blob: 25bfb0eff7720b459ffefa45ec035ec0229c8b1c [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 */
Yehuda Sadeh0c948992010-02-01 16:10:45 -080024#define CEPH_OSDC_PROTOCOL 24 /* server/client */
Sage Weil3ea25f92010-01-25 11:18:17 -080025#define CEPH_MDSC_PROTOCOL 32 /* server/client */
Sage Weil13e38c82009-10-09 16:36:34 -070026#define CEPH_MONC_PROTOCOL 15 /* server/client */
Sage Weil0dee3c22009-10-06 11:31:06 -070027
28
Alex Elderdd6f5e12013-02-15 11:42:30 -060029#define CEPH_INO_ROOT 1
30#define CEPH_INO_CEPH 2 /* hidden .ceph dir */
31#define CEPH_INO_DOTDOT 3 /* used by ceph fuse for parent (..) */
Sage Weil0dee3c22009-10-06 11:31:06 -070032
33/* arbitrary limit on max # of monitors (cluster of 3 is typical) */
34#define CEPH_MAX_MON 31
35
Sage Weil0dee3c22009-10-06 11:31:06 -070036/*
37 * ceph_file_layout - describe data layout for a file/inode
38 */
39struct ceph_file_layout {
40 /* file -> object mapping */
41 __le32 fl_stripe_unit; /* stripe unit, in bytes. must be multiple
42 of page size. */
43 __le32 fl_stripe_count; /* over this many objects */
44 __le32 fl_object_size; /* until objects are this big, then move to
45 new objects */
Sage Weil6c0f3af2010-11-16 11:14:34 -080046 __le32 fl_cas_hash; /* UNUSED. 0 = none; 1 = sha256 */
Sage Weil0dee3c22009-10-06 11:31:06 -070047
48 /* pg -> disk layout */
Sage Weil6c0f3af2010-11-16 11:14:34 -080049 __le32 fl_object_stripe_unit; /* UNUSED. for per-object parity, if any */
Sage Weil0dee3c22009-10-06 11:31:06 -070050
51 /* object -> pg layout */
Alex Elderdd6f5e12013-02-15 11:42:30 -060052 __le32 fl_unused; /* unused; used to be preferred primary for pg (-1 for none) */
Sage Weil0dee3c22009-10-06 11:31:06 -070053 __le32 fl_pg_pool; /* namespace, crush ruleset, rep level */
54} __attribute__ ((packed));
55
Ilya Dryomove8221462014-01-27 17:40:18 +020056#define ceph_file_layout_su(l) ((__s32)le32_to_cpu((l).fl_stripe_unit))
57#define ceph_file_layout_stripe_count(l) \
58 ((__s32)le32_to_cpu((l).fl_stripe_count))
59#define ceph_file_layout_object_size(l) ((__s32)le32_to_cpu((l).fl_object_size))
60#define ceph_file_layout_cas_hash(l) ((__s32)le32_to_cpu((l).fl_cas_hash))
61#define ceph_file_layout_object_su(l) \
62 ((__s32)le32_to_cpu((l).fl_object_stripe_unit))
63#define ceph_file_layout_pg_pool(l) \
64 ((__s32)le32_to_cpu((l).fl_pg_pool))
65
66static inline unsigned ceph_file_layout_stripe_width(struct ceph_file_layout *l)
67{
68 return le32_to_cpu(l->fl_stripe_unit) *
69 le32_to_cpu(l->fl_stripe_count);
70}
71
72/* "period" == bytes before i start on a new set of objects */
73static inline unsigned ceph_file_layout_period(struct ceph_file_layout *l)
74{
75 return le32_to_cpu(l->fl_object_size) *
76 le32_to_cpu(l->fl_stripe_count);
77}
78
Sage Weil752727a2009-10-09 16:38:45 -070079#define CEPH_MIN_STRIPE_UNIT 65536
Sage Weil0dee3c22009-10-06 11:31:06 -070080
Sage Weil752727a2009-10-09 16:38:45 -070081int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
Sage Weil0dee3c22009-10-06 11:31:06 -070082
Sage Weil6c0f3af2010-11-16 11:14:34 -080083struct ceph_dir_layout {
84 __u8 dl_dir_hash; /* see ceph_hash.h for ids */
85 __u8 dl_unused1;
86 __u16 dl_unused2;
87 __u32 dl_unused3;
88} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -070089
Sage Weil4e7a5dc2009-11-18 16:19:57 -080090/* crypto algorithms */
91#define CEPH_CRYPTO_NONE 0x0
92#define CEPH_CRYPTO_AES 0x1
93
Sage Weilcbbfe492010-08-02 15:48:23 -070094#define CEPH_AES_IV "cephsageyudagreg"
95
Sage Weil4e7a5dc2009-11-18 16:19:57 -080096/* security/authentication protocols */
97#define CEPH_AUTH_UNKNOWN 0x0
98#define CEPH_AUTH_NONE 0x1
99#define CEPH_AUTH_CEPHX 0x2
100
Sage Weilca9d93a2010-05-12 14:48:20 -0700101#define CEPH_AUTH_UID_DEFAULT ((__u64) -1)
102
Sage Weil4e7a5dc2009-11-18 16:19:57 -0800103
Sage Weil0dee3c22009-10-06 11:31:06 -0700104/*********************************************
105 * message layer
106 */
107
108/*
109 * message types
110 */
111
112/* misc */
113#define CEPH_MSG_SHUTDOWN 1
114#define CEPH_MSG_PING 2
115
116/* client <-> monitor */
117#define CEPH_MSG_MON_MAP 4
118#define CEPH_MSG_MON_GET_MAP 5
Sage Weil0dee3c22009-10-06 11:31:06 -0700119#define CEPH_MSG_STATFS 13
120#define CEPH_MSG_STATFS_REPLY 14
121#define CEPH_MSG_MON_SUBSCRIBE 15
122#define CEPH_MSG_MON_SUBSCRIBE_ACK 16
Sage Weil4e7a5dc2009-11-18 16:19:57 -0800123#define CEPH_MSG_AUTH 17
124#define CEPH_MSG_AUTH_REPLY 18
Alex Elderdd6f5e12013-02-15 11:42:30 -0600125#define CEPH_MSG_MON_GET_VERSION 19
126#define CEPH_MSG_MON_GET_VERSION_REPLY 20
Sage Weil0dee3c22009-10-06 11:31:06 -0700127
128/* client <-> mds */
Sage Weil0dee3c22009-10-06 11:31:06 -0700129#define CEPH_MSG_MDS_MAP 21
130
131#define CEPH_MSG_CLIENT_SESSION 22
132#define CEPH_MSG_CLIENT_RECONNECT 23
133
134#define CEPH_MSG_CLIENT_REQUEST 24
135#define CEPH_MSG_CLIENT_REQUEST_FORWARD 25
136#define CEPH_MSG_CLIENT_REPLY 26
137#define CEPH_MSG_CLIENT_CAPS 0x310
138#define CEPH_MSG_CLIENT_LEASE 0x311
139#define CEPH_MSG_CLIENT_SNAP 0x312
140#define CEPH_MSG_CLIENT_CAPRELEASE 0x313
141
Sage Weilca9d93a2010-05-12 14:48:20 -0700142/* pool ops */
143#define CEPH_MSG_POOLOP_REPLY 48
144#define CEPH_MSG_POOLOP 49
145
146
Sage Weil0dee3c22009-10-06 11:31:06 -0700147/* osd */
Yehuda Sadeh483fac72011-01-20 16:36:06 -0800148#define CEPH_MSG_OSD_MAP 41
149#define CEPH_MSG_OSD_OP 42
150#define CEPH_MSG_OSD_OPREPLY 43
151#define CEPH_MSG_WATCH_NOTIFY 44
152
153
154/* watch-notify operations */
155enum {
156 WATCH_NOTIFY = 1, /* notifying watcher */
157 WATCH_NOTIFY_COMPLETE = 2, /* notifier notified when done */
158};
159
Sage Weil0dee3c22009-10-06 11:31:06 -0700160
Sage Weilca9d93a2010-05-12 14:48:20 -0700161/* pool operations */
162enum {
163 POOL_OP_CREATE = 0x01,
164 POOL_OP_DELETE = 0x02,
165 POOL_OP_AUID_CHANGE = 0x03,
166 POOL_OP_CREATE_SNAP = 0x11,
167 POOL_OP_DELETE_SNAP = 0x12,
168 POOL_OP_CREATE_UNMANAGED_SNAP = 0x21,
169 POOL_OP_DELETE_UNMANAGED_SNAP = 0x22,
170};
171
Sage Weil13e38c82009-10-09 16:36:34 -0700172struct ceph_mon_request_header {
173 __le64 have_version;
174 __le16 session_mon;
175 __le64 session_mon_tid;
176} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -0700177
178struct ceph_mon_statfs {
Sage Weil13e38c82009-10-09 16:36:34 -0700179 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700180 struct ceph_fsid fsid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700181} __attribute__ ((packed));
182
183struct ceph_statfs {
184 __le64 kb, kb_used, kb_avail;
185 __le64 num_objects;
186} __attribute__ ((packed));
187
188struct ceph_mon_statfs_reply {
189 struct ceph_fsid fsid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700190 __le64 version;
191 struct ceph_statfs st;
192} __attribute__ ((packed));
193
Sage Weilca9d93a2010-05-12 14:48:20 -0700194const char *ceph_pool_op_name(int op);
195
196struct ceph_mon_poolop {
197 struct ceph_mon_request_header monhdr;
198 struct ceph_fsid fsid;
199 __le32 pool;
200 __le32 op;
201 __le64 auid;
202 __le64 snapid;
203 __le32 name_len;
204} __attribute__ ((packed));
205
206struct ceph_mon_poolop_reply {
207 struct ceph_mon_request_header monhdr;
208 struct ceph_fsid fsid;
209 __le32 reply_code;
210 __le32 epoch;
211 char has_data;
212 char data[0];
213} __attribute__ ((packed));
214
215struct ceph_mon_unmanaged_snap {
216 __le64 snapid;
217} __attribute__ ((packed));
218
Sage Weil0dee3c22009-10-06 11:31:06 -0700219struct ceph_osd_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700220 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700221 struct ceph_fsid fsid;
222 __le32 start;
223} __attribute__ ((packed));
224
225struct ceph_mds_getmap {
Sage Weil13e38c82009-10-09 16:36:34 -0700226 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700227 struct ceph_fsid fsid;
228} __attribute__ ((packed));
229
230struct ceph_client_mount {
Sage Weil13e38c82009-10-09 16:36:34 -0700231 struct ceph_mon_request_header monhdr;
Sage Weil0dee3c22009-10-06 11:31:06 -0700232} __attribute__ ((packed));
233
Yehuda Sadeh483fac72011-01-20 16:36:06 -0800234#define CEPH_SUBSCRIBE_ONETIME 1 /* i want only 1 update after have */
235
Sage Weil0dee3c22009-10-06 11:31:06 -0700236struct ceph_mon_subscribe_item {
Yehuda Sadeh483fac72011-01-20 16:36:06 -0800237 __le64 have_version; __le64 have;
Sage Weil0dee3c22009-10-06 11:31:06 -0700238 __u8 onetime;
239} __attribute__ ((packed));
240
Sage Weil07bd10f2009-10-14 17:26:40 -0700241struct ceph_mon_subscribe_ack {
242 __le32 duration; /* seconds */
243 struct ceph_fsid fsid;
244} __attribute__ ((packed));
245
Sage Weil0dee3c22009-10-06 11:31:06 -0700246/*
Alex Elderdd6f5e12013-02-15 11:42:30 -0600247 * mdsmap flags
248 */
249#define CEPH_MDSMAP_DOWN (1<<0) /* cluster deliberately down */
250
251/*
Sage Weil0dee3c22009-10-06 11:31:06 -0700252 * mds states
253 * > 0 -> in
254 * <= 0 -> out
255 */
256#define CEPH_MDS_STATE_DNE 0 /* down, does not exist. */
257#define CEPH_MDS_STATE_STOPPED -1 /* down, once existed, but no subtrees.
258 empty log. */
259#define CEPH_MDS_STATE_BOOT -4 /* up, boot announcement. */
260#define CEPH_MDS_STATE_STANDBY -5 /* up, idle. waiting for assignment. */
261#define CEPH_MDS_STATE_CREATING -6 /* up, creating MDS instance. */
262#define CEPH_MDS_STATE_STARTING -7 /* up, starting previously stopped mds */
263#define CEPH_MDS_STATE_STANDBY_REPLAY -8 /* up, tailing active node's journal */
Alex Elderdd6f5e12013-02-15 11:42:30 -0600264#define CEPH_MDS_STATE_REPLAYONCE -9 /* up, replaying an active node's journal */
Sage Weil0dee3c22009-10-06 11:31:06 -0700265
266#define CEPH_MDS_STATE_REPLAY 8 /* up, replaying journal. */
267#define CEPH_MDS_STATE_RESOLVE 9 /* up, disambiguating distributed
268 operations (import, rename, etc.) */
269#define CEPH_MDS_STATE_RECONNECT 10 /* up, reconnect to clients */
270#define CEPH_MDS_STATE_REJOIN 11 /* up, rejoining distributed cache */
271#define CEPH_MDS_STATE_CLIENTREPLAY 12 /* up, replaying client operations */
272#define CEPH_MDS_STATE_ACTIVE 13 /* up, active */
273#define CEPH_MDS_STATE_STOPPING 14 /* up, but exporting metadata */
274
275extern const char *ceph_mds_state_name(int s);
276
277
278/*
279 * metadata lock types.
280 * - these are bitmasks.. we can compose them
281 * - they also define the lock ordering by the MDS
282 * - a few of these are internal to the mds
283 */
Sage Weildd1c9052010-05-25 16:45:25 -0700284#define CEPH_LOCK_DVERSION 1
285#define CEPH_LOCK_DN 2
286#define CEPH_LOCK_ISNAP 16
287#define CEPH_LOCK_IVERSION 32 /* mds internal */
288#define CEPH_LOCK_IFILE 64
289#define CEPH_LOCK_IAUTH 128
290#define CEPH_LOCK_ILINK 256
291#define CEPH_LOCK_IDFT 512 /* dir frag tree */
292#define CEPH_LOCK_INEST 1024 /* mds internal */
293#define CEPH_LOCK_IXATTR 2048
Sage Weilf0b18d92010-07-07 08:39:03 -0700294#define CEPH_LOCK_IFLOCK 4096 /* advisory file locks */
Sage Weildd1c9052010-05-25 16:45:25 -0700295#define CEPH_LOCK_INO 8192 /* immutable inode bits; not a lock */
Alex Elderdd6f5e12013-02-15 11:42:30 -0600296#define CEPH_LOCK_IPOLICY 16384 /* policy lock on dirs. MDS internal */
Sage Weil0dee3c22009-10-06 11:31:06 -0700297
298/* client_session ops */
299enum {
300 CEPH_SESSION_REQUEST_OPEN,
301 CEPH_SESSION_OPEN,
302 CEPH_SESSION_REQUEST_CLOSE,
303 CEPH_SESSION_CLOSE,
304 CEPH_SESSION_REQUEST_RENEWCAPS,
305 CEPH_SESSION_RENEWCAPS,
306 CEPH_SESSION_STALE,
307 CEPH_SESSION_RECALL_STATE,
Yan, Zheng186e4f72013-11-22 14:48:37 +0800308 CEPH_SESSION_FLUSHMSG,
309 CEPH_SESSION_FLUSHMSG_ACK,
Sage Weil0dee3c22009-10-06 11:31:06 -0700310};
311
312extern const char *ceph_session_op_name(int op);
313
314struct ceph_mds_session_head {
315 __le32 op;
316 __le64 seq;
317 struct ceph_timespec stamp;
318 __le32 max_caps, max_leases;
319} __attribute__ ((packed));
320
321/* client_request */
322/*
323 * metadata ops.
324 * & 0x001000 -> write op
325 * & 0x010000 -> follow symlink (e.g. stat(), not lstat()).
326 & & 0x100000 -> use weird ino/path trace
327 */
328#define CEPH_MDS_OP_WRITE 0x001000
329enum {
330 CEPH_MDS_OP_LOOKUP = 0x00100,
331 CEPH_MDS_OP_GETATTR = 0x00101,
332 CEPH_MDS_OP_LOOKUPHASH = 0x00102,
333 CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
Sage Weil3c454cf2011-04-06 09:31:40 -0700334 CEPH_MDS_OP_LOOKUPINO = 0x00104,
Sage Weil0dee3c22009-10-06 11:31:06 -0700335
336 CEPH_MDS_OP_SETXATTR = 0x01105,
337 CEPH_MDS_OP_RMXATTR = 0x01106,
338 CEPH_MDS_OP_SETLAYOUT = 0x01107,
339 CEPH_MDS_OP_SETATTR = 0x01108,
Greg Farnumfbaad972010-08-02 15:30:08 -0700340 CEPH_MDS_OP_SETFILELOCK= 0x01109,
341 CEPH_MDS_OP_GETFILELOCK= 0x00110,
Greg Farnum571dba52010-09-24 14:56:40 -0700342 CEPH_MDS_OP_SETDIRLAYOUT=0x0110a,
Sage Weil0dee3c22009-10-06 11:31:06 -0700343
344 CEPH_MDS_OP_MKNOD = 0x01201,
345 CEPH_MDS_OP_LINK = 0x01202,
346 CEPH_MDS_OP_UNLINK = 0x01203,
347 CEPH_MDS_OP_RENAME = 0x01204,
348 CEPH_MDS_OP_MKDIR = 0x01220,
349 CEPH_MDS_OP_RMDIR = 0x01221,
350 CEPH_MDS_OP_SYMLINK = 0x01222,
351
Sage Weil3ea25f92010-01-25 11:18:17 -0800352 CEPH_MDS_OP_CREATE = 0x01301,
Sage Weil0dee3c22009-10-06 11:31:06 -0700353 CEPH_MDS_OP_OPEN = 0x00302,
354 CEPH_MDS_OP_READDIR = 0x00305,
355
356 CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
357 CEPH_MDS_OP_MKSNAP = 0x01400,
358 CEPH_MDS_OP_RMSNAP = 0x01401,
359 CEPH_MDS_OP_LSSNAP = 0x00402,
360};
361
362extern const char *ceph_mds_op_name(int op);
363
364
365#define CEPH_SETATTR_MODE 1
366#define CEPH_SETATTR_UID 2
367#define CEPH_SETATTR_GID 4
368#define CEPH_SETATTR_MTIME 8
369#define CEPH_SETATTR_ATIME 16
370#define CEPH_SETATTR_SIZE 32
371#define CEPH_SETATTR_CTIME 64
372
Alex Elderdd6f5e12013-02-15 11:42:30 -0600373/*
374 * Ceph setxattr request flags.
375 */
Yan, Zhengbcdfeb22014-02-11 13:04:19 +0800376#define CEPH_XATTR_CREATE (1 << 0)
377#define CEPH_XATTR_REPLACE (1 << 1)
378#define CEPH_XATTR_REMOVE (1 << 31)
Alex Elderdd6f5e12013-02-15 11:42:30 -0600379
Sage Weil0dee3c22009-10-06 11:31:06 -0700380union ceph_mds_request_args {
381 struct {
382 __le32 mask; /* CEPH_CAP_* */
383 } __attribute__ ((packed)) getattr;
384 struct {
385 __le32 mode;
386 __le32 uid;
387 __le32 gid;
388 struct ceph_timespec mtime;
389 struct ceph_timespec atime;
390 __le64 size, old_size; /* old_size needed by truncate */
391 __le32 mask; /* CEPH_SETATTR_* */
392 } __attribute__ ((packed)) setattr;
393 struct {
394 __le32 frag; /* which dir fragment */
395 __le32 max_entries; /* how many dentries to grab */
Sage Weil23804d92010-05-14 13:06:30 -0700396 __le32 max_bytes;
Sage Weil0dee3c22009-10-06 11:31:06 -0700397 } __attribute__ ((packed)) readdir;
398 struct {
399 __le32 mode;
400 __le32 rdev;
401 } __attribute__ ((packed)) mknod;
402 struct {
403 __le32 mode;
404 } __attribute__ ((packed)) mkdir;
405 struct {
406 __le32 flags;
407 __le32 mode;
408 __le32 stripe_unit; /* layout for newly created file */
409 __le32 stripe_count; /* ... */
410 __le32 object_size;
411 __le32 file_replication;
Sage Weil3469ac1a2012-05-07 15:33:36 -0700412 __le32 unused; /* used to be preferred osd */
Sage Weil0dee3c22009-10-06 11:31:06 -0700413 } __attribute__ ((packed)) open;
414 struct {
415 __le32 flags;
416 } __attribute__ ((packed)) setxattr;
417 struct {
418 struct ceph_file_layout layout;
419 } __attribute__ ((packed)) setlayout;
Greg Farnumfbaad972010-08-02 15:30:08 -0700420 struct {
421 __u8 rule; /* currently fcntl or flock */
422 __u8 type; /* shared, exclusive, remove*/
423 __le64 pid; /* process id requesting the lock */
424 __le64 pid_namespace;
425 __le64 start; /* initial location to lock */
426 __le64 length; /* num bytes to lock from start */
427 __u8 wait; /* will caller wait for lock to become available? */
428 } __attribute__ ((packed)) filelock_change;
Sage Weil0dee3c22009-10-06 11:31:06 -0700429} __attribute__ ((packed));
430
431#define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */
432#define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */
433
434struct ceph_mds_request_head {
Sage Weil6df058c2009-12-22 11:24:33 -0800435 __le64 oldest_client_tid;
Sage Weil0dee3c22009-10-06 11:31:06 -0700436 __le32 mdsmap_epoch; /* on client */
437 __le32 flags; /* CEPH_MDS_FLAG_* */
438 __u8 num_retry, num_fwd; /* count retry, fwd attempts */
439 __le16 num_releases; /* # include cap/lease release records */
440 __le32 op; /* mds op code */
441 __le32 caller_uid, caller_gid;
442 __le64 ino; /* use this ino for openc, mkdir, mknod,
443 etc. (if replaying) */
444 union ceph_mds_request_args args;
445} __attribute__ ((packed));
446
447/* cap/lease release record */
448struct ceph_mds_request_release {
449 __le64 ino, cap_id; /* ino and unique cap id */
450 __le32 caps, wanted; /* new issued, wanted */
451 __le32 seq, issue_seq, mseq;
452 __le32 dname_seq; /* if releasing a dentry lease, a */
453 __le32 dname_len; /* string follows. */
454} __attribute__ ((packed));
455
456/* client reply */
457struct ceph_mds_reply_head {
Sage Weil0dee3c22009-10-06 11:31:06 -0700458 __le32 op;
459 __le32 result;
460 __le32 mdsmap_epoch;
461 __u8 safe; /* true if committed to disk */
462 __u8 is_dentry, is_target; /* true if dentry, target inode records
463 are included with reply */
464} __attribute__ ((packed));
465
466/* one for each node split */
467struct ceph_frag_tree_split {
468 __le32 frag; /* this frag splits... */
469 __le32 by; /* ...by this many bits */
470} __attribute__ ((packed));
471
472struct ceph_frag_tree_head {
473 __le32 nsplits; /* num ceph_frag_tree_split records */
474 struct ceph_frag_tree_split splits[];
475} __attribute__ ((packed));
476
477/* capability issue, for bundling with mds reply */
478struct ceph_mds_reply_cap {
479 __le32 caps, wanted; /* caps issued, wanted */
480 __le64 cap_id;
481 __le32 seq, mseq;
482 __le64 realm; /* snap realm */
483 __u8 flags; /* CEPH_CAP_FLAG_* */
484} __attribute__ ((packed));
485
Yan, Zheng4ee6a912013-11-24 14:43:46 +0800486#define CEPH_CAP_FLAG_AUTH (1 << 0) /* cap is issued by auth mds */
487#define CEPH_CAP_FLAG_RELEASE (1 << 1) /* release the cap */
Sage Weil0dee3c22009-10-06 11:31:06 -0700488
489/* inode record, for bundling with mds reply */
490struct ceph_mds_reply_inode {
491 __le64 ino;
492 __le64 snapid;
493 __le32 rdev;
494 __le64 version; /* inode version */
495 __le64 xattr_version; /* version for xattr blob */
496 struct ceph_mds_reply_cap cap; /* caps issued for this inode */
497 struct ceph_file_layout layout;
498 struct ceph_timespec ctime, mtime, atime;
499 __le32 time_warp_seq;
500 __le64 size, max_size, truncate_size;
501 __le32 truncate_seq;
502 __le32 mode, uid, gid;
503 __le32 nlink;
504 __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */
505 struct ceph_timespec rctime;
506 struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */
507} __attribute__ ((packed));
Sage Weil6c0f3af2010-11-16 11:14:34 -0800508/* followed by frag array, symlink string, dir layout, xattr blob */
Sage Weil0dee3c22009-10-06 11:31:06 -0700509
510/* reply_lease follows dname, and reply_inode */
511struct ceph_mds_reply_lease {
512 __le16 mask; /* lease type(s) */
513 __le32 duration_ms; /* lease duration */
514 __le32 seq;
515} __attribute__ ((packed));
516
517struct ceph_mds_reply_dirfrag {
518 __le32 frag; /* fragment */
519 __le32 auth; /* auth mds, if this is a delegation point */
520 __le32 ndist; /* number of mds' this is replicated on */
521 __le32 dist[];
522} __attribute__ ((packed));
523
Greg Farnumfbaad972010-08-02 15:30:08 -0700524#define CEPH_LOCK_FCNTL 1
525#define CEPH_LOCK_FLOCK 2
526
527#define CEPH_LOCK_SHARED 1
528#define CEPH_LOCK_EXCL 2
529#define CEPH_LOCK_UNLOCK 4
530
531struct ceph_filelock {
532 __le64 start;/* file offset to start lock at */
533 __le64 length; /* num bytes to lock; 0 for all following start */
534 __le64 client; /* which client holds the lock */
535 __le64 pid; /* process id holding the lock on the client */
536 __le64 pid_namespace;
537 __u8 type; /* shared lock, exclusive lock, or unlock */
538} __attribute__ ((packed));
539
540
Sage Weil0dee3c22009-10-06 11:31:06 -0700541/* file access modes */
542#define CEPH_FILE_MODE_PIN 0
543#define CEPH_FILE_MODE_RD 1
544#define CEPH_FILE_MODE_WR 2
545#define CEPH_FILE_MODE_RDWR 3 /* RD | WR */
546#define CEPH_FILE_MODE_LAZY 4 /* lazy io */
547#define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */
548
549int ceph_flags_to_mode(int flags);
550
551
552/* capability bits */
553#define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */
554
555/* generic cap bits */
556#define CEPH_CAP_GSHARED 1 /* client can reads */
557#define CEPH_CAP_GEXCL 2 /* client can read and update */
558#define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */
559#define CEPH_CAP_GRD 8 /* (file) client can read */
560#define CEPH_CAP_GWR 16 /* (file) client can write */
561#define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */
562#define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */
563#define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */
564
Alex Elderdd6f5e12013-02-15 11:42:30 -0600565#define CEPH_CAP_SIMPLE_BITS 2
566#define CEPH_CAP_FILE_BITS 8
567
Sage Weil0dee3c22009-10-06 11:31:06 -0700568/* per-lock shift */
569#define CEPH_CAP_SAUTH 2
570#define CEPH_CAP_SLINK 4
571#define CEPH_CAP_SXATTR 6
Greg Farnumfbaad972010-08-02 15:30:08 -0700572#define CEPH_CAP_SFILE 8
Alex Elderdd6f5e12013-02-15 11:42:30 -0600573#define CEPH_CAP_SFLOCK 20
Sage Weil0dee3c22009-10-06 11:31:06 -0700574
Alex Elderdd6f5e12013-02-15 11:42:30 -0600575#define CEPH_CAP_BITS 22
Sage Weil0dee3c22009-10-06 11:31:06 -0700576
577/* composed values */
578#define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH)
579#define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH)
580#define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK)
581#define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK)
582#define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR)
583#define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR)
584#define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE)
585#define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE)
586#define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE)
587#define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE)
588#define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE)
589#define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE)
590#define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE)
591#define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
592#define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE)
Greg Farnumfbaad972010-08-02 15:30:08 -0700593#define CEPH_CAP_FLOCK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFLOCK)
594#define CEPH_CAP_FLOCK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFLOCK)
595
Sage Weil0dee3c22009-10-06 11:31:06 -0700596
597/* cap masks (for getattr) */
598#define CEPH_STAT_CAP_INODE CEPH_CAP_PIN
599#define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */
600#define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN
601#define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED
602#define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED
603#define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED
604#define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED
605#define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED
606#define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED
607#define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED
608#define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */
609#define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED
610#define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \
611 CEPH_CAP_AUTH_SHARED | \
612 CEPH_CAP_LINK_SHARED | \
613 CEPH_CAP_FILE_SHARED | \
614 CEPH_CAP_XATTR_SHARED)
615
616#define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \
617 CEPH_CAP_LINK_SHARED | \
618 CEPH_CAP_XATTR_SHARED | \
619 CEPH_CAP_FILE_SHARED)
620#define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \
621 CEPH_CAP_FILE_CACHE)
622
623#define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \
624 CEPH_CAP_LINK_EXCL | \
625 CEPH_CAP_XATTR_EXCL | \
626 CEPH_CAP_FILE_EXCL)
627#define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \
628 CEPH_CAP_FILE_EXCL)
629#define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
630#define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
Sage Weil84d95092010-05-25 16:46:24 -0700631 CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \
632 CEPH_CAP_PIN)
Sage Weil0dee3c22009-10-06 11:31:06 -0700633
634#define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
635 CEPH_LOCK_IXATTR)
636
637int ceph_caps_for_mode(int mode);
638
639enum {
640 CEPH_CAP_OP_GRANT, /* mds->client grant */
641 CEPH_CAP_OP_REVOKE, /* mds->client revoke */
642 CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */
643 CEPH_CAP_OP_EXPORT, /* mds has exported the cap */
644 CEPH_CAP_OP_IMPORT, /* mds has imported the cap */
645 CEPH_CAP_OP_UPDATE, /* client->mds update */
646 CEPH_CAP_OP_DROP, /* client->mds drop cap bits */
647 CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */
648 CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */
649 CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */
650 CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
651 CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */
652 CEPH_CAP_OP_RENEW, /* client->mds renewal request */
653};
654
655extern const char *ceph_cap_op_name(int op);
656
657/*
658 * caps message, used for capability callbacks, acks, requests, etc.
659 */
660struct ceph_mds_caps {
661 __le32 op; /* CEPH_CAP_OP_* */
662 __le64 ino, realm;
663 __le64 cap_id;
664 __le32 seq, issue_seq;
665 __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
666 __le32 migrate_seq;
667 __le64 snap_follows;
668 __le32 snap_trace_len;
Sage Weil0dee3c22009-10-06 11:31:06 -0700669
670 /* authlock */
671 __le32 uid, gid, mode;
672
673 /* linklock */
674 __le32 nlink;
675
676 /* xattrlock */
677 __le32 xattr_len;
678 __le64 xattr_version;
679
680 /* filelock */
681 __le64 size, max_size, truncate_size;
682 __le32 truncate_seq;
683 struct ceph_timespec mtime, atime, ctime;
684 struct ceph_file_layout layout;
685 __le32 time_warp_seq;
686} __attribute__ ((packed));
687
Yan, Zheng4ee6a912013-11-24 14:43:46 +0800688struct ceph_mds_cap_peer {
689 __le64 cap_id;
690 __le32 seq;
691 __le32 mseq;
692 __le32 mds;
693 __u8 flags;
694} __attribute__ ((packed));
695
Sage Weil0dee3c22009-10-06 11:31:06 -0700696/* cap release msg head */
697struct ceph_mds_cap_release {
698 __le32 num; /* number of cap_items that follow */
699} __attribute__ ((packed));
700
701struct ceph_mds_cap_item {
702 __le64 ino;
703 __le64 cap_id;
704 __le32 migrate_seq, seq;
705} __attribute__ ((packed));
706
707#define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */
708#define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */
709#define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */
710#define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */
711
712extern const char *ceph_lease_op_name(int o);
713
714/* lease msg header */
715struct ceph_mds_lease {
716 __u8 action; /* CEPH_MDS_LEASE_* */
717 __le16 mask; /* which lease */
718 __le64 ino;
719 __le64 first, last; /* snap range */
720 __le32 seq;
721 __le32 duration_ms; /* duration of renewal */
722} __attribute__ ((packed));
723/* followed by a __le32+string for dname */
724
725/* client reconnect */
726struct ceph_mds_cap_reconnect {
727 __le64 cap_id;
728 __le32 wanted;
729 __le32 issued;
Sage Weil20cb34a2010-05-12 15:21:32 -0700730 __le64 snaprealm;
731 __le64 pathbase; /* base ino for our path to this ino */
732 __le32 flock_len; /* size of flock state blob, if any */
733} __attribute__ ((packed));
734/* followed by flock blob */
735
736struct ceph_mds_cap_reconnect_v1 {
737 __le64 cap_id;
738 __le32 wanted;
739 __le32 issued;
Sage Weil0dee3c22009-10-06 11:31:06 -0700740 __le64 size;
741 struct ceph_timespec mtime, atime;
742 __le64 snaprealm;
743 __le64 pathbase; /* base ino for our path to this ino */
744} __attribute__ ((packed));
Sage Weil0dee3c22009-10-06 11:31:06 -0700745
746struct ceph_mds_snaprealm_reconnect {
747 __le64 ino; /* snap realm base */
748 __le64 seq; /* snap seq for this snap realm */
749 __le64 parent; /* parent realm */
750} __attribute__ ((packed));
751
752/*
753 * snaps
754 */
755enum {
756 CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */
757 CEPH_SNAP_OP_CREATE,
758 CEPH_SNAP_OP_DESTROY,
759 CEPH_SNAP_OP_SPLIT,
760};
761
762extern const char *ceph_snap_op_name(int o);
763
764/* snap msg header */
765struct ceph_mds_snap_head {
766 __le32 op; /* CEPH_SNAP_OP_* */
767 __le64 split; /* ino to split off, if any */
768 __le32 num_split_inos; /* # inos belonging to new child realm */
769 __le32 num_split_realms; /* # child realms udner new child realm */
770 __le32 trace_len; /* size of snap trace blob */
771} __attribute__ ((packed));
772/* followed by split ino list, then split realms, then the trace blob */
773
774/*
775 * encode info about a snaprealm, as viewed by a client
776 */
777struct ceph_mds_snap_realm {
778 __le64 ino; /* ino */
779 __le64 created; /* snap: when created */
780 __le64 parent; /* ino: parent realm */
781 __le64 parent_since; /* snap: same parent since */
782 __le64 seq; /* snap: version */
783 __le32 num_snaps;
784 __le32 num_prior_parent_snaps;
785} __attribute__ ((packed));
786/* followed by my snap list, then prior parent snap list */
787
788#endif