blob: 1e96a9a87d8d416cbdfd97ee8069eacb55fc8fa9 [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 Weilbb097ff2009-10-19 16:17:31 -070022#define CEPH_VERSION_MINOR 17
23#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 */
38#define CEPH_OSD_PROTOCOL 7 /* cluster internal */
39#define CEPH_MDS_PROTOCOL 9 /* cluster internal */
Sage Weil13e38c82009-10-09 16:36:34 -070040#define CEPH_MON_PROTOCOL 5 /* cluster internal */
Sage Weilcfbbcd22009-11-06 16:44:05 -080041#define CEPH_OSDC_PROTOCOL 22 /* server/client */
Sage Weil0dee3c22009-10-06 11:31:06 -070042#define CEPH_MDSC_PROTOCOL 29 /* 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
47
48/* arbitrary limit on max # of monitors (cluster of 3 is typical) */
49#define CEPH_MAX_MON 31
50
51
Sage Weil0dee3c22009-10-06 11:31:06 -070052
53/*
54 * ceph_file_layout - describe data layout for a file/inode
55 */
56struct ceph_file_layout {
57 /* file -> object mapping */
58 __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple
59 of page size. */
60 __le32 fl_stripe_count; /* over this many objects */
61 __le32 fl_object_size; /* until objects are this big, then move to
62 new objects */
63 __le32 fl_cas_hash; /* 0 = none; 1 = sha256 */
64
65 /* pg -> disk layout */
66 __le32 fl_object_stripe_unit; /* for per-object parity, if any */
67
68 /* object -> pg layout */
69 __le32 fl_pg_preferred; /* preferred primary for pg (-1 for none) */
70 __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */
71} __attribute__ ((packed));
72
Sage Weil752727a2009-10-09 16:38:45 -070073#define CEPH_MIN_STRIPE_UNIT 65536
Sage Weil0dee3c22009-10-06 11:31:06 -070074
Sage Weil752727a2009-10-09 16:38:45 -070075int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
Sage Weil0dee3c22009-10-06 11:31:06 -070076
77
Sage Weil4e7a5dc2009-11-18 16:19:57 -080078/* crypto algorithms */
79#define CEPH_CRYPTO_NONE 0x0
80#define CEPH_CRYPTO_AES 0x1
81
82/* security/authentication protocols */
83#define CEPH_AUTH_UNKNOWN 0x0
84#define CEPH_AUTH_NONE 0x1
85#define CEPH_AUTH_CEPHX 0x2
86
87
Sage Weil0dee3c22009-10-06 11:31:06 -070088/*********************************************
89 * message layer
90 */
91
92/*
93 * message types
94 */
95
96/* misc */
97#define CEPH_MSG_SHUTDOWN 1
98#define CEPH_MSG_PING 2
99
100/* client <-> monitor */
101#define CEPH_MSG_MON_MAP 4
102#define CEPH_MSG_MON_GET_MAP 5
Sage Weil0dee3c22009-10-06 11:31:06 -0700103#define CEPH_MSG_STATFS 13
104#define CEPH_MSG_STATFS_REPLY 14
105#define CEPH_MSG_MON_SUBSCRIBE 15
106#define CEPH_MSG_MON_SUBSCRIBE_ACK 16
Sage Weil4e7a5dc2009-11-18 16:19:57 -0800107#define CEPH_MSG_AUTH 17
108#define CEPH_MSG_AUTH_REPLY 18
Sage Weil0dee3c22009-10-06 11:31:06 -0700109
110/* client <-> mds */
Sage Weil0dee3c22009-10-06 11:31:06 -0700111#define CEPH_MSG_MDS_MAP 21
112
113#define CEPH_MSG_CLIENT_SESSION 22
114#define CEPH_MSG_CLIENT_RECONNECT 23
115
116#define CEPH_MSG_CLIENT_REQUEST 24
117#define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
118#define CEPH_MSG_CLIENT_REPLY 26
119#define CEPH_MSG_CLIENT_CAPS 0x310
120#define CEPH_MSG_CLIENT_LEASE 0x311
121#define CEPH_MSG_CLIENT_SNAP 0x312
122#define CEPH_MSG_CLIENT_CAPRELEASE 0x313
123
124/* osd */
Sage Weil0dee3c22009-10-06 11:31:06 -0700125#define CEPH_MSG_OSD_MAP 41
126#define CEPH_MSG_OSD_OP 42
127#define CEPH_MSG_OSD_OPREPLY 43
128
Sage Weil13e38c82009-10-09 16:36:34 -0700129struct ceph_mon_request_header {
130 __le64 have_version;
131 __le16 session_mon;
132 __le64 session_mon_tid;
133} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -0700134
135struct ceph_mon_statfs {
Sage Weil13e38c82009-10-09 16:36:34 -0700136 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700137 struct ceph_fsid fsid;
138 __le64 tid;
139} __attribute__ ((packed));
140
141struct ceph_statfs {
142 __le64 kb, kb_used, kb_avail;
143 __le64 num_objects;
144} __attribute__ ((packed));
145
146struct ceph_mon_statfs_reply {
147 struct ceph_fsid fsid;
148 __le64 tid;
149 __le64 version;
150 struct ceph_statfs st;
151} __attribute__ ((packed));
152
153struct ceph_osd_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700154 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700155 struct ceph_fsid fsid;
156 __le32 start;
157} __attribute__ ((packed));
158
159struct ceph_mds_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700160 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700161 struct ceph_fsid fsid;
162} __attribute__ ((packed));
163
164struct ceph_client_mount {
Sage Weil13e38c82009-10-09 16:36:34 -0700165 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700166} __attribute__ ((packed));
167
168struct ceph_mon_subscribe_item {
Sage Weil13e38c82009-10-09 16:36:34 -0700169 __le64 have_version; __le64 have;
Sage Weil0dee3c22009-10-06 11:31:06 -0700170 __u8 onetime;
171} __attribute__ ((packed));
172
Sage Weil07bd10f2009-10-14 17:26:40 -0700173struct ceph_mon_subscribe_ack {
174 __le32 duration; /* seconds */
175 struct ceph_fsid fsid;
176} __attribute__ ((packed));
177
Sage Weil0dee3c22009-10-06 11:31:06 -0700178/*
179 * mds states
180 * > 0 -> in
181 * <= 0 -> out
182 */
183#define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */
184#define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees.
185 empty log. */
186#define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */
187#define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */
188#define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */
189#define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */
190#define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
191
192#define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */
193#define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed
194 operations (import, rename, etc.) */
195#define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */
196#define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */
197#define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
198#define CEPH_MDS_STATE_ACTIVE 13 /* up, active */
199#define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */
200
201extern const char *ceph_mds_state_name(int s);
202
203
204/*
205 * metadata lock types.
206 * - these are bitmasks.. we can compose them
207 * - they also define the lock ordering by the MDS
208 * - a few of these are internal to the mds
209 */
210#define CEPH_LOCK_DN 1
211#define CEPH_LOCK_ISNAP 2
212#define CEPH_LOCK_IVERSION 4 /* mds internal */
213#define CEPH_LOCK_IFILE 8 /* mds internal */
214#define CEPH_LOCK_IAUTH 32
215#define CEPH_LOCK_ILINK 64
216#define CEPH_LOCK_IDFT 128 /* dir frag tree */
217#define CEPH_LOCK_INEST 256 /* mds internal */
218#define CEPH_LOCK_IXATTR 512
219#define CEPH_LOCK_INO 2048 /* immutable inode bits; not a lock */
220
221/* client_session ops */
222enum {
223 CEPH_SESSION_REQUEST_OPEN,
224 CEPH_SESSION_OPEN,
225 CEPH_SESSION_REQUEST_CLOSE,
226 CEPH_SESSION_CLOSE,
227 CEPH_SESSION_REQUEST_RENEWCAPS,
228 CEPH_SESSION_RENEWCAPS,
229 CEPH_SESSION_STALE,
230 CEPH_SESSION_RECALL_STATE,
231};
232
233extern const char *ceph_session_op_name(int op);
234
235struct ceph_mds_session_head {
236 __le32 op;
237 __le64 seq;
238 struct ceph_timespec stamp;
239 __le32 max_caps, max_leases;
240} __attribute__ ((packed));
241
242/* client_request */
243/*
244 * metadata ops.
245 * & 0x001000 -> write op
246 * & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
247 & & 0x100000 -> use weird ino/path trace
248 */
249#define CEPH_MDS_OP_WRITE 0x001000
250enum {
251 CEPH_MDS_OP_LOOKUP = 0x00100,
252 CEPH_MDS_OP_GETATTR = 0x00101,
253 CEPH_MDS_OP_LOOKUPHASH = 0x00102,
254 CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
255
256 CEPH_MDS_OP_SETXATTR = 0x01105,
257 CEPH_MDS_OP_RMXATTR = 0x01106,
258 CEPH_MDS_OP_SETLAYOUT = 0x01107,
259 CEPH_MDS_OP_SETATTR = 0x01108,
260
261 CEPH_MDS_OP_MKNOD = 0x01201,
262 CEPH_MDS_OP_LINK = 0x01202,
263 CEPH_MDS_OP_UNLINK = 0x01203,
264 CEPH_MDS_OP_RENAME = 0x01204,
265 CEPH_MDS_OP_MKDIR = 0x01220,
266 CEPH_MDS_OP_RMDIR = 0x01221,
267 CEPH_MDS_OP_SYMLINK = 0x01222,
268
269 CEPH_MDS_OP_CREATE = 0x00301,
270 CEPH_MDS_OP_OPEN = 0x00302,
271 CEPH_MDS_OP_READDIR = 0x00305,
272
273 CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
274 CEPH_MDS_OP_MKSNAP = 0x01400,
275 CEPH_MDS_OP_RMSNAP = 0x01401,
276 CEPH_MDS_OP_LSSNAP = 0x00402,
277};
278
279extern const char *ceph_mds_op_name(int op);
280
281
282#define CEPH_SETATTR_MODE 1
283#define CEPH_SETATTR_UID 2
284#define CEPH_SETATTR_GID 4
285#define CEPH_SETATTR_MTIME 8
286#define CEPH_SETATTR_ATIME 16
287#define CEPH_SETATTR_SIZE 32
288#define CEPH_SETATTR_CTIME 64
289
290union ceph_mds_request_args {
291 struct {
292 __le32 mask; /* CEPH_CAP_* */
293 } __attribute__ ((packed)) getattr;
294 struct {
295 __le32 mode;
296 __le32 uid;
297 __le32 gid;
298 struct ceph_timespec mtime;
299 struct ceph_timespec atime;
300 __le64 size, old_size; /* old_size needed by truncate */
301 __le32 mask; /* CEPH_SETATTR_* */
302 } __attribute__ ((packed)) setattr;
303 struct {
304 __le32 frag; /* which dir fragment */
305 __le32 max_entries; /* how many dentries to grab */
306 } __attribute__ ((packed)) readdir;
307 struct {
308 __le32 mode;
309 __le32 rdev;
310 } __attribute__ ((packed)) mknod;
311 struct {
312 __le32 mode;
313 } __attribute__ ((packed)) mkdir;
314 struct {
315 __le32 flags;
316 __le32 mode;
317 __le32 stripe_unit; /* layout for newly created file */
318 __le32 stripe_count; /* ... */
319 __le32 object_size;
320 __le32 file_replication;
321 __le32 preferred;
322 } __attribute__ ((packed)) open;
323 struct {
324 __le32 flags;
325 } __attribute__ ((packed)) setxattr;
326 struct {
327 struct ceph_file_layout layout;
328 } __attribute__ ((packed)) setlayout;
329} __attribute__ ((packed));
330
331#define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */
332#define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */
333
334struct ceph_mds_request_head {
335 __le64 tid, oldest_client_tid;
336 __le32 mdsmap_epoch; /* on client */
337 __le32 flags; /* CEPH_MDS_FLAG_* */
338 __u8 num_retry, num_fwd; /* count retry, fwd attempts */
339 __le16 num_releases; /* # include cap/lease release records */
340 __le32 op; /* mds op code */
341 __le32 caller_uid, caller_gid;
342 __le64 ino; /* use this ino for openc, mkdir, mknod,
343 etc. (if replaying) */
344 union ceph_mds_request_args args;
345} __attribute__ ((packed));
346
347/* cap/lease release record */
348struct ceph_mds_request_release {
349 __le64 ino, cap_id; /* ino and unique cap id */
350 __le32 caps, wanted; /* new issued, wanted */
351 __le32 seq, issue_seq, mseq;
352 __le32 dname_seq; /* if releasing a dentry lease, a */
353 __le32 dname_len; /* string follows. */
354} __attribute__ ((packed));
355
356/* client reply */
357struct ceph_mds_reply_head {
358 __le64 tid;
359 __le32 op;
360 __le32 result;
361 __le32 mdsmap_epoch;
362 __u8 safe; /* true if committed to disk */
363 __u8 is_dentry, is_target; /* true if dentry, target inode records
364 are included with reply */
365} __attribute__ ((packed));
366
367/* one for each node split */
368struct ceph_frag_tree_split {
369 __le32 frag; /* this frag splits... */
370 __le32 by; /* ...by this many bits */
371} __attribute__ ((packed));
372
373struct ceph_frag_tree_head {
374 __le32 nsplits; /* num ceph_frag_tree_split records */
375 struct ceph_frag_tree_split splits[];
376} __attribute__ ((packed));
377
378/* capability issue, for bundling with mds reply */
379struct ceph_mds_reply_cap {
380 __le32 caps, wanted; /* caps issued, wanted */
381 __le64 cap_id;
382 __le32 seq, mseq;
383 __le64 realm; /* snap realm */
384 __u8 flags; /* CEPH_CAP_FLAG_* */
385} __attribute__ ((packed));
386
387#define CEPH_CAP_FLAG_AUTH 1 /* cap is issued by auth mds */
388
389/* inode record, for bundling with mds reply */
390struct ceph_mds_reply_inode {
391 __le64 ino;
392 __le64 snapid;
393 __le32 rdev;
394 __le64 version; /* inode version */
395 __le64 xattr_version; /* version for xattr blob */
396 struct ceph_mds_reply_cap cap; /* caps issued for this inode */
397 struct ceph_file_layout layout;
398 struct ceph_timespec ctime, mtime, atime;
399 __le32 time_warp_seq;
400 __le64 size, max_size, truncate_size;
401 __le32 truncate_seq;
402 __le32 mode, uid, gid;
403 __le32 nlink;
404 __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */
405 struct ceph_timespec rctime;
406 struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */
407} __attribute__ ((packed));
408/* followed by frag array, then symlink string, then xattr blob */
409
410/* reply_lease follows dname, and reply_inode */
411struct ceph_mds_reply_lease {
412 __le16 mask; /* lease type(s) */
413 __le32 duration_ms; /* lease duration */
414 __le32 seq;
415} __attribute__ ((packed));
416
417struct ceph_mds_reply_dirfrag {
418 __le32 frag; /* fragment */
419 __le32 auth; /* auth mds, if this is a delegation point */
420 __le32 ndist; /* number of mds' this is replicated on */
421 __le32 dist[];
422} __attribute__ ((packed));
423
424/* file access modes */
425#define CEPH_FILE_MODE_PIN 0
426#define CEPH_FILE_MODE_RD 1
427#define CEPH_FILE_MODE_WR 2
428#define CEPH_FILE_MODE_RDWR 3 /* RD | WR */
429#define CEPH_FILE_MODE_LAZY 4 /* lazy io */
430#define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */
431
432int ceph_flags_to_mode(int flags);
433
434
435/* capability bits */
436#define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */
437
438/* generic cap bits */
439#define CEPH_CAP_GSHARED 1 /* client can reads */
440#define CEPH_CAP_GEXCL 2 /* client can read and update */
441#define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */
442#define CEPH_CAP_GRD 8 /* (file) client can read */
443#define CEPH_CAP_GWR 16 /* (file) client can write */
444#define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */
445#define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */
446#define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */
447
448/* per-lock shift */
449#define CEPH_CAP_SAUTH 2
450#define CEPH_CAP_SLINK 4
451#define CEPH_CAP_SXATTR 6
452#define CEPH_CAP_SFILE 8 /* goes at the end (uses >2 cap bits) */
453
454#define CEPH_CAP_BITS 16
455
456/* composed values */
457#define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH)
458#define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH)
459#define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK)
460#define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK)
461#define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR)
462#define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR)
463#define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE)
464#define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE)
465#define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE)
466#define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE)
467#define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE)
468#define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE)
469#define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE)
470#define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
471#define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE)
472
473/* cap masks (for getattr) */
474#define CEPH_STAT_CAP_INODE CEPH_CAP_PIN
475#define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */
476#define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN
477#define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED
478#define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED
479#define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED
480#define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED
481#define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED
482#define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED
483#define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED
484#define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */
485#define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED
486#define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \
487 CEPH_CAP_AUTH_SHARED | \
488 CEPH_CAP_LINK_SHARED | \
489 CEPH_CAP_FILE_SHARED | \
490 CEPH_CAP_XATTR_SHARED)
491
492#define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \
493 CEPH_CAP_LINK_SHARED | \
494 CEPH_CAP_XATTR_SHARED | \
495 CEPH_CAP_FILE_SHARED)
496#define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \
497 CEPH_CAP_FILE_CACHE)
498
499#define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \
500 CEPH_CAP_LINK_EXCL | \
501 CEPH_CAP_XATTR_EXCL | \
502 CEPH_CAP_FILE_EXCL)
503#define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \
504 CEPH_CAP_FILE_EXCL)
505#define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
506#define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
507 CEPH_CAP_ANY_FILE_WR | CEPH_CAP_PIN)
508
509#define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
510 CEPH_LOCK_IXATTR)
511
512int ceph_caps_for_mode(int mode);
513
514enum {
515 CEPH_CAP_OP_GRANT, /* mds->client grant */
516 CEPH_CAP_OP_REVOKE, /* mds->client revoke */
517 CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */
518 CEPH_CAP_OP_EXPORT, /* mds has exported the cap */
519 CEPH_CAP_OP_IMPORT, /* mds has imported the cap */
520 CEPH_CAP_OP_UPDATE, /* client->mds update */
521 CEPH_CAP_OP_DROP, /* client->mds drop cap bits */
522 CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */
523 CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */
524 CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */
525 CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
526 CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */
527 CEPH_CAP_OP_RENEW, /* client->mds renewal request */
528};
529
530extern const char *ceph_cap_op_name(int op);
531
532/*
533 * caps message, used for capability callbacks, acks, requests, etc.
534 */
535struct ceph_mds_caps {
536 __le32 op; /* CEPH_CAP_OP_* */
537 __le64 ino, realm;
538 __le64 cap_id;
539 __le32 seq, issue_seq;
540 __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
541 __le32 migrate_seq;
542 __le64 snap_follows;
543 __le32 snap_trace_len;
544 __le64 client_tid; /* for FLUSH(SNAP) -> FLUSH(SNAP)_ACK */
545
546 /* authlock */
547 __le32 uid, gid, mode;
548
549 /* linklock */
550 __le32 nlink;
551
552 /* xattrlock */
553 __le32 xattr_len;
554 __le64 xattr_version;
555
556 /* filelock */
557 __le64 size, max_size, truncate_size;
558 __le32 truncate_seq;
559 struct ceph_timespec mtime, atime, ctime;
560 struct ceph_file_layout layout;
561 __le32 time_warp_seq;
562} __attribute__ ((packed));
563
564/* cap release msg head */
565struct ceph_mds_cap_release {
566 __le32 num; /* number of cap_items that follow */
567} __attribute__ ((packed));
568
569struct ceph_mds_cap_item {
570 __le64 ino;
571 __le64 cap_id;
572 __le32 migrate_seq, seq;
573} __attribute__ ((packed));
574
575#define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */
576#define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */
577#define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */
578#define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */
579
580extern const char *ceph_lease_op_name(int o);
581
582/* lease msg header */
583struct ceph_mds_lease {
584 __u8 action; /* CEPH_MDS_LEASE_* */
585 __le16 mask; /* which lease */
586 __le64 ino;
587 __le64 first, last; /* snap range */
588 __le32 seq;
589 __le32 duration_ms; /* duration of renewal */
590} __attribute__ ((packed));
591/* followed by a __le32+string for dname */
592
593/* client reconnect */
594struct ceph_mds_cap_reconnect {
595 __le64 cap_id;
596 __le32 wanted;
597 __le32 issued;
598 __le64 size;
599 struct ceph_timespec mtime, atime;
600 __le64 snaprealm;
601 __le64 pathbase; /* base ino for our path to this ino */
602} __attribute__ ((packed));
603/* followed by encoded string */
604
605struct ceph_mds_snaprealm_reconnect {
606 __le64 ino; /* snap realm base */
607 __le64 seq; /* snap seq for this snap realm */
608 __le64 parent; /* parent realm */
609} __attribute__ ((packed));
610
611/*
612 * snaps
613 */
614enum {
615 CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */
616 CEPH_SNAP_OP_CREATE,
617 CEPH_SNAP_OP_DESTROY,
618 CEPH_SNAP_OP_SPLIT,
619};
620
621extern const char *ceph_snap_op_name(int o);
622
623/* snap msg header */
624struct ceph_mds_snap_head {
625 __le32 op; /* CEPH_SNAP_OP_* */
626 __le64 split; /* ino to split off, if any */
627 __le32 num_split_inos; /* # inos belonging to new child realm */
628 __le32 num_split_realms; /* # child realms udner new child realm */
629 __le32 trace_len; /* size of snap trace blob */
630} __attribute__ ((packed));
631/* followed by split ino list, then split realms, then the trace blob */
632
633/*
634 * encode info about a snaprealm, as viewed by a client
635 */
636struct ceph_mds_snap_realm {
637 __le64 ino; /* ino */
638 __le64 created; /* snap: when created */
639 __le64 parent; /* ino: parent realm */
640 __le64 parent_since; /* snap: same parent since */
641 __le64 seq; /* snap: version */
642 __le32 num_snaps;
643 __le32 num_prior_parent_snaps;
644} __attribute__ ((packed));
645/* followed by my snap list, then prior parent snap list */
646
647#endif