]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - include/linux/ceph/ceph_fs.h
projects / mirror_ubuntu-bionic-kernel.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
rbd: use snprintf for disk->disk_name
[mirror_ubuntu-bionic-kernel.git] / include / linux / ceph / ceph_fs.h
CommitLineData
0dee3c28
SW		 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
5cd068c2
SW		 12#ifndef CEPH_FS_H
13#define CEPH_FS_H
0dee3c28
SW		 14
15#include "msgr.h"
16#include "rados.h"
17
0dee3c28
SW		 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 */
6df058c0 24#define CEPH_OSD_PROTOCOL 8 /* cluster internal */
ca9d93a2 25#define CEPH_MDS_PROTOCOL 12 /* cluster internal */
13e38c8a 26#define CEPH_MON_PROTOCOL 5 /* cluster internal */
0c948992 27#define CEPH_OSDC_PROTOCOL 24 /* server/client */
3ea25f94 28#define CEPH_MDSC_PROTOCOL 32 /* server/client */
13e38c8a 29#define CEPH_MONC_PROTOCOL 15 /* server/client */
0dee3c28
SW		 30
31
32#define CEPH_INO_ROOT 1
1d1de916 33#define CEPH_INO_CEPH 2 /* hidden .ceph dir */
0dee3c28
SW		 34
35/* arbitrary limit on max # of monitors (cluster of 3 is typical) */
36#define CEPH_MAX_MON 31
37
38
04a419f9
SW		 39/*
40 * feature bits
41 */
9688f19a
SW		 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)
6c0f3af7
SW		 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)
04a419f9 50
0dee3c28
SW		 51
52/*
53 * ceph_file_layout - describe data layout for a file/inode
54 */
55struct 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 */
6c0f3af7 62 __le32 fl_cas_hash; /* UNUSED. 0 = none; 1 = sha256 */
0dee3c28
SW		 63
64 /* pg -> disk layout */
6c0f3af7 65 __le32 fl_object_stripe_unit; /* UNUSED. for per-object parity, if any */
0dee3c28
SW		 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
752727a1 72#define CEPH_MIN_STRIPE_UNIT 65536
0dee3c28 73
752727a1 74int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
0dee3c28 75
6c0f3af7
SW		 76struct 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));
0dee3c28 82
4e7a5dcd
SW		 83/* crypto algorithms */
84#define CEPH_CRYPTO_NONE 0x0
85#define CEPH_CRYPTO_AES 0x1
86
cbbfe499
SW		 87#define CEPH_AES_IV "cephsageyudagreg"
88
4e7a5dcd
SW		 89/* security/authentication protocols */
90#define CEPH_AUTH_UNKNOWN 0x0
91#define CEPH_AUTH_NONE 0x1
92#define CEPH_AUTH_CEPHX 0x2
93
ca9d93a2
SW		 94#define CEPH_AUTH_UID_DEFAULT ((__u64) -1)
95
4e7a5dcd 96
0dee3c28
SW		 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
0dee3c28
SW		 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
4e7a5dcd
SW		 116#define CEPH_MSG_AUTH 17
117#define CEPH_MSG_AUTH_REPLY 18
0dee3c28
SW		 118
119/* client <-> mds */
0dee3c28
SW		 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
ca9d93a2
SW		 133/* pool ops */
134#define CEPH_MSG_POOLOP_REPLY 48
135#define CEPH_MSG_POOLOP 49
136
137
0dee3c28 138/* osd */
483fac71
YS		 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 */
146enum {
147 WATCH_NOTIFY = 1, /* notifying watcher */
148 WATCH_NOTIFY_COMPLETE = 2, /* notifier notified when done */
149};
150
0dee3c28 151
ca9d93a2
SW		 152/* pool operations */
153enum {
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
13e38c8a
SW		 163struct ceph_mon_request_header {
164 __le64 have_version;
165 __le16 session_mon;
166 __le64 session_mon_tid;
167} __attribute__ ((packed));
0dee3c28
SW		 168
169struct ceph_mon_statfs {
13e38c8a 170 struct ceph_mon_request_header monhdr;
0dee3c28 171 struct ceph_fsid fsid;
0dee3c28
SW		 172} __attribute__ ((packed));
173
174struct ceph_statfs {
175 __le64 kb, kb_used, kb_avail;
176 __le64 num_objects;
177} __attribute__ ((packed));
178
179struct ceph_mon_statfs_reply {
180 struct ceph_fsid fsid;
0dee3c28
SW		 181 __le64 version;
182 struct ceph_statfs st;
183} __attribute__ ((packed));
184
ca9d93a2
SW		 185const char *ceph_pool_op_name(int op);
186
187struct 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
197struct 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
206struct ceph_mon_unmanaged_snap {
207 __le64 snapid;
208} __attribute__ ((packed));
209
0dee3c28 210struct ceph_osd_getmap {
13e38c8a 211 struct ceph_mon_request_header monhdr;
0dee3c28
SW		 212 struct ceph_fsid fsid;
213 __le32 start;
214} __attribute__ ((packed));
215
216struct ceph_mds_getmap {
13e38c8a 217 struct ceph_mon_request_header monhdr;
0dee3c28
SW		 218 struct ceph_fsid fsid;
219} __attribute__ ((packed));
220
221struct ceph_client_mount {
13e38c8a 222 struct ceph_mon_request_header monhdr;
0dee3c28
SW		 223} __attribute__ ((packed));
224
483fac71
YS		 225#define CEPH_SUBSCRIBE_ONETIME 1 /* i want only 1 update after have */
226
0dee3c28 227struct ceph_mon_subscribe_item {
483fac71 228 __le64 have_version; __le64 have;
0dee3c28
SW		 229 __u8 onetime;
230} __attribute__ ((packed));
231
07bd10fb
SW		 232struct ceph_mon_subscribe_ack {
233 __le32 duration; /* seconds */
234 struct ceph_fsid fsid;
235} __attribute__ ((packed));
236
0dee3c28
SW		 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
260extern 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 */
dd1c9057
SW		 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
f0b18d9f 279#define CEPH_LOCK_IFLOCK 4096 /* advisory file locks */
dd1c9057 280#define CEPH_LOCK_INO 8192 /* immutable inode bits; not a lock */
0dee3c28
SW		 281
282/* client_session ops */
283enum {
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
294extern const char *ceph_session_op_name(int op);
295
296struct 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
311enum {
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
317 CEPH_MDS_OP_SETXATTR = 0x01105,
318 CEPH_MDS_OP_RMXATTR = 0x01106,
319 CEPH_MDS_OP_SETLAYOUT = 0x01107,
320 CEPH_MDS_OP_SETATTR = 0x01108,
fbaad979
GF		 321 CEPH_MDS_OP_SETFILELOCK= 0x01109,
322 CEPH_MDS_OP_GETFILELOCK= 0x00110,
571dba52 323 CEPH_MDS_OP_SETDIRLAYOUT=0x0110a,
0dee3c28
SW		 324
325 CEPH_MDS_OP_MKNOD = 0x01201,
326 CEPH_MDS_OP_LINK = 0x01202,
327 CEPH_MDS_OP_UNLINK = 0x01203,
328 CEPH_MDS_OP_RENAME = 0x01204,
329 CEPH_MDS_OP_MKDIR = 0x01220,
330 CEPH_MDS_OP_RMDIR = 0x01221,
331 CEPH_MDS_OP_SYMLINK = 0x01222,
332
3ea25f94 333 CEPH_MDS_OP_CREATE = 0x01301,
0dee3c28
SW		 334 CEPH_MDS_OP_OPEN = 0x00302,
335 CEPH_MDS_OP_READDIR = 0x00305,
336
337 CEPH_MDS_OP_LOOKUPSNAP = 0x00400,
338 CEPH_MDS_OP_MKSNAP = 0x01400,
339 CEPH_MDS_OP_RMSNAP = 0x01401,
340 CEPH_MDS_OP_LSSNAP = 0x00402,
341};
342
343extern const char *ceph_mds_op_name(int op);
344
345
346#define CEPH_SETATTR_MODE 1
347#define CEPH_SETATTR_UID 2
348#define CEPH_SETATTR_GID 4
349#define CEPH_SETATTR_MTIME 8
350#define CEPH_SETATTR_ATIME 16
351#define CEPH_SETATTR_SIZE 32
352#define CEPH_SETATTR_CTIME 64
353
354union ceph_mds_request_args {
355 struct {
356 __le32 mask; /* CEPH_CAP_* */
357 } __attribute__ ((packed)) getattr;
358 struct {
359 __le32 mode;
360 __le32 uid;
361 __le32 gid;
362 struct ceph_timespec mtime;
363 struct ceph_timespec atime;
364 __le64 size, old_size; /* old_size needed by truncate */
365 __le32 mask; /* CEPH_SETATTR_* */
366 } __attribute__ ((packed)) setattr;
367 struct {
368 __le32 frag; /* which dir fragment */
369 __le32 max_entries; /* how many dentries to grab */
23804d91 370 __le32 max_bytes;
0dee3c28
SW		 371 } __attribute__ ((packed)) readdir;
372 struct {
373 __le32 mode;
374 __le32 rdev;
375 } __attribute__ ((packed)) mknod;
376 struct {
377 __le32 mode;
378 } __attribute__ ((packed)) mkdir;
379 struct {
380 __le32 flags;
381 __le32 mode;
382 __le32 stripe_unit; /* layout for newly created file */
383 __le32 stripe_count; /* ... */
384 __le32 object_size;
385 __le32 file_replication;
386 __le32 preferred;
387 } __attribute__ ((packed)) open;
388 struct {
389 __le32 flags;
390 } __attribute__ ((packed)) setxattr;
391 struct {
392 struct ceph_file_layout layout;
393 } __attribute__ ((packed)) setlayout;
fbaad979
GF		 394 struct {
395 __u8 rule; /* currently fcntl or flock */
396 __u8 type; /* shared, exclusive, remove*/
397 __le64 pid; /* process id requesting the lock */
398 __le64 pid_namespace;
399 __le64 start; /* initial location to lock */
400 __le64 length; /* num bytes to lock from start */
401 __u8 wait; /* will caller wait for lock to become available? */
402 } __attribute__ ((packed)) filelock_change;
0dee3c28
SW		 403} __attribute__ ((packed));
404
405#define CEPH_MDS_FLAG_REPLAY 1 /* this is a replayed op */
406#define CEPH_MDS_FLAG_WANT_DENTRY 2 /* want dentry in reply */
407
408struct ceph_mds_request_head {
6df058c0 409 __le64 oldest_client_tid;
0dee3c28
SW		 410 __le32 mdsmap_epoch; /* on client */
411 __le32 flags; /* CEPH_MDS_FLAG_* */
412 __u8 num_retry, num_fwd; /* count retry, fwd attempts */
413 __le16 num_releases; /* # include cap/lease release records */
414 __le32 op; /* mds op code */
415 __le32 caller_uid, caller_gid;
416 __le64 ino; /* use this ino for openc, mkdir, mknod,
417 etc. (if replaying) */
418 union ceph_mds_request_args args;
419} __attribute__ ((packed));
420
421/* cap/lease release record */
422struct ceph_mds_request_release {
423 __le64 ino, cap_id; /* ino and unique cap id */
424 __le32 caps, wanted; /* new issued, wanted */
425 __le32 seq, issue_seq, mseq;
426 __le32 dname_seq; /* if releasing a dentry lease, a */
427 __le32 dname_len; /* string follows. */
428} __attribute__ ((packed));
429
430/* client reply */
431struct ceph_mds_reply_head {
0dee3c28
SW		 432 __le32 op;
433 __le32 result;
434 __le32 mdsmap_epoch;
435 __u8 safe; /* true if committed to disk */
436 __u8 is_dentry, is_target; /* true if dentry, target inode records
437 are included with reply */
438} __attribute__ ((packed));
439
440/* one for each node split */
441struct ceph_frag_tree_split {
442 __le32 frag; /* this frag splits... */
443 __le32 by; /* ...by this many bits */
444} __attribute__ ((packed));
445
446struct ceph_frag_tree_head {
447 __le32 nsplits; /* num ceph_frag_tree_split records */
448 struct ceph_frag_tree_split splits[];
449} __attribute__ ((packed));
450
451/* capability issue, for bundling with mds reply */
452struct ceph_mds_reply_cap {
453 __le32 caps, wanted; /* caps issued, wanted */
454 __le64 cap_id;
455 __le32 seq, mseq;
456 __le64 realm; /* snap realm */
457 __u8 flags; /* CEPH_CAP_FLAG_* */
458} __attribute__ ((packed));
459
460#define CEPH_CAP_FLAG_AUTH 1 /* cap is issued by auth mds */
461
462/* inode record, for bundling with mds reply */
463struct ceph_mds_reply_inode {
464 __le64 ino;
465 __le64 snapid;
466 __le32 rdev;
467 __le64 version; /* inode version */
468 __le64 xattr_version; /* version for xattr blob */
469 struct ceph_mds_reply_cap cap; /* caps issued for this inode */
470 struct ceph_file_layout layout;
471 struct ceph_timespec ctime, mtime, atime;
472 __le32 time_warp_seq;
473 __le64 size, max_size, truncate_size;
474 __le32 truncate_seq;
475 __le32 mode, uid, gid;
476 __le32 nlink;
477 __le64 files, subdirs, rbytes, rfiles, rsubdirs; /* dir stats */
478 struct ceph_timespec rctime;
479 struct ceph_frag_tree_head fragtree; /* (must be at end of struct) */
480} __attribute__ ((packed));
6c0f3af7 481/* followed by frag array, symlink string, dir layout, xattr blob */
0dee3c28
SW		 482
483/* reply_lease follows dname, and reply_inode */
484struct ceph_mds_reply_lease {
485 __le16 mask; /* lease type(s) */
486 __le32 duration_ms; /* lease duration */
487 __le32 seq;
488} __attribute__ ((packed));
489
490struct ceph_mds_reply_dirfrag {
491 __le32 frag; /* fragment */
492 __le32 auth; /* auth mds, if this is a delegation point */
493 __le32 ndist; /* number of mds' this is replicated on */
494 __le32 dist[];
495} __attribute__ ((packed));
496
fbaad979
GF		 497#define CEPH_LOCK_FCNTL 1
498#define CEPH_LOCK_FLOCK 2
499
500#define CEPH_LOCK_SHARED 1
501#define CEPH_LOCK_EXCL 2
502#define CEPH_LOCK_UNLOCK 4
503
504struct ceph_filelock {
505 __le64 start;/* file offset to start lock at */
506 __le64 length; /* num bytes to lock; 0 for all following start */
507 __le64 client; /* which client holds the lock */
508 __le64 pid; /* process id holding the lock on the client */
509 __le64 pid_namespace;
510 __u8 type; /* shared lock, exclusive lock, or unlock */
511} __attribute__ ((packed));
512
513
0dee3c28
SW		 514/* file access modes */
515#define CEPH_FILE_MODE_PIN 0
516#define CEPH_FILE_MODE_RD 1
517#define CEPH_FILE_MODE_WR 2
518#define CEPH_FILE_MODE_RDWR 3 /* RD | WR */
519#define CEPH_FILE_MODE_LAZY 4 /* lazy io */
520#define CEPH_FILE_MODE_NUM 8 /* bc these are bit fields.. mostly */
521
522int ceph_flags_to_mode(int flags);
523
524
525/* capability bits */
526#define CEPH_CAP_PIN 1 /* no specific capabilities beyond the pin */
527
528/* generic cap bits */
529#define CEPH_CAP_GSHARED 1 /* client can reads */
530#define CEPH_CAP_GEXCL 2 /* client can read and update */
531#define CEPH_CAP_GCACHE 4 /* (file) client can cache reads */
532#define CEPH_CAP_GRD 8 /* (file) client can read */
533#define CEPH_CAP_GWR 16 /* (file) client can write */
534#define CEPH_CAP_GBUFFER 32 /* (file) client can buffer writes */
535#define CEPH_CAP_GWREXTEND 64 /* (file) client can extend EOF */
536#define CEPH_CAP_GLAZYIO 128 /* (file) client can perform lazy io */
537
538/* per-lock shift */
539#define CEPH_CAP_SAUTH 2
540#define CEPH_CAP_SLINK 4
541#define CEPH_CAP_SXATTR 6
fbaad979
GF		 542#define CEPH_CAP_SFILE 8
543#define CEPH_CAP_SFLOCK 20
0dee3c28 544
fbaad979 545#define CEPH_CAP_BITS 22
0dee3c28
SW		 546
547/* composed values */
548#define CEPH_CAP_AUTH_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SAUTH)
549#define CEPH_CAP_AUTH_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SAUTH)
550#define CEPH_CAP_LINK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SLINK)
551#define CEPH_CAP_LINK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SLINK)
552#define CEPH_CAP_XATTR_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SXATTR)
553#define CEPH_CAP_XATTR_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SXATTR)
554#define CEPH_CAP_FILE(x) (x << CEPH_CAP_SFILE)
555#define CEPH_CAP_FILE_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFILE)
556#define CEPH_CAP_FILE_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFILE)
557#define CEPH_CAP_FILE_CACHE (CEPH_CAP_GCACHE << CEPH_CAP_SFILE)
558#define CEPH_CAP_FILE_RD (CEPH_CAP_GRD << CEPH_CAP_SFILE)
559#define CEPH_CAP_FILE_WR (CEPH_CAP_GWR << CEPH_CAP_SFILE)
560#define CEPH_CAP_FILE_BUFFER (CEPH_CAP_GBUFFER << CEPH_CAP_SFILE)
561#define CEPH_CAP_FILE_WREXTEND (CEPH_CAP_GWREXTEND << CEPH_CAP_SFILE)
562#define CEPH_CAP_FILE_LAZYIO (CEPH_CAP_GLAZYIO << CEPH_CAP_SFILE)
fbaad979
GF		 563#define CEPH_CAP_FLOCK_SHARED (CEPH_CAP_GSHARED << CEPH_CAP_SFLOCK)
564#define CEPH_CAP_FLOCK_EXCL (CEPH_CAP_GEXCL << CEPH_CAP_SFLOCK)
565
0dee3c28
SW		 566
567/* cap masks (for getattr) */
568#define CEPH_STAT_CAP_INODE CEPH_CAP_PIN
569#define CEPH_STAT_CAP_TYPE CEPH_CAP_PIN /* mode >> 12 */
570#define CEPH_STAT_CAP_SYMLINK CEPH_CAP_PIN
571#define CEPH_STAT_CAP_UID CEPH_CAP_AUTH_SHARED
572#define CEPH_STAT_CAP_GID CEPH_CAP_AUTH_SHARED
573#define CEPH_STAT_CAP_MODE CEPH_CAP_AUTH_SHARED
574#define CEPH_STAT_CAP_NLINK CEPH_CAP_LINK_SHARED
575#define CEPH_STAT_CAP_LAYOUT CEPH_CAP_FILE_SHARED
576#define CEPH_STAT_CAP_MTIME CEPH_CAP_FILE_SHARED
577#define CEPH_STAT_CAP_SIZE CEPH_CAP_FILE_SHARED
578#define CEPH_STAT_CAP_ATIME CEPH_CAP_FILE_SHARED /* fixme */
579#define CEPH_STAT_CAP_XATTR CEPH_CAP_XATTR_SHARED
580#define CEPH_STAT_CAP_INODE_ALL (CEPH_CAP_PIN | \
581 CEPH_CAP_AUTH_SHARED | \
582 CEPH_CAP_LINK_SHARED | \
583 CEPH_CAP_FILE_SHARED | \
584 CEPH_CAP_XATTR_SHARED)
585
586#define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED | \
587 CEPH_CAP_LINK_SHARED | \
588 CEPH_CAP_XATTR_SHARED | \
589 CEPH_CAP_FILE_SHARED)
590#define CEPH_CAP_ANY_RD (CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_RD | \
591 CEPH_CAP_FILE_CACHE)
592
593#define CEPH_CAP_ANY_EXCL (CEPH_CAP_AUTH_EXCL | \
594 CEPH_CAP_LINK_EXCL | \
595 CEPH_CAP_XATTR_EXCL | \
596 CEPH_CAP_FILE_EXCL)
597#define CEPH_CAP_ANY_FILE_WR (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_BUFFER | \
598 CEPH_CAP_FILE_EXCL)
599#define CEPH_CAP_ANY_WR (CEPH_CAP_ANY_EXCL | CEPH_CAP_ANY_FILE_WR)
600#define CEPH_CAP_ANY (CEPH_CAP_ANY_RD | CEPH_CAP_ANY_EXCL | \
84d95092
SW		 601 CEPH_CAP_ANY_FILE_WR | CEPH_CAP_FILE_LAZYIO | \
602 CEPH_CAP_PIN)
0dee3c28
SW		 603
604#define CEPH_CAP_LOCKS (CEPH_LOCK_IFILE | CEPH_LOCK_IAUTH | CEPH_LOCK_ILINK | \
605 CEPH_LOCK_IXATTR)
606
607int ceph_caps_for_mode(int mode);
608
609enum {
610 CEPH_CAP_OP_GRANT, /* mds->client grant */
611 CEPH_CAP_OP_REVOKE, /* mds->client revoke */
612 CEPH_CAP_OP_TRUNC, /* mds->client trunc notify */
613 CEPH_CAP_OP_EXPORT, /* mds has exported the cap */
614 CEPH_CAP_OP_IMPORT, /* mds has imported the cap */
615 CEPH_CAP_OP_UPDATE, /* client->mds update */
616 CEPH_CAP_OP_DROP, /* client->mds drop cap bits */
617 CEPH_CAP_OP_FLUSH, /* client->mds cap writeback */
618 CEPH_CAP_OP_FLUSH_ACK, /* mds->client flushed */
619 CEPH_CAP_OP_FLUSHSNAP, /* client->mds flush snapped metadata */
620 CEPH_CAP_OP_FLUSHSNAP_ACK, /* mds->client flushed snapped metadata */
621 CEPH_CAP_OP_RELEASE, /* client->mds release (clean) cap */
622 CEPH_CAP_OP_RENEW, /* client->mds renewal request */
623};
624
625extern const char *ceph_cap_op_name(int op);
626
627/*
628 * caps message, used for capability callbacks, acks, requests, etc.
629 */
630struct ceph_mds_caps {
631 __le32 op; /* CEPH_CAP_OP_* */
632 __le64 ino, realm;
633 __le64 cap_id;
634 __le32 seq, issue_seq;
635 __le32 caps, wanted, dirty; /* latest issued/wanted/dirty */
636 __le32 migrate_seq;
637 __le64 snap_follows;
638 __le32 snap_trace_len;
0dee3c28
SW		 639
640 /* authlock */
641 __le32 uid, gid, mode;
642
643 /* linklock */
644 __le32 nlink;
645
646 /* xattrlock */
647 __le32 xattr_len;
648 __le64 xattr_version;
649
650 /* filelock */
651 __le64 size, max_size, truncate_size;
652 __le32 truncate_seq;
653 struct ceph_timespec mtime, atime, ctime;
654 struct ceph_file_layout layout;
655 __le32 time_warp_seq;
656} __attribute__ ((packed));
657
658/* cap release msg head */
659struct ceph_mds_cap_release {
660 __le32 num; /* number of cap_items that follow */
661} __attribute__ ((packed));
662
663struct ceph_mds_cap_item {
664 __le64 ino;
665 __le64 cap_id;
666 __le32 migrate_seq, seq;
667} __attribute__ ((packed));
668
669#define CEPH_MDS_LEASE_REVOKE 1 /* mds -> client */
670#define CEPH_MDS_LEASE_RELEASE 2 /* client -> mds */
671#define CEPH_MDS_LEASE_RENEW 3 /* client <-> mds */
672#define CEPH_MDS_LEASE_REVOKE_ACK 4 /* client -> mds */
673
674extern const char *ceph_lease_op_name(int o);
675
676/* lease msg header */
677struct ceph_mds_lease {
678 __u8 action; /* CEPH_MDS_LEASE_* */
679 __le16 mask; /* which lease */
680 __le64 ino;
681 __le64 first, last; /* snap range */
682 __le32 seq;
683 __le32 duration_ms; /* duration of renewal */
684} __attribute__ ((packed));
685/* followed by a __le32+string for dname */
686
687/* client reconnect */
688struct ceph_mds_cap_reconnect {
20cb34ae
SW		 689 __le64 cap_id;
690 __le32 wanted;
691 __le32 issued;
692 __le64 snaprealm;
693 __le64 pathbase; /* base ino for our path to this ino */
694 __le32 flock_len; /* size of flock state blob, if any */
695} __attribute__ ((packed));
696/* followed by flock blob */
697
698struct ceph_mds_cap_reconnect_v1 {
0dee3c28
SW		 699 __le64 cap_id;
700 __le32 wanted;
701 __le32 issued;
702 __le64 size;
703 struct ceph_timespec mtime, atime;
704 __le64 snaprealm;
705 __le64 pathbase; /* base ino for our path to this ino */
706} __attribute__ ((packed));
0dee3c28
SW		 707
708struct ceph_mds_snaprealm_reconnect {
709 __le64 ino; /* snap realm base */
710 __le64 seq; /* snap seq for this snap realm */
711 __le64 parent; /* parent realm */
712} __attribute__ ((packed));
713
714/*
715 * snaps
716 */
717enum {
718 CEPH_SNAP_OP_UPDATE, /* CREATE or DESTROY */
719 CEPH_SNAP_OP_CREATE,
720 CEPH_SNAP_OP_DESTROY,
721 CEPH_SNAP_OP_SPLIT,
722};
723
724extern const char *ceph_snap_op_name(int o);
725
726/* snap msg header */
727struct ceph_mds_snap_head {
728 __le32 op; /* CEPH_SNAP_OP_* */
729 __le64 split; /* ino to split off, if any */
730 __le32 num_split_inos; /* # inos belonging to new child realm */
731 __le32 num_split_realms; /* # child realms udner new child realm */
732 __le32 trace_len; /* size of snap trace blob */
733} __attribute__ ((packed));
734/* followed by split ino list, then split realms, then the trace blob */
735
736/*
737 * encode info about a snaprealm, as viewed by a client
738 */
739struct ceph_mds_snap_realm {
740 __le64 ino; /* ino */
741 __le64 created; /* snap: when created */
742 __le64 parent; /* ino: parent realm */
743 __le64 parent_since; /* snap: same parent since */
744 __le64 seq; /* snap: version */
745 __le32 num_snaps;
746 __le32 num_prior_parent_snaps;
747} __attribute__ ((packed));
748/* followed by my snap list, then prior parent snap list */
749
750#endif
ubuntu-bionic-kernel mirror