blob: cf6f4d998a7642567bb9198d90b16f07c6996fa8 [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
Sage Weil5cd068c2010-07-07 08:38:17 -070012#ifndef CEPH_FS_H
13#define CEPH_FS_H
Sage Weil0dee3c22009-10-06 11:31:06 -070014
David Howellsa1ce3922012-10-02 18:01:25 +010015#include <linux/ceph/msgr.h>
16#include <linux/ceph/rados.h>
Sage Weil0dee3c22009-10-06 11:31:06 -070017
18/*
Sage Weil0dee3c22009-10-06 11:31:06 -070019 * 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 */
Sage Weil6df058c2009-12-22 11:24:33 -080024#define CEPH_OSD_PROTOCOL 8 /* cluster internal */
Sage Weilca9d93a2010-05-12 14:48:20 -070025#define CEPH_MDS_PROTOCOL 12 /* cluster internal */
Sage Weil13e38c82009-10-09 16:36:34 -070026#define CEPH_MON_PROTOCOL 5 /* cluster internal */
Yehuda Sadeh0c948992010-02-01 16:10:45 -080027#define CEPH_OSDC_PROTOCOL 24 /* server/client */
Sage Weil3ea25f92010-01-25 11:18:17 -080028#define CEPH_MDSC_PROTOCOL 32 /* server/client */
Sage Weil13e38c82009-10-09 16:36:34 -070029#define CEPH_MONC_PROTOCOL 15 /* server/client */
Sage Weil0dee3c22009-10-06 11:31:06 -070030
31
32#define CEPH_INO_ROOT 1
Sage Weil1d1de9162009-12-02 11:54:25 -080033#define CEPH_INO_CEPH 2 /* hidden .ceph dir */
Sage Weil0dee3c22009-10-06 11:31:06 -070034
35/* arbitrary limit on max # of monitors (cluster of 3 is typical) */
36#define CEPH_MAX_MON 31
37
Sage Weil0dee3c22009-10-06 11:31:06 -070038/*
39 * ceph_file_layout - describe data layout for a file/inode
40 */
41struct ceph_file_layout {
42 /* file -> object mapping */
43 __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple
44 of page size. */
45 __le32 fl_stripe_count; /* over this many objects */
46 __le32 fl_object_size; /* until objects are this big, then move to
47 new objects */
Sage Weil6c0f3af2010-11-16 11:14:34 -080048 __le32 fl_cas_hash; /* UNUSED. 0 = none; 1 = sha256 */
Sage Weil0dee3c22009-10-06 11:31:06 -070049
50 /* pg -> disk layout */
Sage Weil6c0f3af2010-11-16 11:14:34 -080051 __le32 fl_object_stripe_unit; /* UNUSED. for per-object parity, if any */
Sage Weil0dee3c22009-10-06 11:31:06 -070052
53 /* object -> pg layout */
Sage Weil3469ac1a2012-05-07 15:33:36 -070054 __le32 fl_unused; /* unused; used to be preferred primary (-1) */
Sage Weil0dee3c22009-10-06 11:31:06 -070055 __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */
56} __attribute__ ((packed));
57
Sage Weil752727a2009-10-09 16:38:45 -070058#define CEPH_MIN_STRIPE_UNIT 65536
Sage Weil0dee3c22009-10-06 11:31:06 -070059
Sage Weil752727a2009-10-09 16:38:45 -070060int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
Sage Weil0dee3c22009-10-06 11:31:06 -070061
Sage Weil6c0f3af2010-11-16 11:14:34 -080062struct ceph_dir_layout {
63 __u8 dl_dir_hash; /* see ceph_hash.h for ids */
64 __u8 dl_unused1;
65 __u16 dl_unused2;
66 __u32 dl_unused3;
67} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -070068
Sage Weil4e7a5dc2009-11-18 16:19:57 -080069/* crypto algorithms */
70#define CEPH_CRYPTO_NONE 0x0
71#define CEPH_CRYPTO_AES 0x1
72
Sage Weilcbbfe492010-08-02 15:48:23 -070073#define CEPH_AES_IV "cephsageyudagreg"
74
Sage Weil4e7a5dc2009-11-18 16:19:57 -080075/* security/authentication protocols */
76#define CEPH_AUTH_UNKNOWN 0x0
77#define CEPH_AUTH_NONE 0x1
78#define CEPH_AUTH_CEPHX 0x2
79
Sage Weilca9d93a2010-05-12 14:48:20 -070080#define CEPH_AUTH_UID_DEFAULT ((__u64) -1)
81
Sage Weil4e7a5dc2009-11-18 16:19:57 -080082
Sage Weil0dee3c22009-10-06 11:31:06 -070083/*********************************************
84 * message layer
85 */
86
87/*
88 * message types
89 */
90
91/* misc */
92#define CEPH_MSG_SHUTDOWN 1
93#define CEPH_MSG_PING 2
94
95/* client <-> monitor */
96#define CEPH_MSG_MON_MAP 4
97#define CEPH_MSG_MON_GET_MAP 5
Sage Weil0dee3c22009-10-06 11:31:06 -070098#define CEPH_MSG_STATFS 13
99#define CEPH_MSG_STATFS_REPLY 14
100#define CEPH_MSG_MON_SUBSCRIBE 15
101#define CEPH_MSG_MON_SUBSCRIBE_ACK 16
Sage Weil4e7a5dc2009-11-18 16:19:57 -0800102#define CEPH_MSG_AUTH 17
103#define CEPH_MSG_AUTH_REPLY 18
Sage Weil0dee3c22009-10-06 11:31:06 -0700104
105/* client <-> mds */
Sage Weil0dee3c22009-10-06 11:31:06 -0700106#define CEPH_MSG_MDS_MAP 21
107
108#define CEPH_MSG_CLIENT_SESSION 22
109#define CEPH_MSG_CLIENT_RECONNECT 23
110
111#define CEPH_MSG_CLIENT_REQUEST 24
112#define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
113#define CEPH_MSG_CLIENT_REPLY 26
114#define CEPH_MSG_CLIENT_CAPS 0x310
115#define CEPH_MSG_CLIENT_LEASE 0x311
116#define CEPH_MSG_CLIENT_SNAP 0x312
117#define CEPH_MSG_CLIENT_CAPRELEASE 0x313
118
Sage Weilca9d93a2010-05-12 14:48:20 -0700119/* pool ops */
120#define CEPH_MSG_POOLOP_REPLY 48
121#define CEPH_MSG_POOLOP 49
122
123
Sage Weil0dee3c22009-10-06 11:31:06 -0700124/* osd */
Yehuda Sadeh483fac72011-01-20 16:36:06 -0800125#define CEPH_MSG_OSD_MAP 41
126#define CEPH_MSG_OSD_OP 42
127#define CEPH_MSG_OSD_OPREPLY 43
128#define CEPH_MSG_WATCH_NOTIFY 44
129
130
131/* watch-notify operations */
132enum {
133 WATCH_NOTIFY = 1, /* notifying watcher */
134 WATCH_NOTIFY_COMPLETE = 2, /* notifier notified when done */
135};
136
Sage Weil0dee3c22009-10-06 11:31:06 -0700137
Sage Weilca9d93a2010-05-12 14:48:20 -0700138/* pool operations */
139enum {
140 POOL_OP_CREATE = 0x01,
141 POOL_OP_DELETE = 0x02,
142 POOL_OP_AUID_CHANGE = 0x03,
143 POOL_OP_CREATE_SNAP = 0x11,
144 POOL_OP_DELETE_SNAP = 0x12,
145 POOL_OP_CREATE_UNMANAGED_SNAP = 0x21,
146 POOL_OP_DELETE_UNMANAGED_SNAP = 0x22,
147};
148
Sage Weil13e38c82009-10-09 16:36:34 -0700149struct ceph_mon_request_header {
150 __le64 have_version;
151 __le16 session_mon;
152 __le64 session_mon_tid;
153} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -0700154
155struct ceph_mon_statfs {
Sage Weil13e38c82009-10-09 16:36:34 -0700156 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700157 struct ceph_fsid fsid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700158} __attribute__ ((packed));
159
160struct ceph_statfs {
161 __le64 kb, kb_used, kb_avail;
162 __le64 num_objects;
163} __attribute__ ((packed));
164
165struct ceph_mon_statfs_reply {
166 struct ceph_fsid fsid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700167 __le64 version;
168 struct ceph_statfs st;
169} __attribute__ ((packed));
170
Sage Weilca9d93a2010-05-12 14:48:20 -0700171const char *ceph_pool_op_name(int op);
172
173struct ceph_mon_poolop {
174 struct ceph_mon_request_header monhdr;
175 struct ceph_fsid fsid;
176 __le32 pool;
177 __le32 op;
178 __le64 auid;
179 __le64 snapid;
180 __le32 name_len;
181} __attribute__ ((packed));
182
183struct ceph_mon_poolop_reply {
184 struct ceph_mon_request_header monhdr;
185 struct ceph_fsid fsid;
186 __le32 reply_code;
187 __le32 epoch;
188 char has_data;
189 char data[0];
190} __attribute__ ((packed));
191
192struct ceph_mon_unmanaged_snap {
193 __le64 snapid;
194} __attribute__ ((packed));
195
Sage Weil0dee3c22009-10-06 11:31:06 -0700196struct ceph_osd_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700197 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700198 struct ceph_fsid fsid;
199 __le32 start;
200} __attribute__ ((packed));
201
202struct ceph_mds_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700203 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700204 struct ceph_fsid fsid;
205} __attribute__ ((packed));
206
207struct ceph_client_mount {
Sage Weil13e38c82009-10-09 16:36:34 -0700208 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700209} __attribute__ ((packed));
210
Yehuda Sadeh483fac72011-01-20 16:36:06 -0800211#define CEPH_SUBSCRIBE_ONETIME 1 /* i want only 1 update after have */
212
Sage Weil0dee3c22009-10-06 11:31:06 -0700213struct ceph_mon_subscribe_item {
Yehuda Sadeh483fac72011-01-20 16:36:06 -0800214 __le64 have_version; __le64 have;
Sage Weil0dee3c22009-10-06 11:31:06 -0700215 __u8 onetime;
216} __attribute__ ((packed));
217
Sage Weil07bd10f2009-10-14 17:26:40 -0700218struct ceph_mon_subscribe_ack {
219 __le32 duration; /* seconds */
220 struct ceph_fsid fsid;
221} __attribute__ ((packed));
222
Sage Weil0dee3c22009-10-06 11:31:06 -0700223/*
224 * mds states
225 * > 0 -> in
226 * <= 0 -> out
227 */
228#define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */
229#define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees.
230 empty log. */
231#define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */
232#define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */
233#define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */
234#define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */
235#define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
236
237#define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */
238#define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed
239 operations (import, rename, etc.) */
240#define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */
241#define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */
242#define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
243#define CEPH_MDS_STATE_ACTIVE 13 /* up, active */
244#define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */
245
246extern const char *ceph_mds_state_name(int s);
247
248
249/*
250 * metadata lock types.
251 * - these are bitmasks.. we can compose them
252 * - they also define the lock ordering by the MDS
253 * - a few of these are internal to the mds
254 */
Sage Weildd1c9052010-05-25 16:45:25 -0700255#define CEPH_LOCK_DVERSION 1
256#define CEPH_LOCK_DN 2
257#define CEPH_LOCK_ISNAP 16
258#define CEPH_LOCK_IVERSION 32 /* mds internal */
259#define CEPH_LOCK_IFILE 64
260#define CEPH_LOCK_IAUTH 128
261#define CEPH_LOCK_ILINK 256
262#define CEPH_LOCK_IDFT 512 /* dir frag tree */
263#define CEPH_LOCK_INEST 1024 /* mds internal */
264#define CEPH_LOCK_IXATTR 2048
Sage Weilf0b18d92010-07-07 08:39:03 -0700265#define CEPH_LOCK_IFLOCK 4096 /* advisory file locks */
Sage Weildd1c9052010-05-25 16:45:25 -0700266#define CEPH_LOCK_INO 8192 /* immutable inode bits; not a lock */
Sage Weil0dee3c22009-10-06 11:31:06 -0700267
268/* client_session ops */
269enum {
270 CEPH_SESSION_REQUEST_OPEN,
271 CEPH_SESSION_OPEN,
272 CEPH_SESSION_REQUEST_CLOSE,
273 CEPH_SESSION_CLOSE,
274 CEPH_SESSION_REQUEST_RENEWCAPS,
275 CEPH_SESSION_RENEWCAPS,
276 CEPH_SESSION_STALE,
277 CEPH_SESSION_RECALL_STATE,
278};
279
280extern const char *ceph_session_op_name(int op);
281
282struct ceph_mds_session_head {
283 __le32 op;
284 __le64 seq;
285 struct ceph_timespec stamp;
286 __le32 max_caps, max_leases;
287} __attribute__ ((packed));
288
289/* client_request */
290/*
291 * metadata ops.
292 * & 0x001000 -> write op
293 * & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
294 & & 0x100000 -> use weird ino/path trace
295 */
296#define CEPH_MDS_OP_WRITE 0x001000
297enum {
298 CEPH_MDS_OP_LOOKUP = 0x00100,
299 CEPH_MDS_OP_GETATTR = 0x00101,
300 CEPH_MDS_OP_LOOKUPHASH = 0x00102,
301 CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
Sage Weil3c454cf2011-04-06 09:31:40 -0700302 CEPH_MDS_OP_LOOKUPINO = 0x00104,
Sage Weil0dee3c22009-10-06 11:31:06 -0700303
304 CEPH_MDS_OP_SETXATTR = 0x01105,
305 CEPH_MDS_OP_RMXATTR = 0x01106,
306 CEPH_MDS_OP_SETLAYOUT = 0x01107,
307 CEPH_MDS_OP_SETATTR = 0x01108,
Greg Farnumfbaad972010-08-02 15:30:08 -0700308 CEPH_MDS_OP_SETFILELOCK= 0x01109,
309 CEPH_MDS_OP_GETFILELOCK= 0x00110,
Greg Farnum571dba52010-09-24 14:56:40 -0700310 CEPH_MDS_OP_SETDIRLAYOUT=0x0110a,
Sage Weil0dee3c22009-10-06 11:31:06 -0700311
312 CEPH_MDS_OP_MKNOD = 0x01201,
313 CEPH_MDS_OP_LINK = 0x01202,
314 CEPH_MDS_OP_UNLINK = 0x01203,
315 CEPH_MDS_OP_RENAME = 0x01204,
316 CEPH_MDS_OP_MKDIR = 0x01220,
317 CEPH_MDS_OP_RMDIR = 0x01221,
318 CEPH_MDS_OP_SYMLINK = 0x01222,
319
Sage Weil3ea25f92010-01-25 11:18:17 -0800320 CEPH_MDS_OP_CREATE = 0x01301,
Sage Weil0dee3c22009-10-06 11:31:06 -0700321 CEPH_MDS_OP_OPEN = 0x00302,
322 CEPH_MDS_OP_READDIR = 0x00305,
323
324 CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
325 CEPH_MDS_OP_MKSNAP = 0x01400,
326 CEPH_MDS_OP_RMSNAP = 0x01401,
327 CEPH_MDS_OP_LSSNAP = 0x00402,
328};
329
330extern const char *ceph_mds_op_name(int op);
331
332
333#define CEPH_SETATTR_MODE 1
334#define CEPH_SETATTR_UID 2
335#define CEPH_SETATTR_GID 4
336#define CEPH_SETATTR_MTIME 8
337#define CEPH_SETATTR_ATIME 16
338#define CEPH_SETATTR_SIZE 32
339#define CEPH_SETATTR_CTIME 64
340
341union ceph_mds_request_args {
342 struct {
343 __le32 mask; /* CEPH_CAP_* */
344 } __attribute__ ((packed)) getattr;
345 struct {
346 __le32 mode;
347 __le32 uid;
348 __le32 gid;
349 struct ceph_timespec mtime;
350 struct ceph_timespec atime;
351 __le64 size, old_size; /* old_size needed by truncate */
352 __le32 mask; /* CEPH_SETATTR_* */
353 } __attribute__ ((packed)) setattr;
354 struct {
355 __le32 frag; /* which dir fragment */
356 __le32 max_entries; /* how many dentries to grab */
Sage Weil23804d92010-05-14 13:06:30 -0700357 __le32 max_bytes;
Sage Weil0dee3c22009-10-06 11:31:06 -0700358 } __attribute__ ((packed)) readdir;
359 struct {
360 __le32 mode;
361 __le32 rdev;
362 } __attribute__ ((packed)) mknod;
363 struct {
364 __le32 mode;
365 } __attribute__ ((packed)) mkdir;
366 struct {
367 __le32 flags;
368 __le32 mode;
369 __le32 stripe_unit; /* layout for newly created file */
370 __le32 stripe_count; /* ... */
371 __le32 object_size;
372 __le32 file_replication;
Sage Weil3469ac1a2012-05-07 15:33:36 -0700373 __le32 unused; /* used to be preferred osd */
Sage Weil0dee3c22009-10-06 11:31:06 -0700374 } __attribute__ ((packed)) open;
375 struct {
376 __le32 flags;
377 } __attribute__ ((packed)) setxattr;
378 struct {
379 struct ceph_file_layout layout;
380 } __attribute__ ((packed)) setlayout;
Greg Farnumfbaad972010-08-02 15:30:08 -0700381 struct {
382 __u8 rule; /* currently fcntl or flock */
383 __u8 type; /* shared, exclusive, remove*/
384 __le64 pid; /* process id requesting the lock */
385 __le64 pid_namespace;
386 __le64 start; /* initial location to lock */
387 __le64 length; /* num bytes to lock from start */
388 __u8 wait; /* will caller wait for lock to become available? */
389 } __attribute__ ((packed)) filelock_change;
Sage Weil0dee3c22009-10-06 11:31:06 -0700390} __attribute__ ((packed));
391
392#define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */
393#define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */
394
395struct ceph_mds_request_head {
Sage Weil6df058c2009-12-22 11:24:33 -0800396 __le64 oldest_client_tid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700397 __le32 mdsmap_epoch; /* on client */
398 __le32 flags; /* CEPH_MDS_FLAG_* */
399 __u8 num_retry, num_fwd; /* count retry, fwd attempts */
400 __le16 num_releases; /* # include cap/lease release records */
401 __le32 op; /* mds op code */
402 __le32 caller_uid, caller_gid;
403 __le64 ino; /* use this ino for openc, mkdir, mknod,
404 etc. (if replaying) */
405 union ceph_mds_request_args args;
406} __attribute__ ((packed));
407
408/* cap/lease release record */
409struct ceph_mds_request_release {
410 __le64 ino, cap_id; /* ino and unique cap id */
411 __le32 caps, wanted; /* new issued, wanted */
412 __le32 seq, issue_seq, mseq;
413 __le32 dname_seq; /* if releasing a dentry lease, a */
414 __le32 dname_len; /* string follows. */
415} __attribute__ ((packed));
416
417/* client reply */
418struct ceph_mds_reply_head {
Sage Weil0dee3c22009-10-06 11:31:06 -0700419 __le32 op;
420 __le32 result;
421 __le32 mdsmap_epoch;
422 __u8 safe; /* true if committed to disk */
423 __u8 is_dentry, is_target; /* true if dentry, target inode records
424 are included with reply */
425} __attribute__ ((packed));
426
427/* one for each node split */
428struct ceph_frag_tree_split {
429 __le32 frag; /* this frag splits... */
430 __le32 by; /* ...by this many bits */
431} __attribute__ ((packed));
432
433struct ceph_frag_tree_head {
434 __le32 nsplits; /* num ceph_frag_tree_split records */
435 struct ceph_frag_tree_split splits[];
436} __attribute__ ((packed));
437
438/* capability issue, for bundling with mds reply */
439struct ceph_mds_reply_cap {
440 __le32 caps, wanted; /* caps issued, wanted */
441 __le64 cap_id;
442 __le32 seq, mseq;
443 __le64 realm; /* snap realm */
444 __u8 flags; /* CEPH_CAP_FLAG_* */
445} __attribute__ ((packed));
446
447#define CEPH_CAP_FLAG_AUTH 1 /* cap is issued by auth mds */
448
449/* inode record, for bundling with mds reply */
450struct ceph_mds_reply_inode {
451 __le64 ino;
452 __le64 snapid;
453 __le32 rdev;
454 __le64 version; /* inode version */
455 __le64 xattr_version; /* version for xattr blob */
456 struct ceph_mds_reply_cap cap; /* caps issued for this inode */
457 struct ceph_file_layout layout;
458 struct ceph_timespec ctime, mtime, atime;
459 __le32 time_warp_seq;
460 __le64 size, max_size, truncate_size;
461 __le32 truncate_seq;
462 __le32 mode, uid, gid;
463 __le32 nlink;
464 __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */
465 struct ceph_timespec rctime;
466 struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */
467} __attribute__ ((packed));
Sage Weil6c0f3af2010-11-16 11:14:34 -0800468/* followed by frag array, symlink string, dir layout, xattr blob */
Sage Weil0dee3c22009-10-06 11:31:06 -0700469
470/* reply_lease follows dname, and reply_inode */
471struct ceph_mds_reply_lease {
472 __le16 mask; /* lease type(s) */
473 __le32 duration_ms; /* lease duration */
474 __le32 seq;
475} __attribute__ ((packed));
476
477struct ceph_mds_reply_dirfrag {
478 __le32 frag; /* fragment */
479 __le32 auth; /* auth mds, if this is a delegation point */
480 __le32 ndist; /* number of mds' this is replicated on */
481 __le32 dist[];
482} __attribute__ ((packed));
483
Greg Farnumfbaad972010-08-02 15:30:08 -0700484#define CEPH_LOCK_FCNTL 1
485#define CEPH_LOCK_FLOCK 2
486
487#define CEPH_LOCK_SHARED 1
488#define CEPH_LOCK_EXCL 2
489#define CEPH_LOCK_UNLOCK 4
490
491struct ceph_filelock {
492 __le64 start;/* file offset to start lock at */
493 __le64 length; /* num bytes to lock; 0 for all following start */
494 __le64 client; /* which client holds the lock */
495 __le64 pid; /* process id holding the lock on the client */
496 __le64 pid_namespace;
497 __u8 type; /* shared lock, exclusive lock, or unlock */
498} __attribute__ ((packed));
499
500
Sage Weil0dee3c22009-10-06 11:31:06 -0700501/* file access modes */
502#define CEPH_FILE_MODE_PIN 0
503#define CEPH_FILE_MODE_RD 1
504#define CEPH_FILE_MODE_WR 2
505#define CEPH_FILE_MODE_RDWR 3 /* RD | WR */
506#define CEPH_FILE_MODE_LAZY 4 /* lazy io */
507#define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */
508
509int ceph_flags_to_mode(int flags);
510
511
512/* capability bits */
513#define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */
514
515/* generic cap bits */
516#define CEPH_CAP_GSHARED 1 /* client can reads */
517#define CEPH_CAP_GEXCL 2 /* client can read and update */
518#define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */
519#define CEPH_CAP_GRD 8 /* (file) client can read */
520#define CEPH_CAP_GWR 16 /* (file) client can write */
521#define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */
522#define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */
523#define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */
524
525/* per-lock shift */
526#define CEPH_CAP_SAUTH 2
527#define CEPH_CAP_SLINK 4
528#define CEPH_CAP_SXATTR 6
Greg Farnumfbaad972010-08-02 15:30:08 -0700529#define CEPH_CAP_SFILE 8
530#define CEPH_CAP_SFLOCK 20
Sage Weil0dee3c22009-10-06 11:31:06 -0700531
Greg Farnumfbaad972010-08-02 15:30:08 -0700532#define CEPH_CAP_BITS 22
Sage Weil0dee3c22009-10-06 11:31:06 -0700533
534/* composed values */
535#define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH)
536#define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH)
537#define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK)
538#define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK)
539#define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR)
540#define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR)
541#define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE)
542#define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE)
543#define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE)
544#define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE)
545#define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE)
546#define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE)
547#define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE)
548#define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
549#define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE)
Greg Farnumfbaad972010-08-02 15:30:08 -0700550#define CEPH_CAP_FLOCK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFLOCK)
551#define CEPH_CAP_FLOCK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFLOCK)
552
Sage Weil0dee3c22009-10-06 11:31:06 -0700553
554/* cap masks (for getattr) */
555#define CEPH_STAT_CAP_INODE CEPH_CAP_PIN
556#define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */
557#define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN
558#define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED
559#define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED
560#define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED
561#define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED
562#define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED
563#define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED
564#define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED
565#define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */
566#define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED
567#define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \
568 CEPH_CAP_AUTH_SHARED | \
569 CEPH_CAP_LINK_SHARED | \
570 CEPH_CAP_FILE_SHARED | \
571 CEPH_CAP_XATTR_SHARED)
572
573#define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \
574 CEPH_CAP_LINK_SHARED | \
575 CEPH_CAP_XATTR_SHARED | \
576 CEPH_CAP_FILE_SHARED)
577#define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \
578 CEPH_CAP_FILE_CACHE)
579
580#define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \
581 CEPH_CAP_LINK_EXCL | \
582 CEPH_CAP_XATTR_EXCL | \
583 CEPH_CAP_FILE_EXCL)
584#define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \
585 CEPH_CAP_FILE_EXCL)
586#define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
587#define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
Sage Weil84d95092010-05-25 16:46:24 -0700588 CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \
589 CEPH_CAP_PIN)
Sage Weil0dee3c22009-10-06 11:31:06 -0700590
591#define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
592 CEPH_LOCK_IXATTR)
593
594int ceph_caps_for_mode(int mode);
595
596enum {
597 CEPH_CAP_OP_GRANT, /* mds->client grant */
598 CEPH_CAP_OP_REVOKE, /* mds->client revoke */
599 CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */
600 CEPH_CAP_OP_EXPORT, /* mds has exported the cap */
601 CEPH_CAP_OP_IMPORT, /* mds has imported the cap */
602 CEPH_CAP_OP_UPDATE, /* client->mds update */
603 CEPH_CAP_OP_DROP, /* client->mds drop cap bits */
604 CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */
605 CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */
606 CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */
607 CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
608 CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */
609 CEPH_CAP_OP_RENEW, /* client->mds renewal request */
610};
611
612extern const char *ceph_cap_op_name(int op);
613
614/*
615 * caps message, used for capability callbacks, acks, requests, etc.
616 */
617struct ceph_mds_caps {
618 __le32 op; /* CEPH_CAP_OP_* */
619 __le64 ino, realm;
620 __le64 cap_id;
621 __le32 seq, issue_seq;
622 __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
623 __le32 migrate_seq;
624 __le64 snap_follows;
625 __le32 snap_trace_len;
Sage Weil0dee3c22009-10-06 11:31:06 -0700626
627 /* authlock */
628 __le32 uid, gid, mode;
629
630 /* linklock */
631 __le32 nlink;
632
633 /* xattrlock */
634 __le32 xattr_len;
635 __le64 xattr_version;
636
637 /* filelock */
638 __le64 size, max_size, truncate_size;
639 __le32 truncate_seq;
640 struct ceph_timespec mtime, atime, ctime;
641 struct ceph_file_layout layout;
642 __le32 time_warp_seq;
643} __attribute__ ((packed));
644
645/* cap release msg head */
646struct ceph_mds_cap_release {
647 __le32 num; /* number of cap_items that follow */
648} __attribute__ ((packed));
649
650struct ceph_mds_cap_item {
651 __le64 ino;
652 __le64 cap_id;
653 __le32 migrate_seq, seq;
654} __attribute__ ((packed));
655
656#define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */
657#define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */
658#define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */
659#define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */
660
661extern const char *ceph_lease_op_name(int o);
662
663/* lease msg header */
664struct ceph_mds_lease {
665 __u8 action; /* CEPH_MDS_LEASE_* */
666 __le16 mask; /* which lease */
667 __le64 ino;
668 __le64 first, last; /* snap range */
669 __le32 seq;
670 __le32 duration_ms; /* duration of renewal */
671} __attribute__ ((packed));
672/* followed by a __le32+string for dname */
673
674/* client reconnect */
675struct ceph_mds_cap_reconnect {
676 __le64 cap_id;
677 __le32 wanted;
678 __le32 issued;
Sage Weil20cb34a2010-05-12 15:21:32 -0700679 __le64 snaprealm;
680 __le64 pathbase; /* base ino for our path to this ino */
681 __le32 flock_len; /* size of flock state blob, if any */
682} __attribute__ ((packed));
683/* followed by flock blob */
684
685struct ceph_mds_cap_reconnect_v1 {
686 __le64 cap_id;
687 __le32 wanted;
688 __le32 issued;
Sage Weil0dee3c22009-10-06 11:31:06 -0700689 __le64 size;
690 struct ceph_timespec mtime, atime;
691 __le64 snaprealm;
692 __le64 pathbase; /* base ino for our path to this ino */
693} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -0700694
695struct ceph_mds_snaprealm_reconnect {
696 __le64 ino; /* snap realm base */
697 __le64 seq; /* snap seq for this snap realm */
698 __le64 parent; /* parent realm */
699} __attribute__ ((packed));
700
701/*
702 * snaps
703 */
704enum {
705 CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */
706 CEPH_SNAP_OP_CREATE,
707 CEPH_SNAP_OP_DESTROY,
708 CEPH_SNAP_OP_SPLIT,
709};
710
711extern const char *ceph_snap_op_name(int o);
712
713/* snap msg header */
714struct ceph_mds_snap_head {
715 __le32 op; /* CEPH_SNAP_OP_* */
716 __le64 split; /* ino to split off, if any */
717 __le32 num_split_inos; /* # inos belonging to new child realm */
718 __le32 num_split_realms; /* # child realms udner new child realm */
719 __le32 trace_len; /* size of snap trace blob */
720} __attribute__ ((packed));
721/* followed by split ino list, then split realms, then the trace blob */
722
723/*
724 * encode info about a snaprealm, as viewed by a client
725 */
726struct ceph_mds_snap_realm {
727 __le64 ino; /* ino */
728 __le64 created; /* snap: when created */
729 __le64 parent; /* ino: parent realm */
730 __le64 parent_since; /* snap: same parent since */
731 __le64 seq; /* snap: version */
732 __le32 num_snaps;
733 __le32 num_prior_parent_snaps;
734} __attribute__ ((packed));
735/* followed by my snap list, then prior parent snap list */
736
737#endif