xref: /linux-6.15/include/linux/ceph/ceph_fs.h (revision 78f73bf0)
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  * ceph_file_layout - describe data layout for a file/inode
38  */
39 struct ceph_file_layout {
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 #define ceph_file_layout_su(l) ((__s32)le32_to_cpu((l).fl_stripe_unit))
57 #define ceph_file_layout_stripe_count(l) \
58 	((__s32)le32_to_cpu((l).fl_stripe_count))
59 #define ceph_file_layout_object_size(l) ((__s32)le32_to_cpu((l).fl_object_size))
60 #define ceph_file_layout_cas_hash(l) ((__s32)le32_to_cpu((l).fl_cas_hash))
61 #define ceph_file_layout_object_su(l) \
62 	((__s32)le32_to_cpu((l).fl_object_stripe_unit))
63 #define ceph_file_layout_pg_pool(l) \
64 	((__s32)le32_to_cpu((l).fl_pg_pool))
65 
66 static inline unsigned ceph_file_layout_stripe_width(struct ceph_file_layout *l)
67 {
68 	return le32_to_cpu(l->fl_stripe_unit) *
69 		le32_to_cpu(l->fl_stripe_count);
70 }
71 
72 /* "period" == bytes before i start on a new set of objects */
73 static inline unsigned ceph_file_layout_period(struct ceph_file_layout *l)
74 {
75 	return le32_to_cpu(l->fl_object_size) *
76 		le32_to_cpu(l->fl_stripe_count);
77 }
78 
79 #define CEPH_MIN_STRIPE_UNIT 65536
80 
81 int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
82 
83 struct ceph_dir_layout {
84 	__u8   dl_dir_hash;   /* see ceph_hash.h for ids */
85 	__u8   dl_unused1;
86 	__u16  dl_unused2;
87 	__u32  dl_unused3;
88 } __attribute__ ((packed));
89 
90 /* crypto algorithms */
91 #define CEPH_CRYPTO_NONE 0x0
92 #define CEPH_CRYPTO_AES  0x1
93 
94 #define CEPH_AES_IV "cephsageyudagreg"
95 
96 /* security/authentication protocols */
97 #define CEPH_AUTH_UNKNOWN	0x0
98 #define CEPH_AUTH_NONE	 	0x1
99 #define CEPH_AUTH_CEPHX	 	0x2
100 
101 #define CEPH_AUTH_UID_DEFAULT ((__u64) -1)
102 
103 
104 /*********************************************
105  * message layer
106  */
107 
108 /*
109  * message types
110  */
111 
112 /* misc */
113 #define CEPH_MSG_SHUTDOWN               1
114 #define CEPH_MSG_PING                   2
115 
116 /* client <-> monitor */
117 #define CEPH_MSG_MON_MAP                4
118 #define CEPH_MSG_MON_GET_MAP            5
119 #define CEPH_MSG_STATFS                 13
120 #define CEPH_MSG_STATFS_REPLY           14
121 #define CEPH_MSG_MON_SUBSCRIBE          15
122 #define CEPH_MSG_MON_SUBSCRIBE_ACK      16
123 #define CEPH_MSG_AUTH			17
124 #define CEPH_MSG_AUTH_REPLY		18
125 #define CEPH_MSG_MON_GET_VERSION        19
126 #define CEPH_MSG_MON_GET_VERSION_REPLY  20
127 
128 /* client <-> mds */
129 #define CEPH_MSG_MDS_MAP                21
130 
131 #define CEPH_MSG_CLIENT_SESSION         22
132 #define CEPH_MSG_CLIENT_RECONNECT       23
133 
134 #define CEPH_MSG_CLIENT_REQUEST         24
135 #define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
136 #define CEPH_MSG_CLIENT_REPLY           26
137 #define CEPH_MSG_CLIENT_CAPS            0x310
138 #define CEPH_MSG_CLIENT_LEASE           0x311
139 #define CEPH_MSG_CLIENT_SNAP            0x312
140 #define CEPH_MSG_CLIENT_CAPRELEASE      0x313
141 
142 /* pool ops */
143 #define CEPH_MSG_POOLOP_REPLY           48
144 #define CEPH_MSG_POOLOP                 49
145 
146 
147 /* osd */
148 #define CEPH_MSG_OSD_MAP                41
149 #define CEPH_MSG_OSD_OP                 42
150 #define CEPH_MSG_OSD_OPREPLY            43
151 #define CEPH_MSG_WATCH_NOTIFY           44
152 
153 
154 /* watch-notify operations */
155 enum {
156 	CEPH_WATCH_EVENT_NOTIFY		  = 1, /* notifying watcher */
157 	CEPH_WATCH_EVENT_NOTIFY_COMPLETE  = 2, /* notifier notified when done */
158 	CEPH_WATCH_EVENT_DISCONNECT       = 3, /* we were disconnected */
159 };
160 
161 
162 struct ceph_mon_request_header {
163 	__le64 have_version;
164 	__le16 session_mon;
165 	__le64 session_mon_tid;
166 } __attribute__ ((packed));
167 
168 struct ceph_mon_statfs {
169 	struct ceph_mon_request_header monhdr;
170 	struct ceph_fsid fsid;
171 } __attribute__ ((packed));
172 
173 struct ceph_statfs {
174 	__le64 kb, kb_used, kb_avail;
175 	__le64 num_objects;
176 } __attribute__ ((packed));
177 
178 struct ceph_mon_statfs_reply {
179 	struct ceph_fsid fsid;
180 	__le64 version;
181 	struct ceph_statfs st;
182 } __attribute__ ((packed));
183 
184 struct ceph_osd_getmap {
185 	struct ceph_mon_request_header monhdr;
186 	struct ceph_fsid fsid;
187 	__le32 start;
188 } __attribute__ ((packed));
189 
190 struct ceph_mds_getmap {
191 	struct ceph_mon_request_header monhdr;
192 	struct ceph_fsid fsid;
193 } __attribute__ ((packed));
194 
195 struct ceph_client_mount {
196 	struct ceph_mon_request_header monhdr;
197 } __attribute__ ((packed));
198 
199 #define CEPH_SUBSCRIBE_ONETIME    1  /* i want only 1 update after have */
200 
201 struct ceph_mon_subscribe_item {
202 	__le64 start;
203 	__u8 flags;
204 } __attribute__ ((packed));
205 
206 struct ceph_mon_subscribe_ack {
207 	__le32 duration;         /* seconds */
208 	struct ceph_fsid fsid;
209 } __attribute__ ((packed));
210 
211 #define CEPH_FS_CLUSTER_ID_NONE  -1
212 
213 /*
214  * mdsmap flags
215  */
216 #define CEPH_MDSMAP_DOWN    (1<<0)  /* cluster deliberately down */
217 
218 /*
219  * mds states
220  *   > 0 -> in
221  *  <= 0 -> out
222  */
223 #define CEPH_MDS_STATE_DNE          0  /* down, does not exist. */
224 #define CEPH_MDS_STATE_STOPPED     -1  /* down, once existed, but no subtrees.
225 					  empty log. */
226 #define CEPH_MDS_STATE_BOOT        -4  /* up, boot announcement. */
227 #define CEPH_MDS_STATE_STANDBY     -5  /* up, idle.  waiting for assignment. */
228 #define CEPH_MDS_STATE_CREATING    -6  /* up, creating MDS instance. */
229 #define CEPH_MDS_STATE_STARTING    -7  /* up, starting previously stopped mds */
230 #define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
231 #define CEPH_MDS_STATE_REPLAYONCE   -9 /* up, replaying an active node's journal */
232 
233 #define CEPH_MDS_STATE_REPLAY       8  /* up, replaying journal. */
234 #define CEPH_MDS_STATE_RESOLVE      9  /* up, disambiguating distributed
235 					  operations (import, rename, etc.) */
236 #define CEPH_MDS_STATE_RECONNECT    10 /* up, reconnect to clients */
237 #define CEPH_MDS_STATE_REJOIN       11 /* up, rejoining distributed cache */
238 #define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
239 #define CEPH_MDS_STATE_ACTIVE       13 /* up, active */
240 #define CEPH_MDS_STATE_STOPPING     14 /* up, but exporting metadata */
241 
242 extern const char *ceph_mds_state_name(int s);
243 
244 
245 /*
246  * metadata lock types.
247  *  - these are bitmasks.. we can compose them
248  *  - they also define the lock ordering by the MDS
249  *  - a few of these are internal to the mds
250  */
251 #define CEPH_LOCK_DVERSION    1
252 #define CEPH_LOCK_DN          2
253 #define CEPH_LOCK_ISNAP       16
254 #define CEPH_LOCK_IVERSION    32    /* mds internal */
255 #define CEPH_LOCK_IFILE       64
256 #define CEPH_LOCK_IAUTH       128
257 #define CEPH_LOCK_ILINK       256
258 #define CEPH_LOCK_IDFT        512   /* dir frag tree */
259 #define CEPH_LOCK_INEST       1024  /* mds internal */
260 #define CEPH_LOCK_IXATTR      2048
261 #define CEPH_LOCK_IFLOCK      4096  /* advisory file locks */
262 #define CEPH_LOCK_INO         8192  /* immutable inode bits; not a lock */
263 #define CEPH_LOCK_IPOLICY     16384 /* policy lock on dirs. MDS internal */
264 
265 /* client_session ops */
266 enum {
267 	CEPH_SESSION_REQUEST_OPEN,
268 	CEPH_SESSION_OPEN,
269 	CEPH_SESSION_REQUEST_CLOSE,
270 	CEPH_SESSION_CLOSE,
271 	CEPH_SESSION_REQUEST_RENEWCAPS,
272 	CEPH_SESSION_RENEWCAPS,
273 	CEPH_SESSION_STALE,
274 	CEPH_SESSION_RECALL_STATE,
275 	CEPH_SESSION_FLUSHMSG,
276 	CEPH_SESSION_FLUSHMSG_ACK,
277 	CEPH_SESSION_FORCE_RO,
278 };
279 
280 extern const char *ceph_session_op_name(int op);
281 
282 struct ceph_mds_session_head {
283 	__le32 op;
284 	__le64 seq;
285 	struct ceph_timespec stamp;
286 	__le32 max_caps, max_leases;
287 } __attribute__ ((packed));
288 
289 /* client_request */
290 /*
291  * metadata ops.
292  *  & 0x001000 -> write op
293  *  & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
294  &  & 0x100000 -> use weird ino/path trace
295  */
296 #define CEPH_MDS_OP_WRITE        0x001000
297 enum {
298 	CEPH_MDS_OP_LOOKUP     = 0x00100,
299 	CEPH_MDS_OP_GETATTR    = 0x00101,
300 	CEPH_MDS_OP_LOOKUPHASH = 0x00102,
301 	CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
302 	CEPH_MDS_OP_LOOKUPINO  = 0x00104,
303 	CEPH_MDS_OP_LOOKUPNAME = 0x00105,
304 
305 	CEPH_MDS_OP_SETXATTR   = 0x01105,
306 	CEPH_MDS_OP_RMXATTR    = 0x01106,
307 	CEPH_MDS_OP_SETLAYOUT  = 0x01107,
308 	CEPH_MDS_OP_SETATTR    = 0x01108,
309 	CEPH_MDS_OP_SETFILELOCK= 0x01109,
310 	CEPH_MDS_OP_GETFILELOCK= 0x00110,
311 	CEPH_MDS_OP_SETDIRLAYOUT=0x0110a,
312 
313 	CEPH_MDS_OP_MKNOD      = 0x01201,
314 	CEPH_MDS_OP_LINK       = 0x01202,
315 	CEPH_MDS_OP_UNLINK     = 0x01203,
316 	CEPH_MDS_OP_RENAME     = 0x01204,
317 	CEPH_MDS_OP_MKDIR      = 0x01220,
318 	CEPH_MDS_OP_RMDIR      = 0x01221,
319 	CEPH_MDS_OP_SYMLINK    = 0x01222,
320 
321 	CEPH_MDS_OP_CREATE     = 0x01301,
322 	CEPH_MDS_OP_OPEN       = 0x00302,
323 	CEPH_MDS_OP_READDIR    = 0x00305,
324 
325 	CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
326 	CEPH_MDS_OP_MKSNAP     = 0x01400,
327 	CEPH_MDS_OP_RMSNAP     = 0x01401,
328 	CEPH_MDS_OP_LSSNAP     = 0x00402,
329 	CEPH_MDS_OP_RENAMESNAP = 0x01403,
330 };
331 
332 extern const char *ceph_mds_op_name(int op);
333 
334 
335 #define CEPH_SETATTR_MODE   1
336 #define CEPH_SETATTR_UID    2
337 #define CEPH_SETATTR_GID    4
338 #define CEPH_SETATTR_MTIME  8
339 #define CEPH_SETATTR_ATIME 16
340 #define CEPH_SETATTR_SIZE  32
341 #define CEPH_SETATTR_CTIME 64
342 
343 /*
344  * Ceph setxattr request flags.
345  */
346 #define CEPH_XATTR_CREATE  (1 << 0)
347 #define CEPH_XATTR_REPLACE (1 << 1)
348 #define CEPH_XATTR_REMOVE  (1 << 31)
349 
350 /*
351  * readdir request flags;
352  */
353 #define CEPH_READDIR_REPLY_BITFLAGS	(1<<0)
354 
355 /*
356  * readdir reply flags.
357  */
358 #define CEPH_READDIR_FRAG_END		(1<<0)
359 #define CEPH_READDIR_FRAG_COMPLETE	(1<<8)
360 #define CEPH_READDIR_HASH_ORDER		(1<<9)
361 
362 union ceph_mds_request_args {
363 	struct {
364 		__le32 mask;                 /* CEPH_CAP_* */
365 	} __attribute__ ((packed)) getattr;
366 	struct {
367 		__le32 mode;
368 		__le32 uid;
369 		__le32 gid;
370 		struct ceph_timespec mtime;
371 		struct ceph_timespec atime;
372 		__le64 size, old_size;       /* old_size needed by truncate */
373 		__le32 mask;                 /* CEPH_SETATTR_* */
374 	} __attribute__ ((packed)) setattr;
375 	struct {
376 		__le32 frag;                 /* which dir fragment */
377 		__le32 max_entries;          /* how many dentries to grab */
378 		__le32 max_bytes;
379 		__le16 flags;
380 	} __attribute__ ((packed)) readdir;
381 	struct {
382 		__le32 mode;
383 		__le32 rdev;
384 	} __attribute__ ((packed)) mknod;
385 	struct {
386 		__le32 mode;
387 	} __attribute__ ((packed)) mkdir;
388 	struct {
389 		__le32 flags;
390 		__le32 mode;
391 		__le32 stripe_unit;          /* layout for newly created file */
392 		__le32 stripe_count;         /* ... */
393 		__le32 object_size;
394 		__le32 file_replication;
395                __le32 mask;                 /* CEPH_CAP_* */
396                __le32 old_size;
397 	} __attribute__ ((packed)) open;
398 	struct {
399 		__le32 flags;
400 	} __attribute__ ((packed)) setxattr;
401 	struct {
402 		struct ceph_file_layout layout;
403 	} __attribute__ ((packed)) setlayout;
404 	struct {
405 		__u8 rule; /* currently fcntl or flock */
406 		__u8 type; /* shared, exclusive, remove*/
407 		__le64 owner; /* owner of the lock */
408 		__le64 pid; /* process id requesting the lock */
409 		__le64 start; /* initial location to lock */
410 		__le64 length; /* num bytes to lock from start */
411 		__u8 wait; /* will caller wait for lock to become available? */
412 	} __attribute__ ((packed)) filelock_change;
413 } __attribute__ ((packed));
414 
415 #define CEPH_MDS_FLAG_REPLAY        1  /* this is a replayed op */
416 #define CEPH_MDS_FLAG_WANT_DENTRY   2  /* want dentry in reply */
417 
418 struct ceph_mds_request_head {
419 	__le64 oldest_client_tid;
420 	__le32 mdsmap_epoch;           /* on client */
421 	__le32 flags;                  /* CEPH_MDS_FLAG_* */
422 	__u8 num_retry, num_fwd;       /* count retry, fwd attempts */
423 	__le16 num_releases;           /* # include cap/lease release records */
424 	__le32 op;                     /* mds op code */
425 	__le32 caller_uid, caller_gid;
426 	__le64 ino;                    /* use this ino for openc, mkdir, mknod,
427 					  etc. (if replaying) */
428 	union ceph_mds_request_args args;
429 } __attribute__ ((packed));
430 
431 /* cap/lease release record */
432 struct ceph_mds_request_release {
433 	__le64 ino, cap_id;            /* ino and unique cap id */
434 	__le32 caps, wanted;           /* new issued, wanted */
435 	__le32 seq, issue_seq, mseq;
436 	__le32 dname_seq;              /* if releasing a dentry lease, a */
437 	__le32 dname_len;              /* string follows. */
438 } __attribute__ ((packed));
439 
440 /* client reply */
441 struct ceph_mds_reply_head {
442 	__le32 op;
443 	__le32 result;
444 	__le32 mdsmap_epoch;
445 	__u8 safe;                     /* true if committed to disk */
446 	__u8 is_dentry, is_target;     /* true if dentry, target inode records
447 					  are included with reply */
448 } __attribute__ ((packed));
449 
450 /* one for each node split */
451 struct ceph_frag_tree_split {
452 	__le32 frag;                   /* this frag splits... */
453 	__le32 by;                     /* ...by this many bits */
454 } __attribute__ ((packed));
455 
456 struct ceph_frag_tree_head {
457 	__le32 nsplits;                /* num ceph_frag_tree_split records */
458 	struct ceph_frag_tree_split splits[];
459 } __attribute__ ((packed));
460 
461 /* capability issue, for bundling with mds reply */
462 struct ceph_mds_reply_cap {
463 	__le32 caps, wanted;           /* caps issued, wanted */
464 	__le64 cap_id;
465 	__le32 seq, mseq;
466 	__le64 realm;                  /* snap realm */
467 	__u8 flags;                    /* CEPH_CAP_FLAG_* */
468 } __attribute__ ((packed));
469 
470 #define CEPH_CAP_FLAG_AUTH	(1 << 0)  /* cap is issued by auth mds */
471 #define CEPH_CAP_FLAG_RELEASE	(1 << 1)  /* release the cap */
472 
473 /* inode record, for bundling with mds reply */
474 struct ceph_mds_reply_inode {
475 	__le64 ino;
476 	__le64 snapid;
477 	__le32 rdev;
478 	__le64 version;                /* inode version */
479 	__le64 xattr_version;          /* version for xattr blob */
480 	struct ceph_mds_reply_cap cap; /* caps issued for this inode */
481 	struct ceph_file_layout layout;
482 	struct ceph_timespec ctime, mtime, atime;
483 	__le32 time_warp_seq;
484 	__le64 size, max_size, truncate_size;
485 	__le32 truncate_seq;
486 	__le32 mode, uid, gid;
487 	__le32 nlink;
488 	__le64 files, subdirs, rbytes, rfiles, rsubdirs;  /* dir stats */
489 	struct ceph_timespec rctime;
490 	struct ceph_frag_tree_head fragtree;  /* (must be at end of struct) */
491 } __attribute__ ((packed));
492 /* followed by frag array, symlink string, dir layout, xattr blob */
493 
494 /* reply_lease follows dname, and reply_inode */
495 struct ceph_mds_reply_lease {
496 	__le16 mask;            /* lease type(s) */
497 	__le32 duration_ms;     /* lease duration */
498 	__le32 seq;
499 } __attribute__ ((packed));
500 
501 struct ceph_mds_reply_dirfrag {
502 	__le32 frag;            /* fragment */
503 	__le32 auth;            /* auth mds, if this is a delegation point */
504 	__le32 ndist;           /* number of mds' this is replicated on */
505 	__le32 dist[];
506 } __attribute__ ((packed));
507 
508 #define CEPH_LOCK_FCNTL		1
509 #define CEPH_LOCK_FLOCK		2
510 #define CEPH_LOCK_FCNTL_INTR    3
511 #define CEPH_LOCK_FLOCK_INTR    4
512 
513 
514 #define CEPH_LOCK_SHARED   1
515 #define CEPH_LOCK_EXCL     2
516 #define CEPH_LOCK_UNLOCK   4
517 
518 struct ceph_filelock {
519 	__le64 start;/* file offset to start lock at */
520 	__le64 length; /* num bytes to lock; 0 for all following start */
521 	__le64 client; /* which client holds the lock */
522 	__le64 owner; /* owner the lock */
523 	__le64 pid; /* process id holding the lock on the client */
524 	__u8 type; /* shared lock, exclusive lock, or unlock */
525 } __attribute__ ((packed));
526 
527 
528 /* file access modes */
529 #define CEPH_FILE_MODE_PIN        0
530 #define CEPH_FILE_MODE_RD         1
531 #define CEPH_FILE_MODE_WR         2
532 #define CEPH_FILE_MODE_RDWR       3  /* RD | WR */
533 #define CEPH_FILE_MODE_LAZY       4  /* lazy io */
534 #define CEPH_FILE_MODE_NUM        8  /* bc these are bit fields.. mostly */
535 
536 int ceph_flags_to_mode(int flags);
537 
538 #define CEPH_INLINE_NONE	((__u64)-1)
539 
540 /* capability bits */
541 #define CEPH_CAP_PIN         1  /* no specific capabilities beyond the pin */
542 
543 /* generic cap bits */
544 #define CEPH_CAP_GSHARED     1  /* client can reads */
545 #define CEPH_CAP_GEXCL       2  /* client can read and update */
546 #define CEPH_CAP_GCACHE      4  /* (file) client can cache reads */
547 #define CEPH_CAP_GRD         8  /* (file) client can read */
548 #define CEPH_CAP_GWR        16  /* (file) client can write */
549 #define CEPH_CAP_GBUFFER    32  /* (file) client can buffer writes */
550 #define CEPH_CAP_GWREXTEND  64  /* (file) client can extend EOF */
551 #define CEPH_CAP_GLAZYIO   128  /* (file) client can perform lazy io */
552 
553 #define CEPH_CAP_SIMPLE_BITS  2
554 #define CEPH_CAP_FILE_BITS    8
555 
556 /* per-lock shift */
557 #define CEPH_CAP_SAUTH      2
558 #define CEPH_CAP_SLINK      4
559 #define CEPH_CAP_SXATTR     6
560 #define CEPH_CAP_SFILE      8
561 #define CEPH_CAP_SFLOCK    20
562 
563 #define CEPH_CAP_BITS      22
564 
565 /* composed values */
566 #define CEPH_CAP_AUTH_SHARED  (CEPH_CAP_GSHARED  << CEPH_CAP_SAUTH)
567 #define CEPH_CAP_AUTH_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SAUTH)
568 #define CEPH_CAP_LINK_SHARED  (CEPH_CAP_GSHARED  << CEPH_CAP_SLINK)
569 #define CEPH_CAP_LINK_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SLINK)
570 #define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED  << CEPH_CAP_SXATTR)
571 #define CEPH_CAP_XATTR_EXCL    (CEPH_CAP_GEXCL     << CEPH_CAP_SXATTR)
572 #define CEPH_CAP_FILE(x)    (x << CEPH_CAP_SFILE)
573 #define CEPH_CAP_FILE_SHARED   (CEPH_CAP_GSHARED   << CEPH_CAP_SFILE)
574 #define CEPH_CAP_FILE_EXCL     (CEPH_CAP_GEXCL     << CEPH_CAP_SFILE)
575 #define CEPH_CAP_FILE_CACHE    (CEPH_CAP_GCACHE    << CEPH_CAP_SFILE)
576 #define CEPH_CAP_FILE_RD       (CEPH_CAP_GRD       << CEPH_CAP_SFILE)
577 #define CEPH_CAP_FILE_WR       (CEPH_CAP_GWR       << CEPH_CAP_SFILE)
578 #define CEPH_CAP_FILE_BUFFER   (CEPH_CAP_GBUFFER   << CEPH_CAP_SFILE)
579 #define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
580 #define CEPH_CAP_FILE_LAZYIO   (CEPH_CAP_GLAZYIO   << CEPH_CAP_SFILE)
581 #define CEPH_CAP_FLOCK_SHARED  (CEPH_CAP_GSHARED   << CEPH_CAP_SFLOCK)
582 #define CEPH_CAP_FLOCK_EXCL    (CEPH_CAP_GEXCL     << CEPH_CAP_SFLOCK)
583 
584 
585 /* cap masks (for getattr) */
586 #define CEPH_STAT_CAP_INODE    CEPH_CAP_PIN
587 #define CEPH_STAT_CAP_TYPE     CEPH_CAP_PIN  /* mode >> 12 */
588 #define CEPH_STAT_CAP_SYMLINK  CEPH_CAP_PIN
589 #define CEPH_STAT_CAP_UID      CEPH_CAP_AUTH_SHARED
590 #define CEPH_STAT_CAP_GID      CEPH_CAP_AUTH_SHARED
591 #define CEPH_STAT_CAP_MODE     CEPH_CAP_AUTH_SHARED
592 #define CEPH_STAT_CAP_NLINK    CEPH_CAP_LINK_SHARED
593 #define CEPH_STAT_CAP_LAYOUT   CEPH_CAP_FILE_SHARED
594 #define CEPH_STAT_CAP_MTIME    CEPH_CAP_FILE_SHARED
595 #define CEPH_STAT_CAP_SIZE     CEPH_CAP_FILE_SHARED
596 #define CEPH_STAT_CAP_ATIME    CEPH_CAP_FILE_SHARED  /* fixme */
597 #define CEPH_STAT_CAP_XATTR    CEPH_CAP_XATTR_SHARED
598 #define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN |			\
599 				 CEPH_CAP_AUTH_SHARED |	\
600 				 CEPH_CAP_LINK_SHARED |	\
601 				 CEPH_CAP_FILE_SHARED |	\
602 				 CEPH_CAP_XATTR_SHARED)
603 #define CEPH_STAT_CAP_INLINE_DATA (CEPH_CAP_FILE_SHARED | \
604 				   CEPH_CAP_FILE_RD)
605 
606 #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED |			\
607 			      CEPH_CAP_LINK_SHARED |			\
608 			      CEPH_CAP_XATTR_SHARED |			\
609 			      CEPH_CAP_FILE_SHARED)
610 #define CEPH_CAP_ANY_RD   (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD |	\
611 			   CEPH_CAP_FILE_CACHE)
612 
613 #define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL |		\
614 			   CEPH_CAP_LINK_EXCL |		\
615 			   CEPH_CAP_XATTR_EXCL |	\
616 			   CEPH_CAP_FILE_EXCL)
617 #define CEPH_CAP_ANY_FILE_RD (CEPH_CAP_FILE_RD | CEPH_CAP_FILE_CACHE | \
618 			      CEPH_CAP_FILE_SHARED)
619 #define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER |	\
620 			      CEPH_CAP_FILE_EXCL)
621 #define CEPH_CAP_ANY_WR   (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
622 #define CEPH_CAP_ANY      (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
623 			   CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \
624 			   CEPH_CAP_PIN)
625 
626 #define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
627 			CEPH_LOCK_IXATTR)
628 
629 int ceph_caps_for_mode(int mode);
630 
631 enum {
632 	CEPH_CAP_OP_GRANT,         /* mds->client grant */
633 	CEPH_CAP_OP_REVOKE,        /* mds->client revoke */
634 	CEPH_CAP_OP_TRUNC,         /* mds->client trunc notify */
635 	CEPH_CAP_OP_EXPORT,        /* mds has exported the cap */
636 	CEPH_CAP_OP_IMPORT,        /* mds has imported the cap */
637 	CEPH_CAP_OP_UPDATE,        /* client->mds update */
638 	CEPH_CAP_OP_DROP,          /* client->mds drop cap bits */
639 	CEPH_CAP_OP_FLUSH,         /* client->mds cap writeback */
640 	CEPH_CAP_OP_FLUSH_ACK,     /* mds->client flushed */
641 	CEPH_CAP_OP_FLUSHSNAP,     /* client->mds flush snapped metadata */
642 	CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
643 	CEPH_CAP_OP_RELEASE,       /* client->mds release (clean) cap */
644 	CEPH_CAP_OP_RENEW,         /* client->mds renewal request */
645 };
646 
647 extern const char *ceph_cap_op_name(int op);
648 
649 /*
650  * caps message, used for capability callbacks, acks, requests, etc.
651  */
652 struct ceph_mds_caps {
653 	__le32 op;                  /* CEPH_CAP_OP_* */
654 	__le64 ino, realm;
655 	__le64 cap_id;
656 	__le32 seq, issue_seq;
657 	__le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
658 	__le32 migrate_seq;
659 	__le64 snap_follows;
660 	__le32 snap_trace_len;
661 
662 	/* authlock */
663 	__le32 uid, gid, mode;
664 
665 	/* linklock */
666 	__le32 nlink;
667 
668 	/* xattrlock */
669 	__le32 xattr_len;
670 	__le64 xattr_version;
671 
672 	/* filelock */
673 	__le64 size, max_size, truncate_size;
674 	__le32 truncate_seq;
675 	struct ceph_timespec mtime, atime, ctime;
676 	struct ceph_file_layout layout;
677 	__le32 time_warp_seq;
678 } __attribute__ ((packed));
679 
680 struct ceph_mds_cap_peer {
681 	__le64 cap_id;
682 	__le32 seq;
683 	__le32 mseq;
684 	__le32 mds;
685 	__u8   flags;
686 } __attribute__ ((packed));
687 
688 /* cap release msg head */
689 struct ceph_mds_cap_release {
690 	__le32 num;                /* number of cap_items that follow */
691 } __attribute__ ((packed));
692 
693 struct ceph_mds_cap_item {
694 	__le64 ino;
695 	__le64 cap_id;
696 	__le32 migrate_seq, seq;
697 } __attribute__ ((packed));
698 
699 #define CEPH_MDS_LEASE_REVOKE           1  /*    mds  -> client */
700 #define CEPH_MDS_LEASE_RELEASE          2  /* client  -> mds    */
701 #define CEPH_MDS_LEASE_RENEW            3  /* client <-> mds    */
702 #define CEPH_MDS_LEASE_REVOKE_ACK       4  /* client  -> mds    */
703 
704 extern const char *ceph_lease_op_name(int o);
705 
706 /* lease msg header */
707 struct ceph_mds_lease {
708 	__u8 action;            /* CEPH_MDS_LEASE_* */
709 	__le16 mask;            /* which lease */
710 	__le64 ino;
711 	__le64 first, last;     /* snap range */
712 	__le32 seq;
713 	__le32 duration_ms;     /* duration of renewal */
714 } __attribute__ ((packed));
715 /* followed by a __le32+string for dname */
716 
717 /* client reconnect */
718 struct ceph_mds_cap_reconnect {
719 	__le64 cap_id;
720 	__le32 wanted;
721 	__le32 issued;
722 	__le64 snaprealm;
723 	__le64 pathbase;        /* base ino for our path to this ino */
724 	__le32 flock_len;       /* size of flock state blob, if any */
725 } __attribute__ ((packed));
726 /* followed by flock blob */
727 
728 struct ceph_mds_cap_reconnect_v1 {
729 	__le64 cap_id;
730 	__le32 wanted;
731 	__le32 issued;
732 	__le64 size;
733 	struct ceph_timespec mtime, atime;
734 	__le64 snaprealm;
735 	__le64 pathbase;        /* base ino for our path to this ino */
736 } __attribute__ ((packed));
737 
738 struct ceph_mds_snaprealm_reconnect {
739 	__le64 ino;     /* snap realm base */
740 	__le64 seq;     /* snap seq for this snap realm */
741 	__le64 parent;  /* parent realm */
742 } __attribute__ ((packed));
743 
744 /*
745  * snaps
746  */
747 enum {
748 	CEPH_SNAP_OP_UPDATE,  /* CREATE or DESTROY */
749 	CEPH_SNAP_OP_CREATE,
750 	CEPH_SNAP_OP_DESTROY,
751 	CEPH_SNAP_OP_SPLIT,
752 };
753 
754 extern const char *ceph_snap_op_name(int o);
755 
756 /* snap msg header */
757 struct ceph_mds_snap_head {
758 	__le32 op;                /* CEPH_SNAP_OP_* */
759 	__le64 split;             /* ino to split off, if any */
760 	__le32 num_split_inos;    /* # inos belonging to new child realm */
761 	__le32 num_split_realms;  /* # child realms udner new child realm */
762 	__le32 trace_len;         /* size of snap trace blob */
763 } __attribute__ ((packed));
764 /* followed by split ino list, then split realms, then the trace blob */
765 
766 /*
767  * encode info about a snaprealm, as viewed by a client
768  */
769 struct ceph_mds_snap_realm {
770 	__le64 ino;           /* ino */
771 	__le64 created;       /* snap: when created */
772 	__le64 parent;        /* ino: parent realm */
773 	__le64 parent_since;  /* snap: same parent since */
774 	__le64 seq;           /* snap: version */
775 	__le32 num_snaps;
776 	__le32 num_prior_parent_snaps;
777 } __attribute__ ((packed));
778 /* followed by my snap list, then prior parent snap list */
779 
780 #endif
781