parent;        /* ino: parent realm */
780 	__le64 parent_since;  /* snap: same parent since */
781 	__le64 seq;           /* snap: version */
782 	__le32 num_snaps;
783 	__le32 num_prior_parent_snaps;
784 } __attribute__ ((packed));
785 /* followed by my snap list, then prior parent snap list */
786 
787 #endif
788