blob: 0c2241ef3653242657fc56a8c9cd7c49ea89af7b [file] [log] [blame]
Sage Weil0dee3c22009-10-06 11:31:06 -07001/*
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
Sage Weila17d6472010-02-17 13:56:07 -080022#define CEPH_VERSION_MINOR 19
Sage Weilbb097ff2009-10-19 16:17:31 -070023#define CEPH_VERSION_PATCH 0
Sage Weil0dee3c22009-10-06 11:31:06 -070024
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 */
Sage Weil6df058c2009-12-22 11:24:33 -080038#define CEPH_OSD_PROTOCOL 8 /* cluster internal */
Sage Weil0dee3c22009-10-06 11:31:06 -070039#define CEPH_MDS_PROTOCOL 9 /* cluster internal */
Sage Weil13e38c82009-10-09 16:36:34 -070040#define CEPH_MON_PROTOCOL 5 /* cluster internal */
Yehuda Sadeh0c948992010-02-01 16:10:45 -080041#define CEPH_OSDC_PROTOCOL 24 /* server/client */
Sage Weil3ea25f92010-01-25 11:18:17 -080042#define CEPH_MDSC_PROTOCOL 32 /* server/client */
Sage Weil13e38c82009-10-09 16:36:34 -070043#define CEPH_MONC_PROTOCOL 15 /* server/client */
Sage Weil0dee3c22009-10-06 11:31:06 -070044
45
46#define CEPH_INO_ROOT 1
Sage Weil1d1de9162009-12-02 11:54:25 -080047#define CEPH_INO_CEPH 2 /* hidden .ceph dir */
Sage Weil0dee3c22009-10-06 11:31:06 -070048
49/* arbitrary limit on max # of monitors (cluster of 3 is typical) */
50#define CEPH_MAX_MON 31
51
52
Sage Weil04a419f2009-12-23 09:30:21 -080053/*
54 * feature bits
55 */
56#define CEPH_FEATURE_SUPPORTED 0
57#define CEPH_FEATURE_REQUIRED 0
58
Sage Weil0dee3c22009-10-06 11:31:06 -070059
60/*
61 * ceph_file_layout - describe data layout for a file/inode
62 */
63struct ceph_file_layout {
64 /* file -> object mapping */
65 __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple
66 of page size. */
67 __le32 fl_stripe_count; /* over this many objects */
68 __le32 fl_object_size; /* until objects are this big, then move to
69 new objects */
70 __le32 fl_cas_hash; /* 0 = none; 1 = sha256 */
71
72 /* pg -> disk layout */
73 __le32 fl_object_stripe_unit; /* for per-object parity, if any */
74
75 /* object -> pg layout */
76 __le32 fl_pg_preferred; /* preferred primary for pg (-1 for none) */
77 __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */
78} __attribute__ ((packed));
79
Sage Weil752727a2009-10-09 16:38:45 -070080#define CEPH_MIN_STRIPE_UNIT 65536
Sage Weil0dee3c22009-10-06 11:31:06 -070081
Sage Weil752727a2009-10-09 16:38:45 -070082int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
Sage Weil0dee3c22009-10-06 11:31:06 -070083
84
Sage Weil4e7a5dc2009-11-18 16:19:57 -080085/* crypto algorithms */
86#define CEPH_CRYPTO_NONE 0x0
87#define CEPH_CRYPTO_AES 0x1
88
89/* security/authentication protocols */
90#define CEPH_AUTH_UNKNOWN 0x0
91#define CEPH_AUTH_NONE 0x1
92#define CEPH_AUTH_CEPHX 0x2
93
94
Sage Weil0dee3c22009-10-06 11:31:06 -070095/*********************************************
96 * message layer
97 */
98
99/*
100 * message types
101 */
102
103/* misc */
104#define CEPH_MSG_SHUTDOWN 1
105#define CEPH_MSG_PING 2
106
107/* client <-> monitor */
108#define CEPH_MSG_MON_MAP 4
109#define CEPH_MSG_MON_GET_MAP 5
Sage Weil0dee3c22009-10-06 11:31:06 -0700110#define CEPH_MSG_STATFS 13
111#define CEPH_MSG_STATFS_REPLY 14
112#define CEPH_MSG_MON_SUBSCRIBE 15
113#define CEPH_MSG_MON_SUBSCRIBE_ACK 16
Sage Weil4e7a5dc2009-11-18 16:19:57 -0800114#define CEPH_MSG_AUTH 17
115#define CEPH_MSG_AUTH_REPLY 18
Sage Weil0dee3c22009-10-06 11:31:06 -0700116
117/* client <-> mds */
Sage Weil0dee3c22009-10-06 11:31:06 -0700118#define CEPH_MSG_MDS_MAP 21
119
120#define CEPH_MSG_CLIENT_SESSION 22
121#define CEPH_MSG_CLIENT_RECONNECT 23
122
123#define CEPH_MSG_CLIENT_REQUEST 24
124#define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
125#define CEPH_MSG_CLIENT_REPLY 26
126#define CEPH_MSG_CLIENT_CAPS 0x310
127#define CEPH_MSG_CLIENT_LEASE 0x311
128#define CEPH_MSG_CLIENT_SNAP 0x312
129#define CEPH_MSG_CLIENT_CAPRELEASE 0x313
130
131/* osd */
Sage Weil0dee3c22009-10-06 11:31:06 -0700132#define CEPH_MSG_OSD_MAP 41
133#define CEPH_MSG_OSD_OP 42
134#define CEPH_MSG_OSD_OPREPLY 43
135
Sage Weil13e38c82009-10-09 16:36:34 -0700136struct ceph_mon_request_header {
137 __le64 have_version;
138 __le16 session_mon;
139 __le64 session_mon_tid;
140} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -0700141
142struct ceph_mon_statfs {
Sage Weil13e38c82009-10-09 16:36:34 -0700143 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700144 struct ceph_fsid fsid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700145} __attribute__ ((packed));
146
147struct ceph_statfs {
148 __le64 kb, kb_used, kb_avail;
149 __le64 num_objects;
150} __attribute__ ((packed));
151
152struct ceph_mon_statfs_reply {
153 struct ceph_fsid fsid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700154 __le64 version;
155 struct ceph_statfs st;
156} __attribute__ ((packed));
157
158struct ceph_osd_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700159 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700160 struct ceph_fsid fsid;
161 __le32 start;
162} __attribute__ ((packed));
163
164struct ceph_mds_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700165 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700166 struct ceph_fsid fsid;
167} __attribute__ ((packed));
168
169struct ceph_client_mount {
Sage Weil13e38c82009-10-09 16:36:34 -0700170 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700171} __attribute__ ((packed));
172
173struct ceph_mon_subscribe_item {
Sage Weil13e38c82009-10-09 16:36:34 -0700174 __le64 have_version; __le64 have;
Sage Weil0dee3c22009-10-06 11:31:06 -0700175 __u8 onetime;
176} __attribute__ ((packed));
177
Sage Weil07bd10f2009-10-14 17:26:40 -0700178struct ceph_mon_subscribe_ack {
179 __le32 duration; /* seconds */
180 struct ceph_fsid fsid;
181} __attribute__ ((packed));
182
Sage Weil0dee3c22009-10-06 11:31:06 -0700183/*
184 * mds states
185 * > 0 -> in
186 * <= 0 -> out
187 */
188#define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */
189#define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees.
190 empty log. */
191#define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */
192#define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */
193#define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */
194#define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */
195#define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
196
197#define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */
198#define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed
199 operations (import, rename, etc.) */
200#define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */
201#define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */
202#define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
203#define CEPH_MDS_STATE_ACTIVE 13 /* up, active */
204#define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */
205
206extern const char *ceph_mds_state_name(int s);
207
208
209/*
210 * metadata lock types.
211 * - these are bitmasks.. we can compose them
212 * - they also define the lock ordering by the MDS
213 * - a few of these are internal to the mds
214 */
215#define CEPH_LOCK_DN 1
216#define CEPH_LOCK_ISNAP 2
217#define CEPH_LOCK_IVERSION 4 /* mds internal */
218#define CEPH_LOCK_IFILE 8 /* mds internal */
219#define CEPH_LOCK_IAUTH 32
220#define CEPH_LOCK_ILINK 64
221#define CEPH_LOCK_IDFT 128 /* dir frag tree */
222#define CEPH_LOCK_INEST 256 /* mds internal */
223#define CEPH_LOCK_IXATTR 512
224#define CEPH_LOCK_INO 2048 /* immutable inode bits; not a lock */
225
226/* client_session ops */
227enum {
228 CEPH_SESSION_REQUEST_OPEN,
229 CEPH_SESSION_OPEN,
230 CEPH_SESSION_REQUEST_CLOSE,
231 CEPH_SESSION_CLOSE,
232 CEPH_SESSION_REQUEST_RENEWCAPS,
233 CEPH_SESSION_RENEWCAPS,
234 CEPH_SESSION_STALE,
235 CEPH_SESSION_RECALL_STATE,
236};
237
238extern const char *ceph_session_op_name(int op);
239
240struct ceph_mds_session_head {
241 __le32 op;
242 __le64 seq;
243 struct ceph_timespec stamp;
244 __le32 max_caps, max_leases;
245} __attribute__ ((packed));
246
247/* client_request */
248/*
249 * metadata ops.
250 * & 0x001000 -> write op
251 * & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
252 & & 0x100000 -> use weird ino/path trace
253 */
254#define CEPH_MDS_OP_WRITE 0x001000
255enum {
256 CEPH_MDS_OP_LOOKUP = 0x00100,
257 CEPH_MDS_OP_GETATTR = 0x00101,
258 CEPH_MDS_OP_LOOKUPHASH = 0x00102,
259 CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
260
261 CEPH_MDS_OP_SETXATTR = 0x01105,
262 CEPH_MDS_OP_RMXATTR = 0x01106,
263 CEPH_MDS_OP_SETLAYOUT = 0x01107,
264 CEPH_MDS_OP_SETATTR = 0x01108,
265
266 CEPH_MDS_OP_MKNOD = 0x01201,
267 CEPH_MDS_OP_LINK = 0x01202,
268 CEPH_MDS_OP_UNLINK = 0x01203,
269 CEPH_MDS_OP_RENAME = 0x01204,
270 CEPH_MDS_OP_MKDIR = 0x01220,
271 CEPH_MDS_OP_RMDIR = 0x01221,
272 CEPH_MDS_OP_SYMLINK = 0x01222,
273
Sage Weil3ea25f92010-01-25 11:18:17 -0800274 CEPH_MDS_OP_CREATE = 0x01301,
Sage Weil0dee3c22009-10-06 11:31:06 -0700275 CEPH_MDS_OP_OPEN = 0x00302,
276 CEPH_MDS_OP_READDIR = 0x00305,
277
278 CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
279 CEPH_MDS_OP_MKSNAP = 0x01400,
280 CEPH_MDS_OP_RMSNAP = 0x01401,
281 CEPH_MDS_OP_LSSNAP = 0x00402,
282};
283
284extern const char *ceph_mds_op_name(int op);
285
286
287#define CEPH_SETATTR_MODE 1
288#define CEPH_SETATTR_UID 2
289#define CEPH_SETATTR_GID 4
290#define CEPH_SETATTR_MTIME 8
291#define CEPH_SETATTR_ATIME 16
292#define CEPH_SETATTR_SIZE 32
293#define CEPH_SETATTR_CTIME 64
294
295union ceph_mds_request_args {
296 struct {
297 __le32 mask; /* CEPH_CAP_* */
298 } __attribute__ ((packed)) getattr;
299 struct {
300 __le32 mode;
301 __le32 uid;
302 __le32 gid;
303 struct ceph_timespec mtime;
304 struct ceph_timespec atime;
305 __le64 size, old_size; /* old_size needed by truncate */
306 __le32 mask; /* CEPH_SETATTR_* */
307 } __attribute__ ((packed)) setattr;
308 struct {
309 __le32 frag; /* which dir fragment */
310 __le32 max_entries; /* how many dentries to grab */
311 } __attribute__ ((packed)) readdir;
312 struct {
313 __le32 mode;
314 __le32 rdev;
315 } __attribute__ ((packed)) mknod;
316 struct {
317 __le32 mode;
318 } __attribute__ ((packed)) mkdir;
319 struct {
320 __le32 flags;
321 __le32 mode;
322 __le32 stripe_unit; /* layout for newly created file */
323 __le32 stripe_count; /* ... */
324 __le32 object_size;
325 __le32 file_replication;
326 __le32 preferred;
327 } __attribute__ ((packed)) open;
328 struct {
329 __le32 flags;
330 } __attribute__ ((packed)) setxattr;
331 struct {
332 struct ceph_file_layout layout;
333 } __attribute__ ((packed)) setlayout;
334} __attribute__ ((packed));
335
336#define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */
337#define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */
338
339struct ceph_mds_request_head {
Sage Weil6df058c2009-12-22 11:24:33 -0800340 __le64 oldest_client_tid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700341 __le32 mdsmap_epoch; /* on client */
342 __le32 flags; /* CEPH_MDS_FLAG_* */
343 __u8 num_retry, num_fwd; /* count retry, fwd attempts */
344 __le16 num_releases; /* # include cap/lease release records */
345 __le32 op; /* mds op code */
346 __le32 caller_uid, caller_gid;
347 __le64 ino; /* use this ino for openc, mkdir, mknod,
348 etc. (if replaying) */
349 union ceph_mds_request_args args;
350} __attribute__ ((packed));
351
352/* cap/lease release record */
353struct ceph_mds_request_release {
354 __le64 ino, cap_id; /* ino and unique cap id */
355 __le32 caps, wanted; /* new issued, wanted */
356 __le32 seq, issue_seq, mseq;
357 __le32 dname_seq; /* if releasing a dentry lease, a */
358 __le32 dname_len; /* string follows. */
359} __attribute__ ((packed));
360
361/* client reply */
362struct ceph_mds_reply_head {
Sage Weil0dee3c22009-10-06 11:31:06 -0700363 __le32 op;
364 __le32 result;
365 __le32 mdsmap_epoch;
366 __u8 safe; /* true if committed to disk */
367 __u8 is_dentry, is_target; /* true if dentry, target inode records
368 are included with reply */
369} __attribute__ ((packed));
370
371/* one for each node split */
372struct ceph_frag_tree_split {
373 __le32 frag; /* this frag splits... */
374 __le32 by; /* ...by this many bits */
375} __attribute__ ((packed));
376
377struct ceph_frag_tree_head {
378 __le32 nsplits; /* num ceph_frag_tree_split records */
379 struct ceph_frag_tree_split splits[];
380} __attribute__ ((packed));
381
382/* capability issue, for bundling with mds reply */
383struct ceph_mds_reply_cap {
384 __le32 caps, wanted; /* caps issued, wanted */
385 __le64 cap_id;
386 __le32 seq, mseq;
387 __le64 realm; /* snap realm */
388 __u8 flags; /* CEPH_CAP_FLAG_* */
389} __attribute__ ((packed));
390
391#define CEPH_CAP_FLAG_AUTH 1 /* cap is issued by auth mds */
392
393/* inode record, for bundling with mds reply */
394struct ceph_mds_reply_inode {
395 __le64 ino;
396 __le64 snapid;
397 __le32 rdev;
398 __le64 version; /* inode version */
399 __le64 xattr_version; /* version for xattr blob */
400 struct ceph_mds_reply_cap cap; /* caps issued for this inode */
401 struct ceph_file_layout layout;
402 struct ceph_timespec ctime, mtime, atime;
403 __le32 time_warp_seq;
404 __le64 size, max_size, truncate_size;
405 __le32 truncate_seq;
406 __le32 mode, uid, gid;
407 __le32 nlink;
408 __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */
409 struct ceph_timespec rctime;
410 struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */
411} __attribute__ ((packed));
412/* followed by frag array, then symlink string, then xattr blob */
413
414/* reply_lease follows dname, and reply_inode */
415struct ceph_mds_reply_lease {
416 __le16 mask; /* lease type(s) */
417 __le32 duration_ms; /* lease duration */
418 __le32 seq;
419} __attribute__ ((packed));
420
421struct ceph_mds_reply_dirfrag {
422 __le32 frag; /* fragment */
423 __le32 auth; /* auth mds, if this is a delegation point */
424 __le32 ndist; /* number of mds' this is replicated on */
425 __le32 dist[];
426} __attribute__ ((packed));
427
428/* file access modes */
429#define CEPH_FILE_MODE_PIN 0
430#define CEPH_FILE_MODE_RD 1
431#define CEPH_FILE_MODE_WR 2
432#define CEPH_FILE_MODE_RDWR 3 /* RD | WR */
433#define CEPH_FILE_MODE_LAZY 4 /* lazy io */
434#define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */
435
436int ceph_flags_to_mode(int flags);
437
438
439/* capability bits */
440#define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */
441
442/* generic cap bits */
443#define CEPH_CAP_GSHARED 1 /* client can reads */
444#define CEPH_CAP_GEXCL 2 /* client can read and update */
445#define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */
446#define CEPH_CAP_GRD 8 /* (file) client can read */
447#define CEPH_CAP_GWR 16 /* (file) client can write */
448#define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */
449#define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */
450#define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */
451
452/* per-lock shift */
453#define CEPH_CAP_SAUTH 2
454#define CEPH_CAP_SLINK 4
455#define CEPH_CAP_SXATTR 6
456#define CEPH_CAP_SFILE 8 /* goes at the end (uses >2 cap bits) */
457
458#define CEPH_CAP_BITS 16
459
460/* composed values */
461#define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH)
462#define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH)
463#define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK)
464#define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK)
465#define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR)
466#define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR)
467#define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE)
468#define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE)
469#define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE)
470#define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE)
471#define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE)
472#define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE)
473#define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE)
474#define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
475#define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE)
476
477/* cap masks (for getattr) */
478#define CEPH_STAT_CAP_INODE CEPH_CAP_PIN
479#define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */
480#define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN
481#define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED
482#define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED
483#define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED
484#define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED
485#define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED
486#define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED
487#define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED
488#define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */
489#define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED
490#define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \
491 CEPH_CAP_AUTH_SHARED | \
492 CEPH_CAP_LINK_SHARED | \
493 CEPH_CAP_FILE_SHARED | \
494 CEPH_CAP_XATTR_SHARED)
495
496#define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \
497 CEPH_CAP_LINK_SHARED | \
498 CEPH_CAP_XATTR_SHARED | \
499 CEPH_CAP_FILE_SHARED)
500#define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \
501 CEPH_CAP_FILE_CACHE)
502
503#define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \
504 CEPH_CAP_LINK_EXCL | \
505 CEPH_CAP_XATTR_EXCL | \
506 CEPH_CAP_FILE_EXCL)
507#define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \
508 CEPH_CAP_FILE_EXCL)
509#define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
510#define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
511 CEPH_CAP_ANY_FILE_WR | CEPH_CAP_PIN)
512
513#define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
514 CEPH_LOCK_IXATTR)
515
516int ceph_caps_for_mode(int mode);
517
518enum {
519 CEPH_CAP_OP_GRANT, /* mds->client grant */
520 CEPH_CAP_OP_REVOKE, /* mds->client revoke */
521 CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */
522 CEPH_CAP_OP_EXPORT, /* mds has exported the cap */
523 CEPH_CAP_OP_IMPORT, /* mds has imported the cap */
524 CEPH_CAP_OP_UPDATE, /* client->mds update */
525 CEPH_CAP_OP_DROP, /* client->mds drop cap bits */
526 CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */
527 CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */
528 CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */
529 CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
530 CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */
531 CEPH_CAP_OP_RENEW, /* client->mds renewal request */
532};
533
534extern const char *ceph_cap_op_name(int op);
535
536/*
537 * caps message, used for capability callbacks, acks, requests, etc.
538 */
539struct ceph_mds_caps {
540 __le32 op; /* CEPH_CAP_OP_* */
541 __le64 ino, realm;
542 __le64 cap_id;
543 __le32 seq, issue_seq;
544 __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
545 __le32 migrate_seq;
546 __le64 snap_follows;
547 __le32 snap_trace_len;
Sage Weil0dee3c22009-10-06 11:31:06 -0700548
549 /* authlock */
550 __le32 uid, gid, mode;
551
552 /* linklock */
553 __le32 nlink;
554
555 /* xattrlock */
556 __le32 xattr_len;
557 __le64 xattr_version;
558
559 /* filelock */
560 __le64 size, max_size, truncate_size;
561 __le32 truncate_seq;
562 struct ceph_timespec mtime, atime, ctime;
563 struct ceph_file_layout layout;
564 __le32 time_warp_seq;
565} __attribute__ ((packed));
566
567/* cap release msg head */
568struct ceph_mds_cap_release {
569 __le32 num; /* number of cap_items that follow */
570} __attribute__ ((packed));
571
572struct ceph_mds_cap_item {
573 __le64 ino;
574 __le64 cap_id;
575 __le32 migrate_seq, seq;
576} __attribute__ ((packed));
577
578#define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */
579#define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */
580#define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */
581#define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */
582
583extern const char *ceph_lease_op_name(int o);
584
585/* lease msg header */
586struct ceph_mds_lease {
587 __u8 action; /* CEPH_MDS_LEASE_* */
588 __le16 mask; /* which lease */
589 __le64 ino;
590 __le64 first, last; /* snap range */
591 __le32 seq;
592 __le32 duration_ms; /* duration of renewal */
593} __attribute__ ((packed));
594/* followed by a __le32+string for dname */
595
596/* client reconnect */
597struct ceph_mds_cap_reconnect {
598 __le64 cap_id;
599 __le32 wanted;
600 __le32 issued;
601 __le64 size;
602 struct ceph_timespec mtime, atime;
603 __le64 snaprealm;
604 __le64 pathbase; /* base ino for our path to this ino */
605} __attribute__ ((packed));
606/* followed by encoded string */
607
608struct ceph_mds_snaprealm_reconnect {
609 __le64 ino; /* snap realm base */
610 __le64 seq; /* snap seq for this snap realm */
611 __le64 parent; /* parent realm */
612} __attribute__ ((packed));
613
614/*
615 * snaps
616 */
617enum {
618 CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */
619 CEPH_SNAP_OP_CREATE,
620 CEPH_SNAP_OP_DESTROY,
621 CEPH_SNAP_OP_SPLIT,
622};
623
624extern const char *ceph_snap_op_name(int o);
625
626/* snap msg header */
627struct ceph_mds_snap_head {
628 __le32 op; /* CEPH_SNAP_OP_* */
629 __le64 split; /* ino to split off, if any */
630 __le32 num_split_inos; /* # inos belonging to new child realm */
631 __le32 num_split_realms; /* # child realms udner new child realm */
632 __le32 trace_len; /* size of snap trace blob */
633} __attribute__ ((packed));
634/* followed by split ino list, then split realms, then the trace blob */
635
636/*
637 * encode info about a snaprealm, as viewed by a client
638 */
639struct ceph_mds_snap_realm {
640 __le64 ino; /* ino */
641 __le64 created; /* snap: when created */
642 __le64 parent; /* ino: parent realm */
643 __le64 parent_since; /* snap: same parent since */
644 __le64 seq; /* snap: version */
645 __le32 num_snaps;
646 __le32 num_prior_parent_snaps;
647} __attribute__ ((packed));
648/* followed by my snap list, then prior parent snap list */
649
650#endif