/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
* vim:expandtab:shiftwidth=8:tabstop=8:
*
- * lustre/mdt/mdt_internal.h
- * Lustre Metadata Target (mdt) request handler
+ * GPL HEADER START
*
- * Copyright (c) 2006 Cluster File Systems, Inc.
- * Author: Peter Braam <braam@clusterfs.com>
- * Author: Andreas Dilger <adilger@clusterfs.com>
- * Author: Phil Schwan <phil@clusterfs.com>
- * Author: Mike Shaver <shaver@clusterfs.com>
- * Author: Nikita Danilov <nikita@clusterfs.com>
- * Author: Huang Hua <huanghua@clusterfs.com>
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
- * This file is part of the Lustre file system, http://www.lustre.org
- * Lustre is a trademark of Cluster File Systems, Inc.
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 only,
+ * as published by the Free Software Foundation.
*
- * You may have signed or agreed to another license before downloading
- * this software. If so, you are bound by the terms and conditions
- * of that agreement, and the following does not apply to you. See the
- * LICENSE file included with this distribution for more information.
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details (a copy is included
+ * in the LICENSE file that accompanied this code).
*
- * If you did not agree to a different license, then this copy of Lustre
- * is open source software; you can redistribute it and/or modify it
- * under the terms of version 2 of the GNU General Public License as
- * published by the Free Software Foundation.
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
*
- * In either case, Lustre is distributed in the hope that it will be
- * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
- * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * license text for more details.
+ * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
+ * CA 95054 USA or visit www.sun.com if you need additional information or
+ * have any questions.
+ *
+ * GPL HEADER END
+ */
+/*
+ * Copyright 2008 Sun Microsystems, Inc. All rights reserved
+ * Use is subject to license terms.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * lustre/mdt/mdt_internal.h
+ *
+ * Lustre Metadata Target (mdt) request handler
+ *
+ * Author: Peter Braam <braam@clusterfs.com>
+ * Author: Andreas Dilger <adilger@clusterfs.com>
+ * Author: Phil Schwan <phil@clusterfs.com>
+ * Author: Mike Shaver <shaver@clusterfs.com>
+ * Author: Nikita Danilov <nikita@clusterfs.com>
+ * Author: Huang Hua <huanghua@clusterfs.com>
*/
#ifndef _MDT_INTERNAL_H
#include <lustre_disk.h>
#include <lustre_sec.h>
#include <lvfs.h>
+#include <lustre_idmap.h>
+#include <lustre_eacl.h>
-
-/* Data stored per client in the last_rcvd file. In le32 order. */
-struct mdt_client_data {
- __u8 mcd_uuid[40]; /* client UUID */
- __u64 mcd_last_transno; /* last completed transaction ID */
- __u64 mcd_last_xid; /* xid for the last transaction */
- __u32 mcd_last_result; /* result from last RPC */
- __u32 mcd_last_data; /* per-op data (disposition for open &c.) */
- /* for MDS_CLOSE and MDS_DONE_WRITTING requests */
- __u64 mcd_last_close_transno; /* last completed transaction ID */
- __u64 mcd_last_close_xid; /* xid for the last transaction */
- __u32 mcd_last_close_result; /* result from last RPC */
- __u8 mcd_padding[LR_CLIENT_SIZE - 84];
-};
-
-static inline __u64 mcd_last_transno(struct mdt_client_data *mcd)
+static inline __u64 lcd_last_transno(struct lsd_client_data *lcd)
{
- return max(mcd->mcd_last_transno, mcd->mcd_last_close_transno);
+ return max(lcd->lcd_last_transno, lcd->lcd_last_close_transno);
}
-static inline __u64 mcd_last_xid(struct mdt_client_data *mcd)
+static inline __u64 lcd_last_xid(struct lsd_client_data *lcd)
{
- return max(mcd->mcd_last_xid, mcd->mcd_last_close_xid);
+ return max(lcd->lcd_last_xid, lcd->lcd_last_close_xid);
}
/* check if request's xid is equal to last one or not*/
static inline int req_xid_is_last(struct ptlrpc_request *req)
{
- struct mdt_client_data *mcd = req->rq_export->exp_mdt_data.med_mcd;
- return (req->rq_xid == mcd->mcd_last_xid ||
- req->rq_xid == mcd->mcd_last_close_xid);
+ struct lsd_client_data *lcd = req->rq_export->exp_mdt_data.med_lcd;
+ return (req->rq_xid == lcd->lcd_last_xid ||
+ req->rq_xid == lcd->lcd_last_close_xid);
}
-/* copied from lr_server_data.
- * mds data stored at the head of last_rcvd file. In le32 order. */
-struct mdt_server_data {
- __u8 msd_uuid[40]; /* server UUID */
- __u64 msd_last_transno; /* last completed transaction ID */
- __u64 msd_mount_count; /* incarnation number */
- __u32 msd_feature_compat; /* compatible feature flags */
- __u32 msd_feature_rocompat;/* read-only compatible feature flags */
- __u32 msd_feature_incompat;/* incompatible feature flags */
- __u32 msd_server_size; /* size of server data area */
- __u32 msd_client_start; /* start of per-client data area */
- __u16 msd_client_size; /* size of per-client data area */
- //__u16 msd_subdir_count; /* number of subdirectories for objects */
- //__u64 msd_catalog_oid; /* recovery catalog object id */
- //__u32 msd_catalog_ogen; /* recovery catalog inode generation */
- //__u8 msd_peeruuid[40]; /* UUID of MDS associated with this OST */
- //__u32 msd_ost_index; /* index number of OST in LOV */
- //__u32 msd_mdt_index; /* index number of MDT in LMV */
- __u8 msd_padding[LR_SERVER_SIZE - 78];
-};
-
struct mdt_object;
/* file data for open files on MDS */
struct mdt_file_data {
mo_mds_capa :1,
mo_oss_capa :1;
} mdt_opts;
-
+ /* mdt state flags */
+ __u32 mdt_fl_cfglog:1,
+ mdt_fl_synced:1;
/* lock to pretect epoch and write count */
spinlock_t mdt_ioepoch_lock;
__u64 mdt_ioepoch;
__u64 mdt_mount_count;
/* last_rcvd data */
- struct mdt_server_data mdt_msd;
+ struct lr_server_data mdt_lsd;
spinlock_t mdt_client_bitmap_lock;
unsigned long mdt_client_bitmap[(LR_MAX_CLIENTS >> 3) / sizeof(long)];
struct upcall_cache *mdt_identity_cache;
- struct upcall_cache *mdt_rmtacl_cache;
- /* root squash */
- struct rootsquash_info *mdt_rootsquash_info;
+ /* sptlrpc rules */
+ rwlock_t mdt_sptlrpc_lock;
+ struct sptlrpc_rule_set mdt_sptlrpc_rset;
/* capability keys */
unsigned long mdt_capa_timeout;
struct dt_object *mdt_ck_obj;
unsigned long mdt_ck_timeout;
unsigned long mdt_ck_expiry;
- struct timer_list mdt_ck_timer;
+ cfs_timer_t mdt_ck_timer;
struct ptlrpc_thread mdt_ck_thread;
struct lustre_capa_key mdt_capa_keys[2];
unsigned int mdt_capa_conf:1;
struct lprocfs_stats *mdt_stats;
};
-/*XXX copied from mds_internal.h */
-#define MDT_SERVICE_WATCHDOG_TIMEOUT (obd_timeout * 1000)
+#define MDT_SERVICE_WATCHDOG_FACTOR (2000)
#define MDT_ROCOMPAT_SUPP (OBD_ROCOMPAT_LOVOBJID)
#define MDT_INCOMPAT_SUPP (OBD_INCOMPAT_MDT | OBD_INCOMPAT_COMMON_LR)
struct mdt_thread_info {
/*
* XXX: Part One:
- * The following members will be filled expilictly
+ * The following members will be filled explicitly
* with specific data in mdt_thread_info_init().
*/
-
- /*
- * for req-layout interface. This field should be first to be compatible
- * with "struct com_thread_info" in seq and fld.
+ /* TODO: move this into mdt_session_key(with LCT_SESSION), because
+ * request handling may migrate from one server thread to another.
*/
- struct req_capsule mti_pill;
+ struct req_capsule *mti_pill;
+
/* although we have export in req, there are cases when it is not
* available, e.g. closing files upon export destroy */
struct obd_export *mti_exp;
/*
- * number of buffers in reply message.
- */
- int mti_rep_buf_nr;
- /*
- * sizes of reply buffers.
- */
- int mti_rep_buf_size[REQ_MAX_FIELD_NR];
- /*
* A couple of lock handles.
*/
struct mdt_lock_handle mti_lh[MDT_LH_NR];
/*
* XXX: Part Three:
- * The following members will be filled expilictly
+ * The following members will be filled explictly
* with zero in mdt_reint_unpack(), because they are only used
* by reint requests (including mdt_reint_open()).
*/
__u64 mti_replayepoch;
/* server and client data buffers */
- struct mdt_server_data mti_msd;
- struct mdt_client_data mti_mcd;
+ struct lr_server_data mti_lsd;
+ struct lsd_client_data mti_lcd;
loff_t mti_off;
struct txn_param mti_txn_param;
struct lu_buf mti_buf;
extern struct lu_context_key mdt_txn_key;
static inline void mdt_trans_add_cb(const struct thandle *th,
- mdt_cb_t cb_func, void *cb_data)
+ mdt_cb_t cb_func, void *cb_data)
{
struct mdt_txn_info *txi;
-
+
txi = lu_context_key_get(&th->th_ctx, &mdt_txn_key);
LASSERT(txi->txi_cb_count < ARRAY_SIZE(txi->txi_cb));
txi->txi_cb_count++;
}
-static inline struct md_device_operations *mdt_child_ops(struct mdt_device * m)
+static inline const struct md_device_operations *
+mdt_child_ops(struct mdt_device * m)
{
LASSERT(m->mdt_child);
return m->mdt_child->md_ops;
static inline struct md_object *mdt_object_child(struct mdt_object *o)
{
+ LASSERT(o);
return lu2md(lu_object_next(&o->mot_obj.mo_lu));
}
static inline struct ptlrpc_request *mdt_info_req(struct mdt_thread_info *info)
{
- return info->mti_pill.rc_req;
+ return info->mti_pill ? info->mti_pill->rc_req : NULL;
}
static inline __u64 mdt_conn_flags(struct mdt_thread_info *info)
return lu_object_fid(&o->mot_obj.mo_lu);
}
+static inline struct lu_site *mdt_lu_site(const struct mdt_device *mdt)
+{
+ return mdt->mdt_md_dev.md_lu_dev.ld_site;
+}
+
+static inline struct md_site *mdt_md_site(const struct mdt_device *mdt)
+{
+ return lu_site2md(mdt_lu_site(mdt));
+}
+
+static inline void mdt_export_evict(struct obd_export *exp)
+{
+ class_fail_export(exp);
+ class_export_put(exp);
+}
+
int mdt_get_disposition(struct ldlm_reply *rep, int flag);
void mdt_set_disposition(struct mdt_thread_info *info,
struct ldlm_reply *rep, int flag);
const struct lu_attr *attr, const struct lu_fid *fid);
int mdt_getxattr(struct mdt_thread_info *info);
-int mdt_setxattr(struct mdt_thread_info *info);
+int mdt_reint_setxattr(struct mdt_thread_info *info,
+ struct mdt_lock_handle *lh);
void mdt_lock_handle_init(struct mdt_lock_handle *lh);
void mdt_lock_handle_fini(struct mdt_lock_handle *lh);
int mdt_client_new(const struct lu_env *env,
struct mdt_device *mdt);
+int mdt_export_stats_init(struct obd_device *obd,
+ struct obd_export *exp,
+ void *client_nid);
+
int mdt_pin(struct mdt_thread_info* info);
int mdt_lock_new_child(struct mdt_thread_info *info,
void mdt_dump_lmm(int level, const struct lov_mds_md *lmm);
int mdt_check_ucred(struct mdt_thread_info *);
-
int mdt_init_ucred(struct mdt_thread_info *, struct mdt_body *);
-
int mdt_init_ucred_reint(struct mdt_thread_info *);
-
void mdt_exit_ucred(struct mdt_thread_info *);
-int groups_from_list(struct group_info *, gid_t *);
-
-void groups_sort(struct group_info *);
-
/* mdt_idmap.c */
int mdt_init_idmap(struct mdt_thread_info *);
extern struct upcall_cache_ops mdt_identity_upcall_cache_ops;
-struct mdt_identity *mdt_identity_get(struct upcall_cache *, __u32);
+struct md_identity *mdt_identity_get(struct upcall_cache *, __u32);
-void mdt_identity_put(struct upcall_cache *, struct mdt_identity *);
+void mdt_identity_put(struct upcall_cache *, struct md_identity *);
void mdt_flush_identity(struct upcall_cache *, int);
-__u32 mdt_identity_get_setxid_perm(struct mdt_identity *, __u32, lnet_nid_t);
+__u32 mdt_identity_get_perm(struct md_identity *, __u32, lnet_nid_t);
int mdt_pack_remote_perm(struct mdt_thread_info *, struct mdt_object *, void *);
-/* mdt/mdt_rmtacl.c */
-#define MDT_RMTACL_UPCALL_PATH "/usr/sbin/l_facl"
-
-extern struct upcall_cache_ops mdt_rmtacl_upcall_cache_ops;
-
-int mdt_rmtacl_upcall(struct mdt_thread_info *, char *, struct lu_buf *);
-
extern struct lu_context_key mdt_thread_key;
/* debug issues helper starts here*/
static inline void mdt_fail_write(const struct lu_env *env,
struct dt_device *dd, int id)
{
- if (OBD_FAIL_CHECK(id)) {
+ if (OBD_FAIL_CHECK_ORSET(id, OBD_FAIL_ONCE)) {
CERROR(LUSTRE_MDT_NAME": obd_fail_loc=%x, fail write ops\n",
id);
dd->dd_ops->dt_ro(env, dd);
/* We set FAIL_ONCE because we never "un-fail" a device */
- obd_fail_loc |= OBD_FAILED | OBD_FAIL_ONCE;
}
}
RETURN(1);
}
DEBUG_REQ(D_HA, req, "no reply for RESENT req (have "LPD64")",
- req->rq_export->exp_mdt_data.med_mcd->mcd_last_xid);
+ req->rq_export->exp_mdt_data.med_lcd->lcd_last_xid);
}
RETURN(0);
}
-#define MDT_FAIL_CHECK(id) \
-({ \
- if (unlikely(OBD_FAIL_CHECK(id))) \
- CERROR(LUSTRE_MDT_NAME": " #id " test failed\n"); \
- OBD_FAIL_CHECK(id); \
-})
-
-#define MDT_FAIL_CHECK_ONCE(id) \
-({ int _ret_ = 0; \
- if (unlikely(OBD_FAIL_CHECK(id))) { \
- CERROR(LUSTRE_MDT_NAME": *** obd_fail_loc=%x ***\n", id); \
- obd_fail_loc |= OBD_FAILED; \
- if ((id) & OBD_FAIL_ONCE) \
- obd_fail_loc |= OBD_FAIL_ONCE; \
- _ret_ = 1; \
- } \
- _ret_; \
-})
-
-#define MDT_FAIL_RETURN(id, ret) \
-do { \
- if (unlikely(MDT_FAIL_CHECK_ONCE(id))) { \
- RETURN(ret); \
- } \
-} while(0)
-
struct md_ucred *mdt_ucred(const struct mdt_thread_info *info);
static inline int is_identity_get_disabled(struct upcall_cache *cache)
}
/* lprocfs stuff */
+void lprocfs_mdt_init_vars(struct lprocfs_static_vars *lvars);
int mdt_procfs_init(struct mdt_device *mdt, const char *name);
int mdt_procfs_fini(struct mdt_device *mdt);