/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
* vim:expandtab:shiftwidth=8:tabstop=8:
*
- * Extention of lu_object.h for metadata objects
+ * GPL HEADER START
*
- * Copyright (C) 2006 Cluster File Systems, Inc.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
- * This file is part of Lustre, http://www.lustre.org.
+ * 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.
*
- * Lustre is free 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.
+ * 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).
*
- * 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
- * GNU General Public License for more details.
+ * 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
*
- * You should have received a copy of the GNU General Public License
- * along with Lustre; if not, write to the Free Software
- * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ * 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 (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
+ */
+/*
+ * Copyright (c) 2011 Whamcloud, Inc.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * lustre/include/md_object.h
+ *
+ * Extention of lu_object.h for metadata objects
*/
#ifndef _LUSTRE_MD_OBJECT_H
#define _LUSTRE_MD_OBJECT_H
-/*
+/** \defgroup md md
* Sub-class of lu_object with methods common for "meta-data" objects in MDT
* stack.
*
* them, and treat them as directories.
*
* Examples: mdt, cmm, and mdt are implementations of md interface.
+ * @{
*/
/*
* super-class definitions.
*/
-#include <lu_object.h>
+#include <dt_object.h>
#include <lvfs.h>
struct md_device;
struct md_device_operations;
struct md_object;
+struct obd_export;
-
-typedef enum {
+enum {
UCRED_INVALID = -1,
UCRED_INIT = 0,
UCRED_OLD = 1,
- UCRED_NEW = 2,
-} ucred_t;
-
-#define SQUASH_NONE 0x00
-#define SQUASH_UID 0x01
-#define SQUASH_GID 0x02
+ UCRED_NEW = 2
+};
struct md_ucred {
- ucred_t mu_valid;
- __u32 mu_squash;
- __u32 mu_o_uid;
- __u32 mu_o_gid;
- __u32 mu_o_fsuid;
- __u32 mu_o_fsgid;
- __u32 mu_uid;
- __u32 mu_gid;
- __u32 mu_fsuid;
- __u32 mu_fsgid;
- __u32 mu_suppgids[2];
- __u32 mu_cap;
- __u32 mu_umask;
- struct group_info *mu_ginfo;
- struct mdt_identity *mu_identity;
+ __u32 mu_valid;
+ __u32 mu_o_uid;
+ __u32 mu_o_gid;
+ __u32 mu_o_fsuid;
+ __u32 mu_o_fsgid;
+ __u32 mu_uid;
+ __u32 mu_gid;
+ __u32 mu_fsuid;
+ __u32 mu_fsgid;
+ __u32 mu_suppgids[2];
+ cfs_cap_t mu_cap;
+ __u32 mu_umask;
+ cfs_group_info_t *mu_ginfo;
+ struct md_identity *mu_identity;
};
-#define MD_CAPAINFO_MAX 5
+enum {
+ MD_CAPAINFO_MAX = 5
+};
-/* there are at most 5 fids in one operation, see rename, NOTE the last one
+/** there are at most 5 fids in one operation, see rename, NOTE the last one
* is a temporary one used for is_subdir() */
struct md_capainfo {
+ __u32 mc_auth;
+ __u32 mc_padding;
const struct lu_fid *mc_fid[MD_CAPAINFO_MAX];
struct lustre_capa *mc_capa[MD_CAPAINFO_MAX];
};
-/*
+struct md_quota {
+ struct obd_export *mq_exp;
+};
+
+/**
* Implemented in mdd/mdd_handler.c.
*
* XXX should be moved into separate .h/.c together with all md security
*/
struct md_ucred *md_ucred(const struct lu_env *env);
struct md_capainfo *md_capainfo(const struct lu_env *env);
+struct md_quota *md_quota(const struct lu_env *env);
-/* metadata attributes */
+/** metadata attributes */
enum ma_valid {
MA_INODE = (1 << 0),
MA_LOV = (1 << 1),
MA_FLAGS = (1 << 3),
MA_LMV = (1 << 4),
MA_ACL_DEF = (1 << 5),
- MA_LOV_DEF = (1 << 6)
+ MA_LOV_DEF = (1 << 6),
+/* (Layout lock will used #7 here) */
+ MA_HSM = (1 << 8),
+ MA_SOM = (1 << 9),
+ MA_PFID = (1 << 10)
};
typedef enum {
MDT_PDO_LOCK = (1 << 1)
} mdl_type_t;
+struct md_hsm {
+ __u32 mh_flags;
+ __u32 mh_archive_number;
+};
+
+#define IOEPOCH_INVAL 0
+
+struct md_som_data {
+ __u64 msd_ioepoch;
+ __u64 msd_size;
+ __u64 msd_blocks;
+ __u64 msd_mountid;
+};
+
struct md_attr {
__u64 ma_valid;
__u64 ma_need;
struct llog_cookie *ma_cookie;
int ma_cookie_size;
struct lustre_capa *ma_capa;
+ struct md_hsm ma_hsm;
+ struct md_som_data *ma_som;
+ struct lu_fid ma_pfid;
};
-/* Additional parameters for create */
+/** Additional parameters for create */
struct md_op_spec {
union {
- /* symlink target */
+ /** symlink target */
const char *sp_symname;
- /* parent FID for cross-ref mkdir */
+ /** parent FID for cross-ref mkdir */
const struct lu_fid *sp_pfid;
- /* eadata for regular files */
+ /** eadata for regular files */
struct md_spec_reg {
- /* lov objs exist already */
+ /** lov objs exist already */
const struct lu_fid *fid;
- int no_lov_create;
const void *eadata;
int eadatalen;
} sp_ea;
} u;
-
- /* Create flag from client: such as MDS_OPEN_CREAT, and others. */
- __u32 sp_cr_flags;
+ /** don't create lov objects or llog cookie - this replay */
+ int no_create;
- /* Should mdd do lookup sanity check or not. */
+ /** Create flag from client: such as MDS_OPEN_CREAT, and others. */
+ __u64 sp_cr_flags;
+
+ /** Should mdd do lookup sanity check or not. */
int sp_cr_lookup;
- /* Current lock mode for parent dir where create is performing. */
+ /** Current lock mode for parent dir where create is performing. */
mdl_mode_t sp_cr_mode;
- /* Check for split */
+ /** Check for split */
int sp_ck_split;
+
+ /** to create directory */
+ const struct dt_index_features *sp_feat;
};
-/*
+/**
* Operations implemented for each md object (both directory and leaf).
*/
struct md_object_operations {
int (*moo_xattr_del)(const struct lu_env *env, struct md_object *obj,
const char *name);
+ /** \retval number of bytes actually read upon success */
int (*moo_readpage)(const struct lu_env *env, struct md_object *obj,
const struct lu_rdpg *rdpg);
int (*moo_readlink)(const struct lu_env *env, struct md_object *obj,
struct lu_buf *buf);
-
- /* part of cross-ref operation */
+ int (*moo_changelog)(const struct lu_env *env,
+ enum changelog_rec_type type, int flags,
+ struct md_object *obj);
+ /** part of cross-ref operation */
int (*moo_object_create)(const struct lu_env *env,
struct md_object *obj,
const struct md_op_spec *spec,
struct md_object *obj, int flag);
int (*moo_close)(const struct lu_env *env, struct md_object *obj,
- struct md_attr *ma);
-
+ struct md_attr *ma, int mode);
+
int (*moo_capa_get)(const struct lu_env *, struct md_object *,
struct lustre_capa *, int renewal);
+
+ int (*moo_object_sync)(const struct lu_env *, struct md_object *);
+ dt_obj_version_t (*moo_version_get)(const struct lu_env *,
+ struct md_object *);
+ void (*moo_version_set)(const struct lu_env *, struct md_object *,
+ dt_obj_version_t);
+ int (*moo_path)(const struct lu_env *env, struct md_object *obj,
+ char *path, int pathlen, __u64 *recno, int *linkno);
+ int (*moo_file_lock)(const struct lu_env *env, struct md_object *obj,
+ struct lov_mds_md *lmm, struct ldlm_extent *extent,
+ struct lustre_handle *lockh);
+ int (*moo_file_unlock)(const struct lu_env *env, struct md_object *obj,
+ struct lov_mds_md *lmm,
+ struct lustre_handle *lockh);
};
-/*
+/**
* Operations implemented for each directory object.
*/
struct md_dir_operations {
struct md_op_spec *spec,
struct md_attr *ma);
- /* This method is used for creating data object for this meta object*/
+ /** This method is used for creating data object for this meta object*/
int (*mdo_create_data)(const struct lu_env *env, struct md_object *p,
struct md_object *o,
const struct md_op_spec *spec,
struct md_object *cobj, const struct lu_name *lname,
struct md_attr *ma);
- /* partial ops for cross-ref case */
+ /** partial ops for cross-ref case */
int (*mdo_name_insert)(const struct lu_env *env,
struct md_object *obj,
const struct lu_name *lname,
};
struct md_device_operations {
- /* meta-data device related handlers. */
+ /** meta-data device related handlers. */
int (*mdo_root_get)(const struct lu_env *env, struct md_device *m,
struct lu_fid *f);
int *md_size, int *cookie_size);
int (*mdo_statfs)(const struct lu_env *env, struct md_device *m,
- struct kstatfs *sfs);
+ cfs_kstatfs_t *sfs);
int (*mdo_init_capa_ctxt)(const struct lu_env *env, struct md_device *m,
int mode, unsigned long timeout, __u32 alg,
int (*mdo_update_capa_key)(const struct lu_env *env,
struct md_device *m,
struct lustre_capa_key *key);
+
+ int (*mdo_llog_ctxt_get)(const struct lu_env *env,
+ struct md_device *m, int idx, void **h);
+
+ int (*mdo_iocontrol)(const struct lu_env *env, struct md_device *m,
+ unsigned int cmd, int len, void *data);
+
+#ifdef HAVE_QUOTA_SUPPORT
+ struct md_quota_operations {
+ int (*mqo_notify)(const struct lu_env *env,
+ struct md_device *m);
+
+ int (*mqo_setup)(const struct lu_env *env,
+ struct md_device *m,
+ void *data);
+
+ int (*mqo_cleanup)(const struct lu_env *env,
+ struct md_device *m);
+
+ int (*mqo_recovery)(const struct lu_env *env,
+ struct md_device *m);
+
+ int (*mqo_check)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type);
+
+ int (*mqo_on)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type);
+
+ int (*mqo_off)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type);
+
+ int (*mqo_setinfo)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type,
+ __u32 id,
+ struct obd_dqinfo *dqinfo);
+
+ int (*mqo_getinfo)(const struct lu_env *env,
+ const struct md_device *m,
+ __u32 type,
+ __u32 id,
+ struct obd_dqinfo *dqinfo);
+
+ int (*mqo_setquota)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type,
+ __u32 id,
+ struct obd_dqblk *dqblk);
+
+ int (*mqo_getquota)(const struct lu_env *env,
+ const struct md_device *m,
+ __u32 type,
+ __u32 id,
+ struct obd_dqblk *dqblk);
+
+ int (*mqo_getoinfo)(const struct lu_env *env,
+ const struct md_device *m,
+ __u32 type,
+ __u32 id,
+ struct obd_dqinfo *dqinfo);
+
+ int (*mqo_getoquota)(const struct lu_env *env,
+ const struct md_device *m,
+ __u32 type,
+ __u32 id,
+ struct obd_dqblk *dqblk);
+
+ int (*mqo_invalidate)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type);
+
+ int (*mqo_finvalidate)(const struct lu_env *env,
+ struct md_device *m,
+ __u32 type);
+ } mdo_quota;
+#endif
};
enum md_upcall_event {
- /*sync the md layer*/
+ /** Sync the md layer*/
MD_LOV_SYNC = (1 << 0),
- MD_NO_TRANS = (1 << 1), /* Just for split, no need trans, for replay */
- MD_LOV_CONFIG = (1 << 2)
+ /** Just for split, no need trans, for replay */
+ MD_NO_TRANS = (1 << 1),
+ MD_LOV_CONFIG = (1 << 2),
+ /** Trigger quota recovery */
+ MD_LOV_QUOTA = (1 << 3)
};
struct md_upcall {
- /* this lock protects upcall using against its removal
+ /** this lock protects upcall using against its removal
* read lock is for usage the upcall, write - for init/fini */
- struct rw_semaphore mu_upcall_sem;
- /* device to call, upper layer normally */
+ cfs_rw_semaphore_t mu_upcall_sem;
+ /** device to call, upper layer normally */
struct md_device *mu_upcall_dev;
- /* upcall function */
+ /** upcall function */
int (*mu_upcall)(const struct lu_env *env, struct md_device *md,
- enum md_upcall_event ev);
+ enum md_upcall_event ev, void *data);
};
struct md_device {
- struct lu_device md_lu_dev;
- struct md_device_operations *md_ops;
- struct md_upcall md_upcall;
+ struct lu_device md_lu_dev;
+ const struct md_device_operations *md_ops;
+ struct md_upcall md_upcall;
};
static inline void md_upcall_init(struct md_device *m, void *upcl)
{
- init_rwsem(&m->md_upcall.mu_upcall_sem);
+ cfs_init_rwsem(&m->md_upcall.mu_upcall_sem);
m->md_upcall.mu_upcall_dev = NULL;
m->md_upcall.mu_upcall = upcl;
}
static inline void md_upcall_dev_set(struct md_device *m, struct md_device *up)
{
- down_write(&m->md_upcall.mu_upcall_sem);
+ cfs_down_write(&m->md_upcall.mu_upcall_sem);
m->md_upcall.mu_upcall_dev = up;
- up_write(&m->md_upcall.mu_upcall_sem);
+ cfs_up_write(&m->md_upcall.mu_upcall_sem);
}
static inline void md_upcall_fini(struct md_device *m)
{
- down_write(&m->md_upcall.mu_upcall_sem);
+ cfs_down_write(&m->md_upcall.mu_upcall_sem);
m->md_upcall.mu_upcall_dev = NULL;
m->md_upcall.mu_upcall = NULL;
- up_write(&m->md_upcall.mu_upcall_sem);
+ cfs_up_write(&m->md_upcall.mu_upcall_sem);
}
static inline int md_do_upcall(const struct lu_env *env, struct md_device *m,
- enum md_upcall_event ev)
+ enum md_upcall_event ev, void *data)
{
int rc = 0;
- down_read(&m->md_upcall.mu_upcall_sem);
+ cfs_down_read(&m->md_upcall.mu_upcall_sem);
if (m->md_upcall.mu_upcall_dev != NULL &&
m->md_upcall.mu_upcall_dev->md_upcall.mu_upcall != NULL) {
rc = m->md_upcall.mu_upcall_dev->md_upcall.mu_upcall(env,
- m->md_upcall.mu_upcall_dev, ev);
+ m->md_upcall.mu_upcall_dev,
+ ev, data);
}
- up_read(&m->md_upcall.mu_upcall_sem);
+ cfs_up_read(&m->md_upcall.mu_upcall_sem);
return rc;
}
struct md_object {
- struct lu_object mo_lu;
- struct md_object_operations *mo_ops;
- struct md_dir_operations *mo_dir_ops;
+ struct lu_object mo_lu;
+ const struct md_object_operations *mo_ops;
+ const struct md_dir_operations *mo_dir_ops;
+};
+
+/**
+ * md-server site.
+ */
+struct md_site {
+ struct lu_site ms_lu;
+ /**
+ * mds number of this site.
+ */
+ mdsno_t ms_node_id;
+ /**
+ * Fid location database
+ */
+ struct lu_server_fld *ms_server_fld;
+ struct lu_client_fld *ms_client_fld;
+
+ /**
+ * Server Seq Manager
+ */
+ struct lu_server_seq *ms_server_seq;
+
+ /**
+ * Controller Seq Manager
+ */
+ struct lu_server_seq *ms_control_seq;
+ struct obd_export *ms_control_exp;
+
+ /**
+ * Client Seq Manager
+ */
+ struct lu_client_seq *ms_client_seq;
};
static inline int lu_device_is_md(const struct lu_device *d)
static inline struct md_device *lu2md_dev(const struct lu_device *d)
{
- LASSERT(lu_device_is_md(d));
+ LASSERT(IS_ERR(d) || lu_device_is_md(d));
return container_of0(d, struct md_device, md_lu_dev);
}
static inline struct md_object *lu2md(const struct lu_object *o)
{
- LASSERT(lu_device_is_md(o->lo_dev));
+ LASSERT(o == NULL || IS_ERR(o) || lu_device_is_md(o->lo_dev));
return container_of0(o, struct md_object, mo_lu);
}
static inline struct md_device *md_obj2dev(const struct md_object *o)
{
- LASSERT(lu_device_is_md(o->mo_lu.lo_dev));
+ LASSERT(o == NULL || IS_ERR(o) || lu_device_is_md(o->mo_lu.lo_dev));
return container_of0(o->mo_lu.lo_dev, struct md_device, md_lu_dev);
}
+static inline struct md_site *lu_site2md(const struct lu_site *s)
+{
+ return container_of0(s, struct md_site, ms_lu);
+}
+
static inline int md_device_init(struct md_device *md, struct lu_device_type *t)
{
- return lu_device_init(&md->md_lu_dev, t);
+ return lu_device_init(&md->md_lu_dev, t);
}
static inline void md_device_fini(struct md_device *md)
{
- lu_device_fini(&md->md_lu_dev);
+ lu_device_fini(&md->md_lu_dev);
}
-/* md operations */
+static inline struct md_object *md_object_find_slice(const struct lu_env *env,
+ struct md_device *md,
+ const struct lu_fid *f)
+{
+ return lu2md(lu_object_find_slice(env, md2lu_dev(md), f, NULL));
+}
+
+
+/** md operations */
static inline int mo_permission(const struct lu_env *env,
struct md_object *p,
struct md_object *c,
return m->mo_ops->moo_readlink(env, m, buf);
}
+static inline int mo_changelog(const struct lu_env *env,
+ enum changelog_rec_type type,
+ int flags, struct md_object *m)
+{
+ LASSERT(m->mo_ops->moo_changelog);
+ return m->mo_ops->moo_changelog(env, type, flags, m);
+}
+
static inline int mo_attr_set(const struct lu_env *env,
struct md_object *m,
const struct md_attr *at)
static inline int mo_close(const struct lu_env *env,
struct md_object *m,
- struct md_attr *ma)
+ struct md_attr *ma,
+ int mode)
{
LASSERT(m->mo_ops->moo_close);
- return m->mo_ops->moo_close(env, m, ma);
+ return m->mo_ops->moo_close(env, m, ma, mode);
}
static inline int mo_readpage(const struct lu_env *env,
return m->mo_ops->moo_capa_get(env, m, c, renewal);
}
+static inline int mo_path(const struct lu_env *env, struct md_object *m,
+ char *path, int pathlen, __u64 *recno, int *linkno)
+{
+ if (m->mo_ops->moo_path == NULL)
+ return -ENOSYS;
+ return m->mo_ops->moo_path(env, m, path, pathlen, recno, linkno);
+}
+
+static inline int mo_object_sync(const struct lu_env *env, struct md_object *m)
+{
+ LASSERT(m->mo_ops->moo_object_sync);
+ return m->mo_ops->moo_object_sync(env, m);
+}
+
+static inline dt_obj_version_t mo_version_get(const struct lu_env *env,
+ struct md_object *m)
+{
+ LASSERT(m->mo_ops->moo_version_get);
+ return m->mo_ops->moo_version_get(env, m);
+}
+
+static inline void mo_version_set(const struct lu_env *env,
+ struct md_object *m, dt_obj_version_t ver)
+{
+ LASSERT(m->mo_ops->moo_version_set);
+ return m->mo_ops->moo_version_set(env, m, ver);
+}
+
+static inline int mo_file_lock(const struct lu_env *env, struct md_object *m,
+ struct lov_mds_md *lmm,
+ struct ldlm_extent *extent,
+ struct lustre_handle *lockh)
+{
+ LASSERT(m->mo_ops->moo_file_lock);
+ return m->mo_ops->moo_file_lock(env, m, lmm, extent, lockh);
+}
+
+static inline int mo_file_unlock(const struct lu_env *env, struct md_object *m,
+ struct lov_mds_md *lmm,
+ struct lustre_handle *lockh)
+{
+ LASSERT(m->mo_ops->moo_file_unlock);
+ return m->mo_ops->moo_file_unlock(env, m, lmm, lockh);
+}
+
static inline int mdo_lookup(const struct lu_env *env,
struct md_object *p,
const struct lu_name *lname,
}
}
+struct dt_device;
+/**
+ * Structure to hold object information. This is used to create object
+ * \pre llod_dir exist
+ */
+struct lu_local_obj_desc {
+ const char *llod_dir;
+ const char *llod_name;
+ __u32 llod_oid;
+ int llod_is_index;
+ const struct dt_index_features *llod_feat;
+ cfs_list_t llod_linkage;
+};
+
+struct md_object *llo_store_resolve(const struct lu_env *env,
+ struct md_device *md,
+ struct dt_device *dt,
+ const char *path,
+ struct lu_fid *fid);
+
+struct md_object *llo_store_open(const struct lu_env *env,
+ struct md_device *md,
+ struct dt_device *dt,
+ const char *dirname,
+ const char *objname,
+ struct lu_fid *fid);
+
+struct md_object *llo_store_create_index(const struct lu_env *env,
+ struct md_device *md,
+ struct dt_device *dt,
+ const char *dirname,
+ const char *objname,
+ const struct lu_fid *fid,
+ const struct dt_index_features *feat);
+
+struct md_object *llo_store_create(const struct lu_env *env,
+ struct md_device *md,
+ struct dt_device *dt,
+ const char *dirname,
+ const char *objname,
+ const struct lu_fid *fid);
+
+void llo_local_obj_register(struct lu_local_obj_desc *);
+void llo_local_obj_unregister(struct lu_local_obj_desc *);
+
+int llo_local_objects_setup(const struct lu_env *env,
+ struct md_device * md,
+ struct dt_device * dt);
+
+/** @} md */
#endif /* _LINUX_MD_OBJECT_H */