diff options
Diffstat (limited to 'include/linux/ceph/ceph_fs.h')
-rw-r--r-- | include/linux/ceph/ceph_fs.h | 729 |
1 files changed, 729 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..c3c74aef289d --- /dev/null +++ b/include/linux/ceph/ceph_fs.h | |||
@@ -0,0 +1,729 @@ | |||
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 | |||
47 | |||
48 | /* | ||
49 | * ceph_file_layout - describe data layout for a file/inode | ||
50 | */ | ||
51 | struct ceph_file_layout { | ||
52 | /* file -> object mapping */ | ||
53 | __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple | ||
54 | of page size. */ | ||
55 | __le32 fl_stripe_count; /* over this many objects */ | ||
56 | __le32 fl_object_size; /* until objects are this big, then move to | ||
57 | new objects */ | ||
58 | __le32 fl_cas_hash; /* 0 = none; 1 = sha256 */ | ||
59 | |||
60 | /* pg -> disk layout */ | ||
61 | __le32 fl_object_stripe_unit; /* for per-object parity, if any */ | ||
62 | |||
63 | /* object -> pg layout */ | ||
64 | __le32 fl_pg_preferred; /* preferred primary for pg (-1 for none) */ | ||
65 | __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */ | ||
66 | } __attribute__ ((packed)); | ||
67 | |||
68 | #define CEPH_MIN_STRIPE_UNIT 65536 | ||
69 | |||
70 | int ceph_file_layout_is_valid(const struct ceph_file_layout *layout); | ||
71 | |||
72 | |||
73 | /* crypto algorithms */ | ||
74 | #define CEPH_CRYPTO_NONE 0x0 | ||
75 | #define CEPH_CRYPTO_AES 0x1 | ||
76 | |||
77 | #define CEPH_AES_IV "cephsageyudagreg" | ||
78 | |||
79 | /* security/authentication protocols */ | ||
80 | #define CEPH_AUTH_UNKNOWN 0x0 | ||
81 | #define CEPH_AUTH_NONE 0x1 | ||
82 | #define CEPH_AUTH_CEPHX 0x2 | ||
83 | |||
84 | #define CEPH_AUTH_UID_DEFAULT ((__u64) -1) | ||
85 | |||
86 | |||
87 | /********************************************* | ||
88 | * message layer | ||
89 | */ | ||
90 | |||
91 | /* | ||
92 | * message types | ||
93 | */ | ||
94 | |||
95 | /* misc */ | ||
96 | #define CEPH_MSG_SHUTDOWN 1 | ||
97 | #define CEPH_MSG_PING 2 | ||
98 | |||
99 | /* client <-> monitor */ | ||
100 | #define CEPH_MSG_MON_MAP 4 | ||
101 | #define CEPH_MSG_MON_GET_MAP 5 | ||
102 | #define CEPH_MSG_STATFS 13 | ||
103 | #define CEPH_MSG_STATFS_REPLY 14 | ||
104 | #define CEPH_MSG_MON_SUBSCRIBE 15 | ||
105 | #define CEPH_MSG_MON_SUBSCRIBE_ACK 16 | ||
106 | #define CEPH_MSG_AUTH 17 | ||
107 | #define CEPH_MSG_AUTH_REPLY 18 | ||
108 | |||
109 | /* client <-> mds */ | ||
110 | #define CEPH_MSG_MDS_MAP 21 | ||
111 | |||
112 | #define CEPH_MSG_CLIENT_SESSION 22 | ||
113 | #define CEPH_MSG_CLIENT_RECONNECT 23 | ||
114 | |||
115 | #define CEPH_MSG_CLIENT_REQUEST 24 | ||
116 | #define CEPH_MSG_CLIENT_REQUEST_FORWARD 25 | ||
117 | #define CEPH_MSG_CLIENT_REPLY 26 | ||
118 | #define CEPH_MSG_CLIENT_CAPS 0x310 | ||
119 | #define CEPH_MSG_CLIENT_LEASE 0x311 | ||
120 | #define CEPH_MSG_CLIENT_SNAP 0x312 | ||
121 | #define CEPH_MSG_CLIENT_CAPRELEASE 0x313 | ||
122 | |||
123 | /* pool ops */ | ||
124 | #define CEPH_MSG_POOLOP_REPLY 48 | ||
125 | #define CEPH_MSG_POOLOP 49 | ||
126 | |||
127 | |||
128 | /* osd */ | ||
129 | #define CEPH_MSG_OSD_MAP 41 | ||
130 | #define CEPH_MSG_OSD_OP 42 | ||
131 | #define CEPH_MSG_OSD_OPREPLY 43 | ||
132 | |||
133 | /* pool operations */ | ||
134 | enum { | ||
135 | POOL_OP_CREATE = 0x01, | ||
136 | POOL_OP_DELETE = 0x02, | ||
137 | POOL_OP_AUID_CHANGE = 0x03, | ||
138 | POOL_OP_CREATE_SNAP = 0x11, | ||
139 | POOL_OP_DELETE_SNAP = 0x12, | ||
140 | POOL_OP_CREATE_UNMANAGED_SNAP = 0x21, | ||
141 | POOL_OP_DELETE_UNMANAGED_SNAP = 0x22, | ||
142 | }; | ||
143 | |||
144 | struct ceph_mon_request_header { | ||
145 | __le64 have_version; | ||
146 | __le16 session_mon; | ||
147 | __le64 session_mon_tid; | ||
148 | } __attribute__ ((packed)); | ||
149 | |||
150 | struct ceph_mon_statfs { | ||
151 | struct ceph_mon_request_header monhdr; | ||
152 | struct ceph_fsid fsid; | ||
153 | } __attribute__ ((packed)); | ||
154 | |||
155 | struct ceph_statfs { | ||
156 | __le64 kb, kb_used, kb_avail; | ||
157 | __le64 num_objects; | ||
158 | } __attribute__ ((packed)); | ||
159 | |||
160 | struct ceph_mon_statfs_reply { | ||
161 | struct ceph_fsid fsid; | ||
162 | __le64 version; | ||
163 | struct ceph_statfs st; | ||
164 | } __attribute__ ((packed)); | ||
165 | |||
166 | const char *ceph_pool_op_name(int op); | ||
167 | |||
168 | struct ceph_mon_poolop { | ||
169 | struct ceph_mon_request_header monhdr; | ||
170 | struct ceph_fsid fsid; | ||
171 | __le32 pool; | ||
172 | __le32 op; | ||
173 | __le64 auid; | ||
174 | __le64 snapid; | ||
175 | __le32 name_len; | ||
176 | } __attribute__ ((packed)); | ||
177 | |||
178 | struct ceph_mon_poolop_reply { | ||
179 | struct ceph_mon_request_header monhdr; | ||
180 | struct ceph_fsid fsid; | ||
181 | __le32 reply_code; | ||
182 | __le32 epoch; | ||
183 | char has_data; | ||
184 | char data[0]; | ||
185 | } __attribute__ ((packed)); | ||
186 | |||
187 | struct ceph_mon_unmanaged_snap { | ||
188 | __le64 snapid; | ||
189 | } __attribute__ ((packed)); | ||
190 | |||
191 | struct ceph_osd_getmap { | ||
192 | struct ceph_mon_request_header monhdr; | ||
193 | struct ceph_fsid fsid; | ||
194 | __le32 start; | ||
195 | } __attribute__ ((packed)); | ||
196 | |||
197 | struct ceph_mds_getmap { | ||
198 | struct ceph_mon_request_header monhdr; | ||
199 | struct ceph_fsid fsid; | ||
200 | } __attribute__ ((packed)); | ||
201 | |||
202 | struct ceph_client_mount { | ||
203 | struct ceph_mon_request_header monhdr; | ||
204 | } __attribute__ ((packed)); | ||
205 | |||
206 | struct ceph_mon_subscribe_item { | ||
207 | __le64 have_version; __le64 have; | ||
208 | __u8 onetime; | ||
209 | } __attribute__ ((packed)); | ||
210 | |||
211 | struct ceph_mon_subscribe_ack { | ||
212 | __le32 duration; /* seconds */ | ||
213 | struct ceph_fsid fsid; | ||
214 | } __attribute__ ((packed)); | ||
215 | |||
216 | /* | ||
217 | * mds states | ||
218 | * > 0 -> in | ||
219 | * <= 0 -> out | ||
220 | */ | ||
221 | #define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */ | ||
222 | #define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees. | ||
223 | empty log. */ | ||
224 | #define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */ | ||
225 | #define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */ | ||
226 | #define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */ | ||
227 | #define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */ | ||
228 | #define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */ | ||
229 | |||
230 | #define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */ | ||
231 | #define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed | ||
232 | operations (import, rename, etc.) */ | ||
233 | #define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */ | ||
234 | #define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */ | ||
235 | #define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */ | ||
236 | #define CEPH_MDS_STATE_ACTIVE 13 /* up, active */ | ||
237 | #define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */ | ||
238 | |||
239 | extern const char *ceph_mds_state_name(int s); | ||
240 | |||
241 | |||
242 | /* | ||
243 | * metadata lock types. | ||
244 | * - these are bitmasks.. we can compose them | ||
245 | * - they also define the lock ordering by the MDS | ||
246 | * - a few of these are internal to the mds | ||
247 | */ | ||
248 | #define CEPH_LOCK_DVERSION 1 | ||
249 | #define CEPH_LOCK_DN 2 | ||
250 | #define CEPH_LOCK_ISNAP 16 | ||
251 | #define CEPH_LOCK_IVERSION 32 /* mds internal */ | ||
252 | #define CEPH_LOCK_IFILE 64 | ||
253 | #define CEPH_LOCK_IAUTH 128 | ||
254 | #define CEPH_LOCK_ILINK 256 | ||
255 | #define CEPH_LOCK_IDFT 512 /* dir frag tree */ | ||
256 | #define CEPH_LOCK_INEST 1024 /* mds internal */ | ||
257 | #define CEPH_LOCK_IXATTR 2048 | ||
258 | #define CEPH_LOCK_IFLOCK 4096 /* advisory file locks */ | ||
259 | #define CEPH_LOCK_INO 8192 /* immutable inode bits; not a lock */ | ||
260 | |||
261 | /* client_session ops */ | ||
262 | enum { | ||
263 | CEPH_SESSION_REQUEST_OPEN, | ||
264 | CEPH_SESSION_OPEN, | ||
265 | CEPH_SESSION_REQUEST_CLOSE, | ||
266 | CEPH_SESSION_CLOSE, | ||
267 | CEPH_SESSION_REQUEST_RENEWCAPS, | ||
268 | CEPH_SESSION_RENEWCAPS, | ||
269 | CEPH_SESSION_STALE, | ||
270 | CEPH_SESSION_RECALL_STATE, | ||
271 | }; | ||
272 | |||
273 | extern const char *ceph_session_op_name(int op); | ||
274 | |||
275 | struct ceph_mds_session_head { | ||
276 | __le32 op; | ||
277 | __le64 seq; | ||
278 | struct ceph_timespec stamp; | ||
279 | __le32 max_caps, max_leases; | ||
280 | } __attribute__ ((packed)); | ||
281 | |||
282 | /* client_request */ | ||
283 | /* | ||
284 | * metadata ops. | ||
285 | * & 0x001000 -> write op | ||
286 | * & 0x010000 -> follow symlink (e.g. stat(), not lstat()). | ||
287 | & & 0x100000 -> use weird ino/path trace | ||
288 | */ | ||
289 | #define CEPH_MDS_OP_WRITE 0x001000 | ||
290 | enum { | ||
291 | CEPH_MDS_OP_LOOKUP = 0x00100, | ||
292 | CEPH_MDS_OP_GETATTR = 0x00101, | ||
293 | CEPH_MDS_OP_LOOKUPHASH = 0x00102, | ||
294 | CEPH_MDS_OP_LOOKUPPARENT = 0x00103, | ||
295 | |||
296 | CEPH_MDS_OP_SETXATTR = 0x01105, | ||
297 | CEPH_MDS_OP_RMXATTR = 0x01106, | ||
298 | CEPH_MDS_OP_SETLAYOUT = 0x01107, | ||
299 | CEPH_MDS_OP_SETATTR = 0x01108, | ||
300 | CEPH_MDS_OP_SETFILELOCK= 0x01109, | ||
301 | CEPH_MDS_OP_GETFILELOCK= 0x00110, | ||
302 | CEPH_MDS_OP_SETDIRLAYOUT=0x0110a, | ||
303 | |||
304 | CEPH_MDS_OP_MKNOD = 0x01201, | ||
305 | CEPH_MDS_OP_LINK = 0x01202, | ||
306 | CEPH_MDS_OP_UNLINK = 0x01203, | ||
307 | CEPH_MDS_OP_RENAME = 0x01204, | ||
308 | CEPH_MDS_OP_MKDIR = 0x01220, | ||
309 | CEPH_MDS_OP_RMDIR = 0x01221, | ||
310 | CEPH_MDS_OP_SYMLINK = 0x01222, | ||
311 | |||
312 | CEPH_MDS_OP_CREATE = 0x01301, | ||
313 | CEPH_MDS_OP_OPEN = 0x00302, | ||
314 | CEPH_MDS_OP_READDIR = 0x00305, | ||
315 | |||
316 | CEPH_MDS_OP_LOOKUPSNAP = 0x00400, | ||
317 | CEPH_MDS_OP_MKSNAP = 0x01400, | ||
318 | CEPH_MDS_OP_RMSNAP = 0x01401, | ||
319 | CEPH_MDS_OP_LSSNAP = 0x00402, | ||
320 | }; | ||
321 | |||
322 | extern const char *ceph_mds_op_name(int op); | ||
323 | |||
324 | |||
325 | #define CEPH_SETATTR_MODE 1 | ||
326 | #define CEPH_SETATTR_UID 2 | ||
327 | #define CEPH_SETATTR_GID 4 | ||
328 | #define CEPH_SETATTR_MTIME 8 | ||
329 | #define CEPH_SETATTR_ATIME 16 | ||
330 | #define CEPH_SETATTR_SIZE 32 | ||
331 | #define CEPH_SETATTR_CTIME 64 | ||
332 | |||
333 | union ceph_mds_request_args { | ||
334 | struct { | ||
335 | __le32 mask; /* CEPH_CAP_* */ | ||
336 | } __attribute__ ((packed)) getattr; | ||
337 | struct { | ||
338 | __le32 mode; | ||
339 | __le32 uid; | ||
340 | __le32 gid; | ||
341 | struct ceph_timespec mtime; | ||
342 | struct ceph_timespec atime; | ||
343 | __le64 size, old_size; /* old_size needed by truncate */ | ||
344 | __le32 mask; /* CEPH_SETATTR_* */ | ||
345 | } __attribute__ ((packed)) setattr; | ||
346 | struct { | ||
347 | __le32 frag; /* which dir fragment */ | ||
348 | __le32 max_entries; /* how many dentries to grab */ | ||
349 | __le32 max_bytes; | ||
350 | } __attribute__ ((packed)) readdir; | ||
351 | struct { | ||
352 | __le32 mode; | ||
353 | __le32 rdev; | ||
354 | } __attribute__ ((packed)) mknod; | ||
355 | struct { | ||
356 | __le32 mode; | ||
357 | } __attribute__ ((packed)) mkdir; | ||
358 | struct { | ||
359 | __le32 flags; | ||
360 | __le32 mode; | ||
361 | __le32 stripe_unit; /* layout for newly created file */ | ||
362 | __le32 stripe_count; /* ... */ | ||
363 | __le32 object_size; | ||
364 | __le32 file_replication; | ||
365 | __le32 preferred; | ||
366 | } __attribute__ ((packed)) open; | ||
367 | struct { | ||
368 | __le32 flags; | ||
369 | } __attribute__ ((packed)) setxattr; | ||
370 | struct { | ||
371 | struct ceph_file_layout layout; | ||
372 | } __attribute__ ((packed)) setlayout; | ||
373 | struct { | ||
374 | __u8 rule; /* currently fcntl or flock */ | ||
375 | __u8 type; /* shared, exclusive, remove*/ | ||
376 | __le64 pid; /* process id requesting the lock */ | ||
377 | __le64 pid_namespace; | ||
378 | __le64 start; /* initial location to lock */ | ||
379 | __le64 length; /* num bytes to lock from start */ | ||
380 | __u8 wait; /* will caller wait for lock to become available? */ | ||
381 | } __attribute__ ((packed)) filelock_change; | ||
382 | } __attribute__ ((packed)); | ||
383 | |||
384 | #define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */ | ||
385 | #define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */ | ||
386 | |||
387 | struct ceph_mds_request_head { | ||
388 | __le64 oldest_client_tid; | ||
389 | __le32 mdsmap_epoch; /* on client */ | ||
390 | __le32 flags; /* CEPH_MDS_FLAG_* */ | ||
391 | __u8 num_retry, num_fwd; /* count retry, fwd attempts */ | ||
392 | __le16 num_releases; /* # include cap/lease release records */ | ||
393 | __le32 op; /* mds op code */ | ||
394 | __le32 caller_uid, caller_gid; | ||
395 | __le64 ino; /* use this ino for openc, mkdir, mknod, | ||
396 | etc. (if replaying) */ | ||
397 | union ceph_mds_request_args args; | ||
398 | } __attribute__ ((packed)); | ||
399 | |||
400 | /* cap/lease release record */ | ||
401 | struct ceph_mds_request_release { | ||
402 | __le64 ino, cap_id; /* ino and unique cap id */ | ||
403 | __le32 caps, wanted; /* new issued, wanted */ | ||
404 | __le32 seq, issue_seq, mseq; | ||
405 | __le32 dname_seq; /* if releasing a dentry lease, a */ | ||
406 | __le32 dname_len; /* string follows. */ | ||
407 | } __attribute__ ((packed)); | ||
408 | |||
409 | /* client reply */ | ||
410 | struct ceph_mds_reply_head { | ||
411 | __le32 op; | ||
412 | __le32 result; | ||
413 | __le32 mdsmap_epoch; | ||
414 | __u8 safe; /* true if committed to disk */ | ||
415 | __u8 is_dentry, is_target; /* true if dentry, target inode records | ||
416 | are included with reply */ | ||
417 | } __attribute__ ((packed)); | ||
418 | |||
419 | /* one for each node split */ | ||
420 | struct ceph_frag_tree_split { | ||
421 | __le32 frag; /* this frag splits... */ | ||
422 | __le32 by; /* ...by this many bits */ | ||
423 | } __attribute__ ((packed)); | ||
424 | |||
425 | struct ceph_frag_tree_head { | ||
426 | __le32 nsplits; /* num ceph_frag_tree_split records */ | ||
427 | struct ceph_frag_tree_split splits[]; | ||
428 | } __attribute__ ((packed)); | ||
429 | |||
430 | /* capability issue, for bundling with mds reply */ | ||
431 | struct ceph_mds_reply_cap { | ||
432 | __le32 caps, wanted; /* caps issued, wanted */ | ||
433 | __le64 cap_id; | ||
434 | __le32 seq, mseq; | ||
435 | __le64 realm; /* snap realm */ | ||
436 | __u8 flags; /* CEPH_CAP_FLAG_* */ | ||
437 | } __attribute__ ((packed)); | ||
438 | |||
439 | #define CEPH_CAP_FLAG_AUTH 1 /* cap is issued by auth mds */ | ||
440 | |||
441 | /* inode record, for bundling with mds reply */ | ||
442 | struct ceph_mds_reply_inode { | ||
443 | __le64 ino; | ||
444 | __le64 snapid; | ||
445 | __le32 rdev; | ||
446 | __le64 version; /* inode version */ | ||
447 | __le64 xattr_version; /* version for xattr blob */ | ||
448 | struct ceph_mds_reply_cap cap; /* caps issued for this inode */ | ||
449 | struct ceph_file_layout layout; | ||
450 | struct ceph_timespec ctime, mtime, atime; | ||
451 | __le32 time_warp_seq; | ||
452 | __le64 size, max_size, truncate_size; | ||
453 | __le32 truncate_seq; | ||
454 | __le32 mode, uid, gid; | ||
455 | __le32 nlink; | ||
456 | __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */ | ||
457 | struct ceph_timespec rctime; | ||
458 | struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */ | ||
459 | } __attribute__ ((packed)); | ||
460 | /* followed by frag array, then symlink string, then xattr blob */ | ||
461 | |||
462 | /* reply_lease follows dname, and reply_inode */ | ||
463 | struct ceph_mds_reply_lease { | ||
464 | __le16 mask; /* lease type(s) */ | ||
465 | __le32 duration_ms; /* lease duration */ | ||
466 | __le32 seq; | ||
467 | } __attribute__ ((packed)); | ||
468 | |||
469 | struct ceph_mds_reply_dirfrag { | ||
470 | __le32 frag; /* fragment */ | ||
471 | __le32 auth; /* auth mds, if this is a delegation point */ | ||
472 | __le32 ndist; /* number of mds' this is replicated on */ | ||
473 | __le32 dist[]; | ||
474 | } __attribute__ ((packed)); | ||
475 | |||
476 | #define CEPH_LOCK_FCNTL 1 | ||
477 | #define CEPH_LOCK_FLOCK 2 | ||
478 | |||
479 | #define CEPH_LOCK_SHARED 1 | ||
480 | #define CEPH_LOCK_EXCL 2 | ||
481 | #define CEPH_LOCK_UNLOCK 4 | ||
482 | |||
483 | struct ceph_filelock { | ||
484 | __le64 start;/* file offset to start lock at */ | ||
485 | __le64 length; /* num bytes to lock; 0 for all following start */ | ||
486 | __le64 client; /* which client holds the lock */ | ||
487 | __le64 pid; /* process id holding the lock on the client */ | ||
488 | __le64 pid_namespace; | ||
489 | __u8 type; /* shared lock, exclusive lock, or unlock */ | ||
490 | } __attribute__ ((packed)); | ||
491 | |||
492 | |||
493 | /* file access modes */ | ||
494 | #define CEPH_FILE_MODE_PIN 0 | ||
495 | #define CEPH_FILE_MODE_RD 1 | ||
496 | #define CEPH_FILE_MODE_WR 2 | ||
497 | #define CEPH_FILE_MODE_RDWR 3 /* RD | WR */ | ||
498 | #define CEPH_FILE_MODE_LAZY 4 /* lazy io */ | ||
499 | #define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */ | ||
500 | |||
501 | int ceph_flags_to_mode(int flags); | ||
502 | |||
503 | |||
504 | /* capability bits */ | ||
505 | #define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */ | ||
506 | |||
507 | /* generic cap bits */ | ||
508 | #define CEPH_CAP_GSHARED 1 /* client can reads */ | ||
509 | #define CEPH_CAP_GEXCL 2 /* client can read and update */ | ||
510 | #define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */ | ||
511 | #define CEPH_CAP_GRD 8 /* (file) client can read */ | ||
512 | #define CEPH_CAP_GWR 16 /* (file) client can write */ | ||
513 | #define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */ | ||
514 | #define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */ | ||
515 | #define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */ | ||
516 | |||
517 | /* per-lock shift */ | ||
518 | #define CEPH_CAP_SAUTH 2 | ||
519 | #define CEPH_CAP_SLINK 4 | ||
520 | #define CEPH_CAP_SXATTR 6 | ||
521 | #define CEPH_CAP_SFILE 8 | ||
522 | #define CEPH_CAP_SFLOCK 20 | ||
523 | |||
524 | #define CEPH_CAP_BITS 22 | ||
525 | |||
526 | /* composed values */ | ||
527 | #define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH) | ||
528 | #define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH) | ||
529 | #define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK) | ||
530 | #define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK) | ||
531 | #define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR) | ||
532 | #define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR) | ||
533 | #define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE) | ||
534 | #define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE) | ||
535 | #define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE) | ||
536 | #define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE) | ||
537 | #define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE) | ||
538 | #define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE) | ||
539 | #define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE) | ||
540 | #define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE) | ||
541 | #define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE) | ||
542 | #define CEPH_CAP_FLOCK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFLOCK) | ||
543 | #define CEPH_CAP_FLOCK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFLOCK) | ||
544 | |||
545 | |||
546 | /* cap masks (for getattr) */ | ||
547 | #define CEPH_STAT_CAP_INODE CEPH_CAP_PIN | ||
548 | #define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */ | ||
549 | #define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN | ||
550 | #define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED | ||
551 | #define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED | ||
552 | #define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED | ||
553 | #define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED | ||
554 | #define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED | ||
555 | #define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED | ||
556 | #define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED | ||
557 | #define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */ | ||
558 | #define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED | ||
559 | #define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \ | ||
560 | CEPH_CAP_AUTH_SHARED | \ | ||
561 | CEPH_CAP_LINK_SHARED | \ | ||
562 | CEPH_CAP_FILE_SHARED | \ | ||
563 | CEPH_CAP_XATTR_SHARED) | ||
564 | |||
565 | #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \ | ||
566 | CEPH_CAP_LINK_SHARED | \ | ||
567 | CEPH_CAP_XATTR_SHARED | \ | ||
568 | CEPH_CAP_FILE_SHARED) | ||
569 | #define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \ | ||
570 | CEPH_CAP_FILE_CACHE) | ||
571 | |||
572 | #define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \ | ||
573 | CEPH_CAP_LINK_EXCL | \ | ||
574 | CEPH_CAP_XATTR_EXCL | \ | ||
575 | CEPH_CAP_FILE_EXCL) | ||
576 | #define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \ | ||
577 | CEPH_CAP_FILE_EXCL) | ||
578 | #define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR) | ||
579 | #define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \ | ||
580 | CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \ | ||
581 | CEPH_CAP_PIN) | ||
582 | |||
583 | #define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \ | ||
584 | CEPH_LOCK_IXATTR) | ||
585 | |||
586 | int ceph_caps_for_mode(int mode); | ||
587 | |||
588 | enum { | ||
589 | CEPH_CAP_OP_GRANT, /* mds->client grant */ | ||
590 | CEPH_CAP_OP_REVOKE, /* mds->client revoke */ | ||
591 | CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */ | ||
592 | CEPH_CAP_OP_EXPORT, /* mds has exported the cap */ | ||
593 | CEPH_CAP_OP_IMPORT, /* mds has imported the cap */ | ||
594 | CEPH_CAP_OP_UPDATE, /* client->mds update */ | ||
595 | CEPH_CAP_OP_DROP, /* client->mds drop cap bits */ | ||
596 | CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */ | ||
597 | CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */ | ||
598 | CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */ | ||
599 | CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */ | ||
600 | CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */ | ||
601 | CEPH_CAP_OP_RENEW, /* client->mds renewal request */ | ||
602 | }; | ||
603 | |||
604 | extern const char *ceph_cap_op_name(int op); | ||
605 | |||
606 | /* | ||
607 | * caps message, used for capability callbacks, acks, requests, etc. | ||
608 | */ | ||
609 | struct ceph_mds_caps { | ||
610 | __le32 op; /* CEPH_CAP_OP_* */ | ||
611 | __le64 ino, realm; | ||
612 | __le64 cap_id; | ||
613 | __le32 seq, issue_seq; | ||
614 | __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */ | ||
615 | __le32 migrate_seq; | ||
616 | __le64 snap_follows; | ||
617 | __le32 snap_trace_len; | ||
618 | |||
619 | /* authlock */ | ||
620 | __le32 uid, gid, mode; | ||
621 | |||
622 | /* linklock */ | ||
623 | __le32 nlink; | ||
624 | |||
625 | /* xattrlock */ | ||
626 | __le32 xattr_len; | ||
627 | __le64 xattr_version; | ||
628 | |||
629 | /* filelock */ | ||
630 | __le64 size, max_size, truncate_size; | ||
631 | __le32 truncate_seq; | ||
632 | struct ceph_timespec mtime, atime, ctime; | ||
633 | struct ceph_file_layout layout; | ||
634 | __le32 time_warp_seq; | ||
635 | } __attribute__ ((packed)); | ||
636 | |||
637 | /* cap release msg head */ | ||
638 | struct ceph_mds_cap_release { | ||
639 | __le32 num; /* number of cap_items that follow */ | ||
640 | } __attribute__ ((packed)); | ||
641 | |||
642 | struct ceph_mds_cap_item { | ||
643 | __le64 ino; | ||
644 | __le64 cap_id; | ||
645 | __le32 migrate_seq, seq; | ||
646 | } __attribute__ ((packed)); | ||
647 | |||
648 | #define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */ | ||
649 | #define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */ | ||
650 | #define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */ | ||
651 | #define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */ | ||
652 | |||
653 | extern const char *ceph_lease_op_name(int o); | ||
654 | |||
655 | /* lease msg header */ | ||
656 | struct ceph_mds_lease { | ||
657 | __u8 action; /* CEPH_MDS_LEASE_* */ | ||
658 | __le16 mask; /* which lease */ | ||
659 | __le64 ino; | ||
660 | __le64 first, last; /* snap range */ | ||
661 | __le32 seq; | ||
662 | __le32 duration_ms; /* duration of renewal */ | ||
663 | } __attribute__ ((packed)); | ||
664 | /* followed by a __le32+string for dname */ | ||
665 | |||
666 | /* client reconnect */ | ||
667 | struct ceph_mds_cap_reconnect { | ||
668 | __le64 cap_id; | ||
669 | __le32 wanted; | ||
670 | __le32 issued; | ||
671 | __le64 snaprealm; | ||
672 | __le64 pathbase; /* base ino for our path to this ino */ | ||
673 | __le32 flock_len; /* size of flock state blob, if any */ | ||
674 | } __attribute__ ((packed)); | ||
675 | /* followed by flock blob */ | ||
676 | |||
677 | struct ceph_mds_cap_reconnect_v1 { | ||
678 | __le64 cap_id; | ||
679 | __le32 wanted; | ||
680 | __le32 issued; | ||
681 | __le64 size; | ||
682 | struct ceph_timespec mtime, atime; | ||
683 | __le64 snaprealm; | ||
684 | __le64 pathbase; /* base ino for our path to this ino */ | ||
685 | } __attribute__ ((packed)); | ||
686 | |||
687 | struct ceph_mds_snaprealm_reconnect { | ||
688 | __le64 ino; /* snap realm base */ | ||
689 | __le64 seq; /* snap seq for this snap realm */ | ||
690 | __le64 parent; /* parent realm */ | ||
691 | } __attribute__ ((packed)); | ||
692 | |||
693 | /* | ||
694 | * snaps | ||
695 | */ | ||
696 | enum { | ||
697 | CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */ | ||
698 | CEPH_SNAP_OP_CREATE, | ||
699 | CEPH_SNAP_OP_DESTROY, | ||
700 | CEPH_SNAP_OP_SPLIT, | ||
701 | }; | ||
702 | |||
703 | extern const char *ceph_snap_op_name(int o); | ||
704 | |||
705 | /* snap msg header */ | ||
706 | struct ceph_mds_snap_head { | ||
707 | __le32 op; /* CEPH_SNAP_OP_* */ | ||
708 | __le64 split; /* ino to split off, if any */ | ||
709 | __le32 num_split_inos; /* # inos belonging to new child realm */ | ||
710 | __le32 num_split_realms; /* # child realms udner new child realm */ | ||
711 | __le32 trace_len; /* size of snap trace blob */ | ||
712 | } __attribute__ ((packed)); | ||
713 | /* followed by split ino list, then split realms, then the trace blob */ | ||
714 | |||
715 | /* | ||
716 | * encode info about a snaprealm, as viewed by a client | ||
717 | */ | ||
718 | struct ceph_mds_snap_realm { | ||
719 | __le64 ino; /* ino */ | ||
720 | __le64 created; /* snap: when created */ | ||
721 | __le64 parent; /* ino: parent realm */ | ||
722 | __le64 parent_since; /* snap: same parent since */ | ||
723 | __le64 seq; /* snap: version */ | ||
724 | __le32 num_snaps; | ||
725 | __le32 num_prior_parent_snaps; | ||
726 | } __attribute__ ((packed)); | ||
727 | /* followed by my snap list, then prior parent snap list */ | ||
728 | |||
729 | #endif | ||