diff options
Diffstat (limited to 'include/linux/ceph/ceph_fs.h')
-rw-r--r-- | include/linux/ceph/ceph_fs.h | 751 |
1 files changed, 751 insertions, 0 deletions
diff --git a/include/linux/ceph/ceph_fs.h b/include/linux/ceph/ceph_fs.h new file mode 100644 index 000000000000..b8c60694b2b0 --- /dev/null +++ b/include/linux/ceph/ceph_fs.h | |||
@@ -0,0 +1,751 @@ | |||
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 "msgr.h" | ||
16 | #include "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_OSD_PROTOCOL 8 /* cluster internal */ | ||
25 | #define CEPH_MDS_PROTOCOL 12 /* cluster internal */ | ||
26 | #define CEPH_MON_PROTOCOL 5 /* cluster internal */ | ||
27 | #define CEPH_OSDC_PROTOCOL 24 /* server/client */ | ||
28 | #define CEPH_MDSC_PROTOCOL 32 /* server/client */ | ||
29 | #define CEPH_MONC_PROTOCOL 15 /* server/client */ | ||
30 | |||
31 | |||
32 | #define CEPH_INO_ROOT 1 | ||
33 | #define CEPH_INO_CEPH 2 /* hidden .ceph dir */ | ||
34 | |||
35 | /* arbitrary limit on max # of monitors (cluster of 3 is typical) */ | ||
36 | #define CEPH_MAX_MON 31 | ||
37 | |||
38 | |||
39 | /* | ||
40 | * feature bits | ||
41 | */ | ||
42 | #define CEPH_FEATURE_UID (1<<0) | ||
43 | #define CEPH_FEATURE_NOSRCADDR (1<<1) | ||
44 | #define CEPH_FEATURE_MONCLOCKCHECK (1<<2) | ||
45 | #define CEPH_FEATURE_FLOCK (1<<3) | ||
46 | #define CEPH_FEATURE_SUBSCRIBE2 (1<<4) | ||
47 | #define CEPH_FEATURE_MONNAMES (1<<5) | ||
48 | #define CEPH_FEATURE_RECONNECT_SEQ (1<<6) | ||
49 | #define CEPH_FEATURE_DIRLAYOUTHASH (1<<7) | ||
50 | |||
51 | |||
52 | /* | ||
53 | * ceph_file_layout - describe data layout for a file/inode | ||
54 | */ | ||
55 | struct ceph_file_layout { | ||
56 | /* file -> object mapping */ | ||
57 | __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple | ||
58 | of page size. */ | ||
59 | __le32 fl_stripe_count; /* over this many objects */ | ||
60 | __le32 fl_object_size; /* until objects are this big, then move to | ||
61 | new objects */ | ||
62 | __le32 fl_cas_hash; /* UNUSED. 0 = none; 1 = sha256 */ | ||
63 | |||
64 | /* pg -> disk layout */ | ||
65 | __le32 fl_object_stripe_unit; /* UNUSED. for per-object parity, if any */ | ||
66 | |||
67 | /* object -> pg layout */ | ||
68 | __le32 fl_pg_preferred; /* preferred primary for pg (-1 for none) */ | ||
69 | __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */ | ||
70 | } __attribute__ ((packed)); | ||
71 | |||
72 | #define CEPH_MIN_STRIPE_UNIT 65536 | ||
73 | |||
74 | int ceph_file_layout_is_valid(const struct ceph_file_layout *layout); | ||
75 | |||
76 | struct ceph_dir_layout { | ||
77 | __u8 dl_dir_hash; /* see ceph_hash.h for ids */ | ||
78 | __u8 dl_unused1; | ||
79 | __u16 dl_unused2; | ||
80 | __u32 dl_unused3; | ||
81 | } __attribute__ ((packed)); | ||
82 | |||
83 | /* crypto algorithms */ | ||
84 | #define CEPH_CRYPTO_NONE 0x0 | ||
85 | #define CEPH_CRYPTO_AES 0x1 | ||
86 | |||
87 | #define CEPH_AES_IV "cephsageyudagreg" | ||
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 | #define CEPH_AUTH_UID_DEFAULT ((__u64) -1) | ||
95 | |||
96 | |||
97 | /********************************************* | ||
98 | * message layer | ||
99 | */ | ||
100 | |||
101 | /* | ||
102 | * message types | ||
103 | */ | ||
104 | |||
105 | /* misc */ | ||
106 | #define CEPH_MSG_SHUTDOWN 1 | ||
107 | #define CEPH_MSG_PING 2 | ||
108 | |||
109 | /* client <-> monitor */ | ||
110 | #define CEPH_MSG_MON_MAP 4 | ||
111 | #define CEPH_MSG_MON_GET_MAP 5 | ||
112 | #define CEPH_MSG_STATFS 13 | ||
113 | #define CEPH_MSG_STATFS_REPLY 14 | ||
114 | #define CEPH_MSG_MON_SUBSCRIBE 15 | ||
115 | #define CEPH_MSG_MON_SUBSCRIBE_ACK 16 | ||
116 | #define CEPH_MSG_AUTH 17 | ||
117 | #define CEPH_MSG_AUTH_REPLY 18 | ||
118 | |||
119 | /* client <-> mds */ | ||
120 | #define CEPH_MSG_MDS_MAP 21 | ||
121 | |||
122 | #define CEPH_MSG_CLIENT_SESSION 22 | ||
123 | #define CEPH_MSG_CLIENT_RECONNECT 23 | ||
124 | |||
125 | #define CEPH_MSG_CLIENT_REQUEST 24 | ||
126 | #define CEPH_MSG_CLIENT_REQUEST_FORWARD 25 | ||
127 | #define CEPH_MSG_CLIENT_REPLY 26 | ||
128 | #define CEPH_MSG_CLIENT_CAPS 0x310 | ||
129 | #define CEPH_MSG_CLIENT_LEASE 0x311 | ||
130 | #define CEPH_MSG_CLIENT_SNAP 0x312 | ||
131 | #define CEPH_MSG_CLIENT_CAPRELEASE 0x313 | ||
132 | |||
133 | /* pool ops */ | ||
134 | #define CEPH_MSG_POOLOP_REPLY 48 | ||
135 | #define CEPH_MSG_POOLOP 49 | ||
136 | |||
137 | |||
138 | /* osd */ | ||
139 | #define CEPH_MSG_OSD_MAP 41 | ||
140 | #define CEPH_MSG_OSD_OP 42 | ||
141 | #define CEPH_MSG_OSD_OPREPLY 43 | ||
142 | #define CEPH_MSG_WATCH_NOTIFY 44 | ||
143 | |||
144 | |||
145 | /* watch-notify operations */ | ||
146 | enum { | ||
147 | WATCH_NOTIFY = 1, /* notifying watcher */ | ||
148 | WATCH_NOTIFY_COMPLETE = 2, /* notifier notified when done */ | ||
149 | }; | ||
150 | |||
151 | |||
152 | /* pool operations */ | ||
153 | enum { | ||
154 | POOL_OP_CREATE = 0x01, | ||
155 | POOL_OP_DELETE = 0x02, | ||
156 | POOL_OP_AUID_CHANGE = 0x03, | ||
157 | POOL_OP_CREATE_SNAP = 0x11, | ||
158 | POOL_OP_DELETE_SNAP = 0x12, | ||
159 | POOL_OP_CREATE_UNMANAGED_SNAP = 0x21, | ||
160 | POOL_OP_DELETE_UNMANAGED_SNAP = 0x22, | ||
161 | }; | ||
162 | |||
163 | struct ceph_mon_request_header { | ||
164 | __le64 have_version; | ||
165 | __le16 session_mon; | ||
166 | __le64 session_mon_tid; | ||
167 | } __attribute__ ((packed)); | ||
168 | |||
169 | struct ceph_mon_statfs { | ||
170 | struct ceph_mon_request_header monhdr; | ||
171 | struct ceph_fsid fsid; | ||
172 | } __attribute__ ((packed)); | ||
173 | |||
174 | struct ceph_statfs { | ||
175 | __le64 kb, kb_used, kb_avail; | ||
176 | __le64 num_objects; | ||
177 | } __attribute__ ((packed)); | ||
178 | |||
179 | struct ceph_mon_statfs_reply { | ||
180 | struct ceph_fsid fsid; | ||
181 | __le64 version; | ||
182 | struct ceph_statfs st; | ||
183 | } __attribute__ ((packed)); | ||
184 | |||
185 | const char *ceph_pool_op_name(int op); | ||
186 | |||
187 | struct ceph_mon_poolop { | ||
188 | struct ceph_mon_request_header monhdr; | ||
189 | struct ceph_fsid fsid; | ||
190 | __le32 pool; | ||
191 | __le32 op; | ||
192 | __le64 auid; | ||
193 | __le64 snapid; | ||
194 | __le32 name_len; | ||
195 | } __attribute__ ((packed)); | ||
196 | |||
197 | struct ceph_mon_poolop_reply { | ||
198 | struct ceph_mon_request_header monhdr; | ||
199 | struct ceph_fsid fsid; | ||
200 | __le32 reply_code; | ||
201 | __le32 epoch; | ||
202 | char has_data; | ||
203 | char data[0]; | ||
204 | } __attribute__ ((packed)); | ||
205 | |||
206 | struct ceph_mon_unmanaged_snap { | ||
207 | __le64 snapid; | ||
208 | } __attribute__ ((packed)); | ||
209 | |||
210 | struct ceph_osd_getmap { | ||
211 | struct ceph_mon_request_header monhdr; | ||
212 | struct ceph_fsid fsid; | ||
213 | __le32 start; | ||
214 | } __attribute__ ((packed)); | ||
215 | |||
216 | struct ceph_mds_getmap { | ||
217 | struct ceph_mon_request_header monhdr; | ||
218 | struct ceph_fsid fsid; | ||
219 | } __attribute__ ((packed)); | ||
220 | |||
221 | struct ceph_client_mount { | ||
222 | struct ceph_mon_request_header monhdr; | ||
223 | } __attribute__ ((packed)); | ||
224 | |||
225 | #define CEPH_SUBSCRIBE_ONETIME 1 /* i want only 1 update after have */ | ||
226 | |||
227 | struct ceph_mon_subscribe_item { | ||
228 | __le64 have_version; __le64 have; | ||
229 | __u8 onetime; | ||
230 | } __attribute__ ((packed)); | ||
231 | |||
232 | struct ceph_mon_subscribe_ack { | ||
233 | __le32 duration; /* seconds */ | ||
234 | struct ceph_fsid fsid; | ||
235 | } __attribute__ ((packed)); | ||
236 | |||
237 | /* | ||
238 | * mds states | ||
239 | * > 0 -> in | ||
240 | * <= 0 -> out | ||
241 | */ | ||
242 | #define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */ | ||
243 | #define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees. | ||
244 | empty log. */ | ||
245 | #define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */ | ||
246 | #define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */ | ||
247 | #define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */ | ||
248 | #define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */ | ||
249 | #define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */ | ||
250 | |||
251 | #define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */ | ||
252 | #define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed | ||
253 | operations (import, rename, etc.) */ | ||
254 | #define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */ | ||
255 | #define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */ | ||
256 | #define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */ | ||
257 | #define CEPH_MDS_STATE_ACTIVE 13 /* up, active */ | ||
258 | #define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */ | ||
259 | |||
260 | extern const char *ceph_mds_state_name(int s); | ||
261 | |||
262 | |||
263 | /* | ||
264 | * metadata lock types. | ||
265 | * - these are bitmasks.. we can compose them | ||
266 | * - they also define the lock ordering by the MDS | ||
267 | * - a few of these are internal to the mds | ||
268 | */ | ||
269 | #define CEPH_LOCK_DVERSION 1 | ||
270 | #define CEPH_LOCK_DN 2 | ||
271 | #define CEPH_LOCK_ISNAP 16 | ||
272 | #define CEPH_LOCK_IVERSION 32 /* mds internal */ | ||
273 | #define CEPH_LOCK_IFILE 64 | ||
274 | #define CEPH_LOCK_IAUTH 128 | ||
275 | #define CEPH_LOCK_ILINK 256 | ||
276 | #define CEPH_LOCK_IDFT 512 /* dir frag tree */ | ||
277 | #define CEPH_LOCK_INEST 1024 /* mds internal */ | ||
278 | #define CEPH_LOCK_IXATTR 2048 | ||
279 | #define CEPH_LOCK_IFLOCK 4096 /* advisory file locks */ | ||
280 | #define CEPH_LOCK_INO 8192 /* immutable inode bits; not a lock */ | ||
281 | |||
282 | /* client_session ops */ | ||
283 | enum { | ||
284 | CEPH_SESSION_REQUEST_OPEN, | ||
285 | CEPH_SESSION_OPEN, | ||
286 | CEPH_SESSION_REQUEST_CLOSE, | ||
287 | CEPH_SESSION_CLOSE, | ||
288 | CEPH_SESSION_REQUEST_RENEWCAPS, | ||
289 | CEPH_SESSION_RENEWCAPS, | ||
290 | CEPH_SESSION_STALE, | ||
291 | CEPH_SESSION_RECALL_STATE, | ||
292 | }; | ||
293 | |||
294 | extern const char *ceph_session_op_name(int op); | ||
295 | |||
296 | struct ceph_mds_session_head { | ||
297 | __le32 op; | ||
298 | __le64 seq; | ||
299 | struct ceph_timespec stamp; | ||
300 | __le32 max_caps, max_leases; | ||
301 | } __attribute__ ((packed)); | ||
302 | |||
303 | /* client_request */ | ||
304 | /* | ||
305 | * metadata ops. | ||
306 | * & 0x001000 -> write op | ||
307 | * & 0x010000 -> follow symlink (e.g. stat(), not lstat()). | ||
308 | & & 0x100000 -> use weird ino/path trace | ||
309 | */ | ||
310 | #define CEPH_MDS_OP_WRITE 0x001000 | ||
311 | enum { | ||
312 | CEPH_MDS_OP_LOOKUP = 0x00100, | ||
313 | CEPH_MDS_OP_GETATTR = 0x00101, | ||
314 | CEPH_MDS_OP_LOOKUPHASH = 0x00102, | ||
315 | CEPH_MDS_OP_LOOKUPPARENT = 0x00103, | ||
316 | CEPH_MDS_OP_LOOKUPINO = 0x00104, | ||
317 | |||
318 | CEPH_MDS_OP_SETXATTR = 0x01105, | ||
319 | CEPH_MDS_OP_RMXATTR = 0x01106, | ||
320 | CEPH_MDS_OP_SETLAYOUT = 0x01107, | ||
321 | CEPH_MDS_OP_SETATTR = 0x01108, | ||
322 | CEPH_MDS_OP_SETFILELOCK= 0x01109, | ||
323 | CEPH_MDS_OP_GETFILELOCK= 0x00110, | ||
324 | CEPH_MDS_OP_SETDIRLAYOUT=0x0110a, | ||
325 | |||
326 | CEPH_MDS_OP_MKNOD = 0x01201, | ||
327 | CEPH_MDS_OP_LINK = 0x01202, | ||
328 | CEPH_MDS_OP_UNLINK = 0x01203, | ||
329 | CEPH_MDS_OP_RENAME = 0x01204, | ||
330 | CEPH_MDS_OP_MKDIR = 0x01220, | ||
331 | CEPH_MDS_OP_RMDIR = 0x01221, | ||
332 | CEPH_MDS_OP_SYMLINK = 0x01222, | ||
333 | |||
334 | CEPH_MDS_OP_CREATE = 0x01301, | ||
335 | CEPH_MDS_OP_OPEN = 0x00302, | ||
336 | CEPH_MDS_OP_READDIR = 0x00305, | ||
337 | |||
338 | CEPH_MDS_OP_LOOKUPSNAP = 0x00400, | ||
339 | CEPH_MDS_OP_MKSNAP = 0x01400, | ||
340 | CEPH_MDS_OP_RMSNAP = 0x01401, | ||
341 | CEPH_MDS_OP_LSSNAP = 0x00402, | ||
342 | }; | ||
343 | |||
344 | extern const char *ceph_mds_op_name(int op); | ||
345 | |||
346 | |||
347 | #define CEPH_SETATTR_MODE 1 | ||
348 | #define CEPH_SETATTR_UID 2 | ||
349 | #define CEPH_SETATTR_GID 4 | ||
350 | #define CEPH_SETATTR_MTIME 8 | ||
351 | #define CEPH_SETATTR_ATIME 16 | ||
352 | #define CEPH_SETATTR_SIZE 32 | ||
353 | #define CEPH_SETATTR_CTIME 64 | ||
354 | |||
355 | union ceph_mds_request_args { | ||
356 | struct { | ||
357 | __le32 mask; /* CEPH_CAP_* */ | ||
358 | } __attribute__ ((packed)) getattr; | ||
359 | struct { | ||
360 | __le32 mode; | ||
361 | __le32 uid; | ||
362 | __le32 gid; | ||
363 | struct ceph_timespec mtime; | ||
364 | struct ceph_timespec atime; | ||
365 | __le64 size, old_size; /* old_size needed by truncate */ | ||
366 | __le32 mask; /* CEPH_SETATTR_* */ | ||
367 | } __attribute__ ((packed)) setattr; | ||
368 | struct { | ||
369 | __le32 frag; /* which dir fragment */ | ||
370 | __le32 max_entries; /* how many dentries to grab */ | ||
371 | __le32 max_bytes; | ||
372 | } __attribute__ ((packed)) readdir; | ||
373 | struct { | ||
374 | __le32 mode; | ||
375 | __le32 rdev; | ||
376 | } __attribute__ ((packed)) mknod; | ||
377 | struct { | ||
378 | __le32 mode; | ||
379 | } __attribute__ ((packed)) mkdir; | ||
380 | struct { | ||
381 | __le32 flags; | ||
382 | __le32 mode; | ||
383 | __le32 stripe_unit; /* layout for newly created file */ | ||
384 | __le32 stripe_count; /* ... */ | ||
385 | __le32 object_size; | ||
386 | __le32 file_replication; | ||
387 | __le32 preferred; | ||
388 | } __attribute__ ((packed)) open; | ||
389 | struct { | ||
390 | __le32 flags; | ||
391 | } __attribute__ ((packed)) setxattr; | ||
392 | struct { | ||
393 | struct ceph_file_layout layout; | ||
394 | } __attribute__ ((packed)) setlayout; | ||
395 | struct { | ||
396 | __u8 rule; /* currently fcntl or flock */ | ||
397 | __u8 type; /* shared, exclusive, remove*/ | ||
398 | __le64 pid; /* process id requesting the lock */ | ||
399 | __le64 pid_namespace; | ||
400 | __le64 start; /* initial location to lock */ | ||
401 | __le64 length; /* num bytes to lock from start */ | ||
402 | __u8 wait; /* will caller wait for lock to become available? */ | ||
403 | } __attribute__ ((packed)) filelock_change; | ||
404 | } __attribute__ ((packed)); | ||
405 | |||
406 | #define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */ | ||
407 | #define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */ | ||
408 | |||
409 | struct ceph_mds_request_head { | ||
410 | __le64 oldest_client_tid; | ||
411 | __le32 mdsmap_epoch; /* on client */ | ||
412 | __le32 flags; /* CEPH_MDS_FLAG_* */ | ||
413 | __u8 num_retry, num_fwd; /* count retry, fwd attempts */ | ||
414 | __le16 num_releases; /* # include cap/lease release records */ | ||
415 | __le32 op; /* mds op code */ | ||
416 | __le32 caller_uid, caller_gid; | ||
417 | __le64 ino; /* use this ino for openc, mkdir, mknod, | ||
418 | etc. (if replaying) */ | ||
419 | union ceph_mds_request_args args; | ||
420 | } __attribute__ ((packed)); | ||
421 | |||
422 | /* cap/lease release record */ | ||
423 | struct ceph_mds_request_release { | ||
424 | __le64 ino, cap_id; /* ino and unique cap id */ | ||
425 | __le32 caps, wanted; /* new issued, wanted */ | ||
426 | __le32 seq, issue_seq, mseq; | ||
427 | __le32 dname_seq; /* if releasing a dentry lease, a */ | ||
428 | __le32 dname_len; /* string follows. */ | ||
429 | } __attribute__ ((packed)); | ||
430 | |||
431 | /* client reply */ | ||
432 | struct ceph_mds_reply_head { | ||
433 | __le32 op; | ||
434 | __le32 result; | ||
435 | __le32 mdsmap_epoch; | ||
436 | __u8 safe; /* true if committed to disk */ | ||
437 | __u8 is_dentry, is_target; /* true if dentry, target inode records | ||
438 | are included with reply */ | ||
439 | } __attribute__ ((packed)); | ||
440 | |||
441 | /* one for each node split */ | ||
442 | struct ceph_frag_tree_split { | ||
443 | __le32 frag; /* this frag splits... */ | ||
444 | __le32 by; /* ...by this many bits */ | ||
445 | } __attribute__ ((packed)); | ||
446 | |||
447 | struct ceph_frag_tree_head { | ||
448 | __le32 nsplits; /* num ceph_frag_tree_split records */ | ||
449 | struct ceph_frag_tree_split splits[]; | ||
450 | } __attribute__ ((packed)); | ||
451 | |||
452 | /* capability issue, for bundling with mds reply */ | ||
453 | struct ceph_mds_reply_cap { | ||
454 | __le32 caps, wanted; /* caps issued, wanted */ | ||
455 | __le64 cap_id; | ||
456 | __le32 seq, mseq; | ||
457 | __le64 realm; /* snap realm */ | ||
458 | __u8 flags; /* CEPH_CAP_FLAG_* */ | ||
459 | } __attribute__ ((packed)); | ||
460 | |||
461 | #define CEPH_CAP_FLAG_AUTH 1 /* cap is issued by auth mds */ | ||
462 | |||
463 | /* inode record, for bundling with mds reply */ | ||
464 | struct ceph_mds_reply_inode { | ||
465 | __le64 ino; | ||
466 | __le64 snapid; | ||
467 | __le32 rdev; | ||
468 | __le64 version; /* inode version */ | ||
469 | __le64 xattr_version; /* version for xattr blob */ | ||
470 | struct ceph_mds_reply_cap cap; /* caps issued for this inode */ | ||
471 | struct ceph_file_layout layout; | ||
472 | struct ceph_timespec ctime, mtime, atime; | ||
473 | __le32 time_warp_seq; | ||
474 | __le64 size, max_size, truncate_size; | ||
475 | __le32 truncate_seq; | ||
476 | __le32 mode, uid, gid; | ||
477 | __le32 nlink; | ||
478 | __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */ | ||
479 | struct ceph_timespec rctime; | ||
480 | struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */ | ||
481 | } __attribute__ ((packed)); | ||
482 | /* followed by frag array, symlink string, dir layout, xattr blob */ | ||
483 | |||
484 | /* reply_lease follows dname, and reply_inode */ | ||
485 | struct ceph_mds_reply_lease { | ||
486 | __le16 mask; /* lease type(s) */ | ||
487 | __le32 duration_ms; /* lease duration */ | ||
488 | __le32 seq; | ||
489 | } __attribute__ ((packed)); | ||
490 | |||
491 | struct ceph_mds_reply_dirfrag { | ||
492 | __le32 frag; /* fragment */ | ||
493 | __le32 auth; /* auth mds, if this is a delegation point */ | ||
494 | __le32 ndist; /* number of mds' this is replicated on */ | ||
495 | __le32 dist[]; | ||
496 | } __attribute__ ((packed)); | ||
497 | |||
498 | #define CEPH_LOCK_FCNTL 1 | ||
499 | #define CEPH_LOCK_FLOCK 2 | ||
500 | |||
501 | #define CEPH_LOCK_SHARED 1 | ||
502 | #define CEPH_LOCK_EXCL 2 | ||
503 | #define CEPH_LOCK_UNLOCK 4 | ||
504 | |||
505 | struct ceph_filelock { | ||
506 | __le64 start;/* file offset to start lock at */ | ||
507 | __le64 length; /* num bytes to lock; 0 for all following start */ | ||
508 | __le64 client; /* which client holds the lock */ | ||
509 | __le64 pid; /* process id holding the lock on the client */ | ||
510 | __le64 pid_namespace; | ||
511 | __u8 type; /* shared lock, exclusive lock, or unlock */ | ||
512 | } __attribute__ ((packed)); | ||
513 | |||
514 | |||
515 | /* file access modes */ | ||
516 | #define CEPH_FILE_MODE_PIN 0 | ||
517 | #define CEPH_FILE_MODE_RD 1 | ||
518 | #define CEPH_FILE_MODE_WR 2 | ||
519 | #define CEPH_FILE_MODE_RDWR 3 /* RD | WR */ | ||
520 | #define CEPH_FILE_MODE_LAZY 4 /* lazy io */ | ||
521 | #define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */ | ||
522 | |||
523 | int ceph_flags_to_mode(int flags); | ||
524 | |||
525 | |||
526 | /* capability bits */ | ||
527 | #define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */ | ||
528 | |||
529 | /* generic cap bits */ | ||
530 | #define CEPH_CAP_GSHARED 1 /* client can reads */ | ||
531 | #define CEPH_CAP_GEXCL 2 /* client can read and update */ | ||
532 | #define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */ | ||
533 | #define CEPH_CAP_GRD 8 /* (file) client can read */ | ||
534 | #define CEPH_CAP_GWR 16 /* (file) client can write */ | ||
535 | #define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */ | ||
536 | #define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */ | ||
537 | #define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */ | ||
538 | |||
539 | /* per-lock shift */ | ||
540 | #define CEPH_CAP_SAUTH 2 | ||
541 | #define CEPH_CAP_SLINK 4 | ||
542 | #define CEPH_CAP_SXATTR 6 | ||
543 | #define CEPH_CAP_SFILE 8 | ||
544 | #define CEPH_CAP_SFLOCK 20 | ||
545 | |||
546 | #define CEPH_CAP_BITS 22 | ||
547 | |||
548 | /* composed values */ | ||
549 | #define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH) | ||
550 | #define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH) | ||
551 | #define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK) | ||
552 | #define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK) | ||
553 | #define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR) | ||
554 | #define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR) | ||
555 | #define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE) | ||
556 | #define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE) | ||
557 | #define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE) | ||
558 | #define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE) | ||
559 | #define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE) | ||
560 | #define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE) | ||
561 | #define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE) | ||
562 | #define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE) | ||
563 | #define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE) | ||
564 | #define CEPH_CAP_FLOCK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFLOCK) | ||
565 | #define CEPH_CAP_FLOCK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFLOCK) | ||
566 | |||
567 | |||
568 | /* cap masks (for getattr) */ | ||
569 | #define CEPH_STAT_CAP_INODE CEPH_CAP_PIN | ||
570 | #define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */ | ||
571 | #define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN | ||
572 | #define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED | ||
573 | #define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED | ||
574 | #define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED | ||
575 | #define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED | ||
576 | #define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED | ||
577 | #define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED | ||
578 | #define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED | ||
579 | #define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */ | ||
580 | #define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED | ||
581 | #define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \ | ||
582 | CEPH_CAP_AUTH_SHARED | \ | ||
583 | CEPH_CAP_LINK_SHARED | \ | ||
584 | CEPH_CAP_FILE_SHARED | \ | ||
585 | CEPH_CAP_XATTR_SHARED) | ||
586 | |||
587 | #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \ | ||
588 | CEPH_CAP_LINK_SHARED | \ | ||
589 | CEPH_CAP_XATTR_SHARED | \ | ||
590 | CEPH_CAP_FILE_SHARED) | ||
591 | #define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \ | ||
592 | CEPH_CAP_FILE_CACHE) | ||
593 | |||
594 | #define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \ | ||
595 | CEPH_CAP_LINK_EXCL | \ | ||
596 | CEPH_CAP_XATTR_EXCL | \ | ||
597 | CEPH_CAP_FILE_EXCL) | ||
598 | #define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \ | ||
599 | CEPH_CAP_FILE_EXCL) | ||
600 | #define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR) | ||
601 | #define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \ | ||
602 | CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \ | ||
603 | CEPH_CAP_PIN) | ||
604 | |||
605 | #define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \ | ||
606 | CEPH_LOCK_IXATTR) | ||
607 | |||
608 | int ceph_caps_for_mode(int mode); | ||
609 | |||
610 | enum { | ||
611 | CEPH_CAP_OP_GRANT, /* mds->client grant */ | ||
612 | CEPH_CAP_OP_REVOKE, /* mds->client revoke */ | ||
613 | CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */ | ||
614 | CEPH_CAP_OP_EXPORT, /* mds has exported the cap */ | ||
615 | CEPH_CAP_OP_IMPORT, /* mds has imported the cap */ | ||
616 | CEPH_CAP_OP_UPDATE, /* client->mds update */ | ||
617 | CEPH_CAP_OP_DROP, /* client->mds drop cap bits */ | ||
618 | CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */ | ||
619 | CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */ | ||
620 | CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */ | ||
621 | CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */ | ||
622 | CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */ | ||
623 | CEPH_CAP_OP_RENEW, /* client->mds renewal request */ | ||
624 | }; | ||
625 | |||
626 | extern const char *ceph_cap_op_name(int op); | ||
627 | |||
628 | /* | ||
629 | * caps message, used for capability callbacks, acks, requests, etc. | ||
630 | */ | ||
631 | struct ceph_mds_caps { | ||
632 | __le32 op; /* CEPH_CAP_OP_* */ | ||
633 | __le64 ino, realm; | ||
634 | __le64 cap_id; | ||
635 | __le32 seq, issue_seq; | ||
636 | __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */ | ||
637 | __le32 migrate_seq; | ||
638 | __le64 snap_follows; | ||
639 | __le32 snap_trace_len; | ||
640 | |||
641 | /* authlock */ | ||
642 | __le32 uid, gid, mode; | ||
643 | |||
644 | /* linklock */ | ||
645 | __le32 nlink; | ||
646 | |||
647 | /* xattrlock */ | ||
648 | __le32 xattr_len; | ||
649 | __le64 xattr_version; | ||
650 | |||
651 | /* filelock */ | ||
652 | __le64 size, max_size, truncate_size; | ||
653 | __le32 truncate_seq; | ||
654 | struct ceph_timespec mtime, atime, ctime; | ||
655 | struct ceph_file_layout layout; | ||
656 | __le32 time_warp_seq; | ||
657 | } __attribute__ ((packed)); | ||
658 | |||
659 | /* cap release msg head */ | ||
660 | struct ceph_mds_cap_release { | ||
661 | __le32 num; /* number of cap_items that follow */ | ||
662 | } __attribute__ ((packed)); | ||
663 | |||
664 | struct ceph_mds_cap_item { | ||
665 | __le64 ino; | ||
666 | __le64 cap_id; | ||
667 | __le32 migrate_seq, seq; | ||
668 | } __attribute__ ((packed)); | ||
669 | |||
670 | #define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */ | ||
671 | #define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */ | ||
672 | #define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */ | ||
673 | #define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */ | ||
674 | |||
675 | extern const char *ceph_lease_op_name(int o); | ||
676 | |||
677 | /* lease msg header */ | ||
678 | struct ceph_mds_lease { | ||
679 | __u8 action; /* CEPH_MDS_LEASE_* */ | ||
680 | __le16 mask; /* which lease */ | ||
681 | __le64 ino; | ||
682 | __le64 first, last; /* snap range */ | ||
683 | __le32 seq; | ||
684 | __le32 duration_ms; /* duration of renewal */ | ||
685 | } __attribute__ ((packed)); | ||
686 | /* followed by a __le32+string for dname */ | ||
687 | |||
688 | /* client reconnect */ | ||
689 | struct ceph_mds_cap_reconnect { | ||
690 | __le64 cap_id; | ||
691 | __le32 wanted; | ||
692 | __le32 issued; | ||
693 | __le64 snaprealm; | ||
694 | __le64 pathbase; /* base ino for our path to this ino */ | ||
695 | __le32 flock_len; /* size of flock state blob, if any */ | ||
696 | } __attribute__ ((packed)); | ||
697 | /* followed by flock blob */ | ||
698 | |||
699 | struct ceph_mds_cap_reconnect_v1 { | ||
700 | __le64 cap_id; | ||
701 | __le32 wanted; | ||
702 | __le32 issued; | ||
703 | __le64 size; | ||
704 | struct ceph_timespec mtime, atime; | ||
705 | __le64 snaprealm; | ||
706 | __le64 pathbase; /* base ino for our path to this ino */ | ||
707 | } __attribute__ ((packed)); | ||
708 | |||
709 | struct ceph_mds_snaprealm_reconnect { | ||
710 | __le64 ino; /* snap realm base */ | ||
711 | __le64 seq; /* snap seq for this snap realm */ | ||
712 | __le64 parent; /* parent realm */ | ||
713 | } __attribute__ ((packed)); | ||
714 | |||
715 | /* | ||
716 | * snaps | ||
717 | */ | ||
718 | enum { | ||
719 | CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */ | ||
720 | CEPH_SNAP_OP_CREATE, | ||
721 | CEPH_SNAP_OP_DESTROY, | ||
722 | CEPH_SNAP_OP_SPLIT, | ||
723 | }; | ||
724 | |||
725 | extern const char *ceph_snap_op_name(int o); | ||
726 | |||
727 | /* snap msg header */ | ||
728 | struct ceph_mds_snap_head { | ||
729 | __le32 op; /* CEPH_SNAP_OP_* */ | ||
730 | __le64 split; /* ino to split off, if any */ | ||
731 | __le32 num_split_inos; /* # inos belonging to new child realm */ | ||
732 | __le32 num_split_realms; /* # child realms udner new child realm */ | ||
733 | __le32 trace_len; /* size of snap trace blob */ | ||
734 | } __attribute__ ((packed)); | ||
735 | /* followed by split ino list, then split realms, then the trace blob */ | ||
736 | |||
737 | /* | ||
738 | * encode info about a snaprealm, as viewed by a client | ||
739 | */ | ||
740 | struct ceph_mds_snap_realm { | ||
741 | __le64 ino; /* ino */ | ||
742 | __le64 created; /* snap: when created */ | ||
743 | __le64 parent; /* ino: parent realm */ | ||
744 | __le64 parent_since; /* snap: same parent since */ | ||
745 | __le64 seq; /* snap: version */ | ||
746 | __le32 num_snaps; | ||
747 | __le32 num_prior_parent_snaps; | ||
748 | } __attribute__ ((packed)); | ||
749 | /* followed by my snap list, then prior parent snap list */ | ||
750 | |||
751 | #endif | ||