X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fmdt%2Fmdt_internal.h;h=aa6c6cc547e33a0c3ac9c38e15f6d26101f596b7;hb=57e224145f7c390e33c295d95c735934d0c54883;hp=aabf8d7f5d33dda3fc96dbde626d92088df16b57;hpb=a1efc41f79a5f1d04e7427d76265e0e95b97e58c;p=fs%2Flustre-release.git diff --git a/lustre/mdt/mdt_internal.h b/lustre/mdt/mdt_internal.h index aabf8d7..aa6c6cc 100644 --- a/lustre/mdt/mdt_internal.h +++ b/lustre/mdt/mdt_internal.h @@ -1,34 +1,48 @@ /* -*- 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 - * Author: Andreas Dilger - * Author: Phil Schwan - * Author: Mike Shaver - * Author: Nikita Danilov - * Author: Huang Hua + * 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 + * Author: Andreas Dilger + * Author: Phil Schwan + * Author: Mike Shaver + * Author: Nikita Danilov + * Author: Huang Hua */ #ifndef _MDT_INTERNAL_H @@ -46,72 +60,26 @@ * struct lustre_handle */ #include +#include +#include #include -#include #include #include #include -/* LR_CLIENT_SIZE, etc. */ -#include #include #include #include #include - - -/* 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) -{ - return max(mcd->mcd_last_transno, mcd->mcd_last_close_transno); -} - -static inline __u64 mcd_last_xid(struct mdt_client_data *mcd) -{ - return max(mcd->mcd_last_xid, mcd->mcd_last_close_xid); -} +#include /* 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 { @@ -141,6 +109,8 @@ struct mdt_device { /* underlying device */ struct md_device *mdt_child; struct dt_device *mdt_bottom; + /** target device */ + struct lu_target mdt_lut; /* * Options bit-fields. */ @@ -149,7 +119,8 @@ struct mdt_device { mo_acl :1, mo_compat_resname:1, mo_mds_capa :1, - mo_oss_capa :1; + mo_oss_capa :1, + mo_cos :1; } mdt_opts; /* mdt state flags */ __u32 mdt_fl_cfglog:1, @@ -158,25 +129,13 @@ struct mdt_device { spinlock_t mdt_ioepoch_lock; __u64 mdt_ioepoch; - /* Transaction related stuff here */ - spinlock_t mdt_transno_lock; - __u64 mdt_last_transno; - /* transaction callbacks */ struct dt_txn_callback mdt_txn_cb; - /* last_rcvd file */ - struct dt_object *mdt_last_rcvd; /* these values should be updated from lov if necessary. * or should be placed somewhere else. */ int mdt_max_mdsize; int mdt_max_cookiesize; - __u64 mdt_mount_count; - - /* last_rcvd data */ - struct mdt_server_data mdt_msd; - spinlock_t mdt_client_bitmap_lock; - unsigned long mdt_client_bitmap[(LR_MAX_CLIENTS >> 3) / sizeof(long)]; struct upcall_cache *mdt_identity_cache; @@ -195,14 +154,31 @@ struct mdt_device { struct lustre_capa_key mdt_capa_keys[2]; unsigned int mdt_capa_conf:1; + /* root squash */ + uid_t mdt_squash_uid; + gid_t mdt_squash_gid; + struct list_head mdt_nosquash_nids; + char *mdt_nosquash_str; + int mdt_nosquash_strlen; + struct rw_semaphore mdt_squash_sem; + cfs_proc_dir_entry_t *mdt_proc_entry; struct lprocfs_stats *mdt_stats; + int mdt_sec_level; }; -/*XXX copied from mds_internal.h */ -#define MDT_SERVICE_WATCHDOG_TIMEOUT (obd_timeout * 1000) +#define mdt_transno_lock mdt_lut.lut_translock +#define mdt_last_transno mdt_lut.lut_last_transno +#define mdt_last_rcvd mdt_lut.lut_last_rcvd +#define mdt_mount_count mdt_lut.lut_mount_count +#define mdt_lsd mdt_lut.lut_lsd +#define mdt_client_bitmap_lock mdt_lut.lut_client_bitmap_lock +#define mdt_client_bitmap mdt_lut.lut_client_bitmap + +#define MDT_SERVICE_WATCHDOG_FACTOR (2) #define MDT_ROCOMPAT_SUPP (OBD_ROCOMPAT_LOVOBJID) #define MDT_INCOMPAT_SUPP (OBD_INCOMPAT_MDT | OBD_INCOMPAT_COMMON_LR) +#define MDT_COS_DEFAULT (1) struct mdt_object { struct lu_object_header mot_header; @@ -333,7 +309,7 @@ struct mdt_thread_info { /* * XXX: Part Three: - * The following members will be filled expilictly + * The following members will be filled explicitly * with zero in mdt_reint_unpack(), because they are only used * by reint requests (including mdt_reint_open()). */ @@ -343,6 +319,9 @@ struct mdt_thread_info { */ struct mdt_reint_record mti_rr; + /** md objects included in operation */ + struct mdt_object *mti_mos[PTLRPC_NUM_VERSIONS]; + /* * Operation specification (currently create and lookup) */ @@ -381,8 +360,8 @@ struct mdt_thread_info { __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; @@ -390,8 +369,14 @@ struct mdt_thread_info { /* Ops object filename */ struct lu_name mti_name; + struct md_attr mti_tmp_attr; }; +#define mti_parent mti_mos[0] +#define mti_child mti_mos[1] +#define mti_parent1 mti_mos[2] +#define mti_child1 mti_mos[3] + typedef void (*mdt_cb_t)(const struct mdt_device *mdt, __u64 transno, void *data, int err); struct mdt_commit_cb { @@ -399,6 +384,11 @@ struct mdt_commit_cb { void *mdt_cb_data; }; +enum mdt_txn_op { + MDT_TXN_CAPA_KEYS_WRITE_OP, + MDT_TXN_LAST_RCVD_WRITE_OP, +}; + /* * Info allocated per-transaction. */ @@ -406,26 +396,27 @@ struct mdt_commit_cb { struct mdt_txn_info { __u64 txi_transno; unsigned int txi_cb_count; - struct mdt_commit_cb txi_cb[MDT_MAX_COMMIT_CB]; + struct lut_commit_cb txi_cb[MDT_MAX_COMMIT_CB]; }; 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) + lut_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)); /* add new callback */ - txi->txi_cb[txi->txi_cb_count].mdt_cb_func = cb_func; - txi->txi_cb[txi->txi_cb_count].mdt_cb_data = cb_data; + txi->txi_cb[txi->txi_cb_count].lut_cb_func = cb_func; + txi->txi_cb[txi->txi_cb_count].lut_cb_data = cb_data; 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; @@ -474,6 +465,22 @@ static inline const struct lu_fid *mdt_object_fid(struct mdt_object *o) 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); @@ -533,10 +540,10 @@ void mdt_reconstruct_generic(struct mdt_thread_info *mti, struct mdt_lock_handle *lhc); extern void target_recovery_fini(struct obd_device *obd); -extern void target_recovery_init(struct obd_device *obd, +extern void target_recovery_init(struct lu_target *lut, svc_handler_t handler); int mdt_fs_setup(const struct lu_env *, struct mdt_device *, - struct obd_device *); + struct obd_device *, struct lustre_sb_info *lsi); void mdt_fs_cleanup(const struct lu_env *, struct mdt_device *); int mdt_client_del(const struct lu_env *env, @@ -547,6 +554,10 @@ int mdt_client_add(const struct lu_env *env, 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, @@ -577,8 +588,12 @@ void mdt_shrink_reply(struct mdt_thread_info *info); int mdt_handle_last_unlink(struct mdt_thread_info *, struct mdt_object *, const struct md_attr *); void mdt_reconstruct_open(struct mdt_thread_info *, struct mdt_lock_handle *); + +void mdt_trans_credit_init(const struct lu_env *env, + struct mdt_device *mdt, + enum mdt_txn_op op); struct thandle* mdt_trans_start(const struct lu_env *env, - struct mdt_device *mdt, int credits); + struct mdt_device *mdt); void mdt_trans_stop(const struct lu_env *env, struct mdt_device *mdt, struct thandle *th); int mdt_record_write(const struct lu_env *env, @@ -597,23 +612,19 @@ 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 mdt_version_get_check(struct mdt_thread_info *, int); /* mdt_idmap.c */ +int mdt_init_sec_level(struct mdt_thread_info *); int mdt_init_idmap(struct mdt_thread_info *); - void mdt_cleanup_idmap(struct mdt_export_data *); - int mdt_handle_idmap(struct mdt_thread_info *); - int ptlrpc_user_desc_do_idmap(struct ptlrpc_request *, struct ptlrpc_user_desc *); - void mdt_body_reverse_idmap(struct mdt_thread_info *, struct mdt_body *); - int mdt_remote_perm_reverse_idmap(struct ptlrpc_request *, struct mdt_remote_perm *); - int mdt_fix_attr_ucred(struct mdt_thread_info *, __u32); static inline struct mdt_device *mdt_dev(struct lu_device *d) @@ -670,7 +681,7 @@ static inline int mdt_check_resent(struct mdt_thread_info *info, 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); } @@ -682,13 +693,15 @@ static inline int is_identity_get_disabled(struct upcall_cache *cache) return cache ? (strcmp(cache->uc_upcall, "NONE") == 0) : 1; } +int mdt_blocking_ast(struct ldlm_lock*, struct ldlm_lock_desc*, void*, int); + /* Issues dlm lock on passed @ns, @f stores it lock handle into @lh. */ static inline int mdt_fid_lock(struct ldlm_namespace *ns, struct lustre_handle *lh, ldlm_mode_t mode, ldlm_policy_data_t *policy, const struct ldlm_res_id *res_id, - int flags) + int flags, const __u64 *client_cookie) { int rc; @@ -696,9 +709,9 @@ static inline int mdt_fid_lock(struct ldlm_namespace *ns, LASSERT(lh != NULL); rc = ldlm_cli_enqueue_local(ns, res_id, LDLM_IBITS, policy, - mode, &flags, ldlm_blocking_ast, - ldlm_completion_ast, NULL, NULL, - 0, NULL, lh); + mode, &flags, mdt_blocking_ast, + ldlm_completion_ast, + NULL, NULL, 0, NULL, client_cookie, lh); return rc == ELDLM_OK ? 0 : -EIO; } @@ -751,6 +764,9 @@ static inline struct lu_name *mdt_name_copy(struct lu_name *tlname, return tlname; } +void mdt_enable_cos(struct mdt_device *, int); +int mdt_cos_is_enabled(struct mdt_device *); + /* lprocfs stuff */ void lprocfs_mdt_init_vars(struct lprocfs_static_vars *lvars); int mdt_procfs_init(struct mdt_device *mdt, const char *name); @@ -773,11 +789,11 @@ static inline void mdt_set_capainfo(struct mdt_thread_info *info, int offset, const struct lu_fid *fid, struct lustre_capa *capa) { - struct mdt_device *dev = info->mti_mdt; struct md_capainfo *ci; LASSERT(offset >= 0 && offset <= MD_CAPAINFO_MAX); - if (!dev->mdt_opts.mo_mds_capa) + if (!info->mti_mdt->mdt_opts.mo_mds_capa || + !(info->mti_exp->exp_connect_flags & OBD_CONNECT_MDS_CAPA)) return; ci = md_capainfo(info->mti_env); @@ -810,5 +826,9 @@ static inline void mdt_dump_capainfo(struct mdt_thread_info *info) } } +static inline struct obd_device *mdt2obd_dev(const struct mdt_device *mdt) +{ + return mdt->mdt_md_dev.md_lu_dev.ld_obd; +} #endif /* __KERNEL__ */ #endif /* _MDT_H */