1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
4 * Copyright (C) 2001-2003 Cluster File Systems, Inc. <info@clusterfs.com>
6 * This file is part of Lustre, http://www.lustre.org.
8 * Lustre is free software; you can redistribute it and/or
9 * modify it under the terms of version 2 of the GNU General Public
10 * License as published by the Free Software Foundation.
12 * Lustre is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Lustre; if not, write to the Free Software
19 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
21 * MDS data structures.
22 * See also lustre_idl.h for wire formats of requests.
30 # include <linux/fs.h>
31 # include <linux/dcache.h>
34 #include <linux/lustre_handles.h>
35 #include <linux/kp30.h>
36 #include <linux/lustre_idl.h>
37 #include <linux/lustre_lib.h>
38 #include <linux/lustre_dlm.h>
39 #include <linux/lustre_log.h>
40 #include <linux/lustre_export.h>
42 struct ldlm_lock_desc;
44 struct ptlrpc_connection;
47 struct ptlrpc_request;
51 #define LUSTRE_MDS_NAME "mds"
52 #define LUSTRE_MDT_NAME "mdt"
53 #define LUSTRE_MDC_NAME "mdc"
55 struct mds_update_record {
57 struct lustre_id *ur_id1;
58 struct lustre_id *ur_id2;
66 struct llog_cookie *ur_logcookies;
67 struct iattr ur_iattr;
68 struct lvfs_ucred ur_uc;
75 #define ur_fsuid ur_uc.luc_fsuid
76 #define ur_fsgid ur_uc.luc_fsgid
77 #define ur_cap ur_uc.luc_cap
78 #define ur_uid ur_uc.luc_uid
81 #define MDS_LR_SERVER_SIZE 512
83 #define MDS_LR_CLIENT_START 8192
84 #define MDS_LR_CLIENT_SIZE 128
85 #if MDS_LR_CLIENT_START < MDS_LR_SERVER_SIZE
86 #error "Can't have MDS_LR_CLIENT_START < MDS_LR_SERVER_SIZE"
89 #define MDS_CLIENT_SLOTS 17
91 #define MDS_ROCOMPAT_LOVOBJID 0x00000001
92 #define MDS_ROCOMPAT_SUPP (MDS_ROCOMPAT_LOVOBJID)
94 #define MDS_INCOMPAT_SUPP (0)
96 #define MDS_MASTER_OBD 1
97 #define MDS_CACHE_OBD 0
99 /*flags for indicate the record are come from cmobd reint or
101 #define REC_REINT_CREATE 0x0001
103 /* Data stored per server at the head of the last_rcvd file. In le32 order.
104 * Try to keep this the same as fsd_server_data so we might one day merge. */
105 struct mds_server_data {
106 __u8 msd_uuid[40]; /* server UUID */
107 __u64 msd_last_transno; /* last completed transaction ID */
108 __u64 msd_mount_count; /* MDS incarnation number */
110 __u32 msd_feature_compat; /* compatible feature flags */
111 __u32 msd_feature_rocompat;/* read-only compatible feature flags */
112 __u32 msd_feature_incompat;/* incompatible feature flags */
113 __u32 msd_server_size; /* size of server data area */
114 __u32 msd_client_start; /* start of per-client data area */
115 __u16 msd_client_size; /* size of per-client data area */
116 __u16 msd_subdir_count; /* number of subdirectories for objects */
117 __u64 msd_catalog_oid; /* recovery catalog object id */
118 __u32 msd_catalog_ogen; /* recovery catalog inode generation */
119 __u8 msd_peeruuid[40]; /* UUID of LOV/OSC associated with MDS */
120 __u8 msd_padding[MDS_LR_SERVER_SIZE - 140];
123 /* Data stored per client in the last_rcvd file. In le32 order. */
124 struct mds_client_data {
125 __u8 mcd_uuid[40]; /* client UUID */
126 __u64 mcd_last_transno; /* last completed transaction ID */
127 __u64 mcd_last_xid; /* xid for the last transaction */
128 __u32 mcd_last_result; /* result from last RPC */
129 __u32 mcd_last_data; /* per-op data (disposition for open &c.) */
130 __u8 mcd_padding[MDS_LR_CLIENT_SIZE - 64];
133 /* file data for open files on MDS */
134 struct mds_file_data {
135 struct portals_handle mfd_handle; /* must be first */
136 atomic_t mfd_refcount;
137 struct list_head mfd_list;
140 struct dentry *mfd_dentry;
143 /* group hash table */
144 struct mds_grp_hash_entry {
145 struct list_head ge_hash;
146 struct group_info *ge_group_info;
149 atomic_t ge_refcount;
150 wait_queue_head_t ge_waitq;
151 long ge_acquisition_time;
152 unsigned long ge_acquire_expire;
153 unsigned long ge_expire;
156 #define MDSGRP_HASH_SIZE (128)
157 #define MDSGRP_HASH_INDEX(id) ((id) & (MDSGRP_HASH_SIZE - 1))
158 #define MDSGRP_UPCALL_MAXPATH (1024)
160 struct mds_grp_hash {
161 struct list_head gh_table[MDSGRP_HASH_SIZE];
163 char gh_upcall[MDSGRP_UPCALL_MAXPATH];
165 int gh_acquire_expire;
166 unsigned int gh_allow_setgroups:1;
169 /* mds/mds_reint.c */
170 int mds_reint_rec(struct mds_update_record *r, int offset,
171 struct ptlrpc_request *req, struct lustre_handle *);
176 mds_id2dentry(struct obd_device *obd, struct lustre_id *id,
177 struct vfsmount **mnt);
180 mds_id2locked_dentry(struct obd_device *obd, struct lustre_id *id,
181 struct vfsmount **mnt, int lock_mode,
182 struct lustre_handle *lockh, int *pmode,
183 char *name, int namelen, __u64 lockpart);
184 int mds_update_server_data(struct obd_device *, int force_sync);
185 int mds_update_last_fid(struct obd_device *obd, int force_sync);
188 int mds_fs_setup(struct obd_device *obddev, struct vfsmount *mnt);
189 int mds_fs_cleanup(struct obd_device *obddev, int failover);
194 /* mdc/mdc_locks.c */
195 int it_disposition(struct lookup_intent *it, int flag);
196 void it_set_disposition(struct lookup_intent *it, int flag);
197 int it_open_error(int phase, struct lookup_intent *it);
198 int mdc_set_lock_data(struct obd_export *exp, __u64 *lockh, void *data);
199 int mdc_change_cbdata(struct obd_export *exp, struct lustre_id *id,
200 ldlm_iterator_t it, void *data);
201 int mdc_intent_lock(struct obd_export *exp, struct lustre_id *parent,
202 const char *name, int len, void *lmm, int lmmsize,
203 struct lustre_id *child, struct lookup_intent *, int,
204 struct ptlrpc_request **reqp,
205 ldlm_blocking_callback cb_blocking);
206 int mdc_enqueue(struct obd_export *exp,
208 struct lookup_intent *it,
210 struct mdc_op_data *data,
211 struct lustre_handle *lockh,
214 ldlm_completion_callback cb_completion,
215 ldlm_blocking_callback cb_blocking,
218 /* mdc/mdc_request.c */
219 int mdc_get_secdesc_size(void);
220 void mdc_pack_secdesc(struct ptlrpc_request *req, int size);
221 int mdc_req2lustre_md(struct obd_export *exp_mdc, struct ptlrpc_request *req,
222 unsigned int offset, struct obd_export *exp_osc,
223 struct lustre_md *md);
224 int mdc_getstatus(struct obd_export *exp, struct lustre_id *rootid);
225 int mdc_getattr(struct obd_export *exp, struct lustre_id *id,
226 unsigned long valid, unsigned int ea_size,
227 struct ptlrpc_request **request);
228 int mdc_getattr_name(struct obd_export *exp, struct lustre_id *id,
229 char *filename, int namelen, unsigned long valid,
230 unsigned int ea_size, struct ptlrpc_request **request);
231 int mdc_setattr(struct obd_export *exp, struct mdc_op_data *data,
232 struct iattr *iattr, void *ea, int ealen, void *ea2, int ea2len,
233 struct ptlrpc_request **request);
234 int mdc_open(struct obd_export *exp, obd_id ino, int type, int flags,
235 struct lov_mds_md *lmm, int lmm_size, struct lustre_handle *fh,
236 struct ptlrpc_request **);
238 struct obd_client_handle;
240 int mdc_set_open_replay_data(struct obd_export *exp,
241 struct obd_client_handle *och,
242 struct ptlrpc_request *open_req);
243 int mdc_clear_open_replay_data(struct obd_export *exp,
244 struct obd_client_handle *och);
245 int mdc_close(struct obd_export *, struct obdo *, struct obd_client_handle *,
246 struct ptlrpc_request **);
247 int mdc_readpage(struct obd_export *exp, struct lustre_id *id,
248 __u64, struct page *, struct ptlrpc_request **);
249 int mdc_create(struct obd_export *exp, struct mdc_op_data *op_data,
250 const void *data, int datalen, int mode, __u32 uid, __u32 gid,
251 __u64 rdev, struct ptlrpc_request **request);
252 int mdc_unlink(struct obd_export *exp, struct mdc_op_data *data,
253 struct ptlrpc_request **request);
254 int mdc_link(struct obd_export *exp, struct mdc_op_data *data,
255 struct ptlrpc_request **);
256 int mdc_rename(struct obd_export *exp, struct mdc_op_data *data,
257 const char *old, int oldlen, const char *new, int newlen,
258 struct ptlrpc_request **request);
259 int mdc_sync(struct obd_export *exp, struct lustre_id *id,
260 struct ptlrpc_request **);
261 int mdc_create_client(struct obd_uuid uuid, struct ptlrpc_client *cl);
263 /* store the generation of a newly-created inode in |req| for replay. */
264 int mdc_store_inode_generation(struct obd_export *exp,
265 struct ptlrpc_request *req,
266 int reqoff, int repoff);
268 int mdc_llog_process(struct obd_export *, char *, llog_cb_t,
271 int mdc_done_writing(struct obd_export *, struct obdo *);
273 /* ioctls for trying requests */
274 #define IOC_REQUEST_TYPE 'f'
275 #define IOC_REQUEST_MIN_NR 30
277 #define IOC_REQUEST_GETATTR _IOWR('f', 30, long)
278 #define IOC_REQUEST_READPAGE _IOWR('f', 31, long)
279 #define IOC_REQUEST_SETATTR _IOWR('f', 32, long)
280 #define IOC_REQUEST_CREATE _IOWR('f', 33, long)
281 #define IOC_REQUEST_OPEN _IOWR('f', 34, long)
282 #define IOC_REQUEST_CLOSE _IOWR('f', 35, long)
283 #define IOC_REQUEST_MAX_NR 35
285 #define MDS_CHECK_RESENT(req, reconstruct) \
287 if (lustre_msg_get_flags(req->rq_reqmsg) & MSG_RESENT) { \
288 struct mds_client_data *mcd = \
289 req->rq_export->exp_mds_data.med_mcd; \
290 if (mcd->mcd_last_xid == req->rq_xid) { \
294 DEBUG_REQ(D_HA, req, "no reply for RESENT req (have "LPD64")", \
295 mcd->mcd_last_xid); \