X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Finclude%2Fobd.h;h=7941a8374b24a2dd6fd5e7f51cbfac1cfddb09a8;hp=5ef03cc26fc683e841c0a6fd564fbf93843ed490;hb=a6db2409b0282b0a0e44d5d4e6eeb154b5c62dff;hpb=fda99d855f31b29259c0d72a415fa1f4ba289c7f diff --git a/lustre/include/obd.h b/lustre/include/obd.h index 5ef03cc..7941a83 100644 --- a/lustre/include/obd.h +++ b/lustre/include/obd.h @@ -1,6 +1,4 @@ -/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- - * vim:expandtab:shiftwidth=8:tabstop=8: - * +/* * GPL HEADER START * * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. @@ -26,8 +24,10 @@ * GPL HEADER END */ /* - * Copyright 2008 Sun Microsystems, Inc. All rights reserved + * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. + * + * Copyright (c) 2011, 2012, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -54,17 +54,15 @@ #define IOC_MDC_TYPE 'i' #define IOC_MDC_MIN_NR 20 -/* Moved to lustre_user.h -#define IOC_MDC_LOOKUP _IOWR(IOC_MDC_TYPE, 20, struct obd_ioctl_data *) -#define IOC_MDC_GETSTRIPE _IOWR(IOC_MDC_TYPE, 21, struct lov_mds_md *) */ #define IOC_MDC_MAX_NR 50 #include -#include +#ifdef HAVE_SERVER_SUPPORT +# include +#endif #include #include #include -#include #include #include @@ -73,14 +71,6 @@ #define MAX_OBD_DEVICES 8192 -/* this is really local to the OSC */ -struct loi_oap_pages { - cfs_list_t lop_pending; - cfs_list_t lop_urgent; - cfs_list_t lop_pending_group; - int lop_num_pending; -}; - struct osc_async_rc { int ar_rc; int ar_force_sync; @@ -88,24 +78,17 @@ struct osc_async_rc { }; struct lov_oinfo { /* per-stripe data structure */ - __u64 loi_id; /* object ID on the target OST */ - __u64 loi_gr; /* object group on the target OST */ + struct ost_id loi_oi; /* object ID/Sequence on the target OST */ int loi_ost_idx; /* OST stripe index in lov_tgt_desc->tgts */ int loi_ost_gen; /* generation of this loi_ost_idx */ - /* used by the osc to keep track of what objects to build into rpcs */ - struct loi_oap_pages loi_read_lop; - struct loi_oap_pages loi_write_lop; - cfs_list_t loi_ready_item; - cfs_list_t loi_hp_ready_item; - cfs_list_t loi_write_item; - cfs_list_t loi_read_item; - unsigned long loi_kms_valid:1; __u64 loi_kms; /* known minimum size */ struct ost_lvb loi_lvb; struct osc_async_rc loi_ar; }; +#define loi_id loi_oi.oi_id +#define loi_seq loi_oi.oi_seq static inline void loi_kms_set(struct lov_oinfo *oinfo, __u64 kms) { @@ -115,33 +98,27 @@ static inline void loi_kms_set(struct lov_oinfo *oinfo, __u64 kms) static inline void loi_init(struct lov_oinfo *loi) { - CFS_INIT_LIST_HEAD(&loi->loi_read_lop.lop_pending); - CFS_INIT_LIST_HEAD(&loi->loi_read_lop.lop_urgent); - CFS_INIT_LIST_HEAD(&loi->loi_read_lop.lop_pending_group); - CFS_INIT_LIST_HEAD(&loi->loi_write_lop.lop_pending); - CFS_INIT_LIST_HEAD(&loi->loi_write_lop.lop_urgent); - CFS_INIT_LIST_HEAD(&loi->loi_write_lop.lop_pending_group); - CFS_INIT_LIST_HEAD(&loi->loi_ready_item); - CFS_INIT_LIST_HEAD(&loi->loi_hp_ready_item); - CFS_INIT_LIST_HEAD(&loi->loi_write_item); - CFS_INIT_LIST_HEAD(&loi->loi_read_item); } struct lov_stripe_md { - cfs_spinlock_t lsm_lock; + cfs_atomic_t lsm_refc; + spinlock_t lsm_lock; pid_t lsm_lock_owner; /* debugging */ + /* maximum possible file size, might change as OSTs status changes, + * e.g. disconnected, deactivated */ + __u64 lsm_maxbytes; struct { /* Public members. */ __u64 lw_object_id; /* lov object id */ - __u64 lw_object_gr; /* lov object group */ - __u64 lw_maxbytes; /* maximum possible file size */ + __u64 lw_object_seq; /* lov object seq */ /* LOV-private members start here -- only for use in lov/. */ __u32 lw_magic; __u32 lw_stripe_size; /* size of the stripe */ __u32 lw_pattern; /* striping pattern (RAID0, RAID1) */ - unsigned lw_stripe_count; /* number of objects being striped over */ + __u16 lw_stripe_count; /* number of objects being striped over */ + __u16 lw_layout_gen; /* generation of the layout */ char lw_pool_name[LOV_MAXPOOLNAME]; /* pool name */ } lsm_wire; @@ -149,9 +126,9 @@ struct lov_stripe_md { }; #define lsm_object_id lsm_wire.lw_object_id -#define lsm_object_gr lsm_wire.lw_object_gr -#define lsm_maxbytes lsm_wire.lw_maxbytes +#define lsm_object_seq lsm_wire.lw_object_seq #define lsm_magic lsm_wire.lw_magic +#define lsm_layout_gen lsm_wire.lw_layout_gen #define lsm_stripe_size lsm_wire.lw_stripe_size #define lsm_pattern lsm_wire.lw_pattern #define lsm_stripe_count lsm_wire.lw_stripe_count @@ -171,8 +148,9 @@ struct obd_info { - while lock handling, the flags obtained on the enqueue request are set here. - while stats, the flags used for control delay/resend. + - while setattr, the flags used for distinguish punch operation */ - int oi_flags; + __u64 oi_flags; /* Lock handle specific for every OSC lock. */ struct lustre_handle *oi_lockh; /* lsm data specific for every OSC. */ @@ -189,6 +167,8 @@ struct obd_info { /* oss capability, its type is obd_capa in client to avoid copy. * in contrary its type is lustre_capa in OSS. */ void *oi_capa; + /* transfer jobid from ost_sync() to filter_sync()... */ + char *oi_jobid; }; /* compare all relevant fields. */ @@ -202,6 +182,57 @@ static inline int lov_stripe_md_cmp(struct lov_stripe_md *m1, return memcmp(&m1->lsm_wire, &m2->lsm_wire, sizeof m1->lsm_wire); } +static inline int lov_lum_lsm_cmp(struct lov_user_md *lum, + struct lov_stripe_md *lsm) +{ + if (lsm->lsm_magic != lum->lmm_magic) + return 1; + if ((lsm->lsm_stripe_count != 0) && (lum->lmm_stripe_count != 0) && + (lsm->lsm_stripe_count != lum->lmm_stripe_count)) + return 2; + if ((lsm->lsm_stripe_size != 0) && (lum->lmm_stripe_size != 0) && + (lsm->lsm_stripe_size != lum->lmm_stripe_size)) + return 3; + if ((lsm->lsm_pattern != 0) && (lum->lmm_pattern != 0) && + (lsm->lsm_pattern != lum->lmm_pattern)) + return 4; + if ((lsm->lsm_magic == LOV_MAGIC_V3) && + (strncmp(lsm->lsm_pool_name, + ((struct lov_user_md_v3 *)lum)->lmm_pool_name, + LOV_MAXPOOLNAME) != 0)) + return 5; + return 0; +} + +static inline int lov_lum_swab_if_needed(struct lov_user_md_v3 *lumv3, + int *lmm_magic, + struct lov_user_md *lum) +{ + if (lum && cfs_copy_from_user(lumv3, lum,sizeof(struct lov_user_md_v1))) + return -EFAULT; + + *lmm_magic = lumv3->lmm_magic; + + if (*lmm_magic == __swab32(LOV_USER_MAGIC_V1)) { + lustre_swab_lov_user_md_v1((struct lov_user_md_v1 *)lumv3); + *lmm_magic = LOV_USER_MAGIC_V1; + } else if (*lmm_magic == LOV_USER_MAGIC_V3) { + if (lum && cfs_copy_from_user(lumv3, lum, sizeof(*lumv3))) + return -EFAULT; + } else if (*lmm_magic == __swab32(LOV_USER_MAGIC_V3)) { + if (lum && cfs_copy_from_user(lumv3, lum, sizeof(*lumv3))) + return -EFAULT; + lustre_swab_lov_user_md_v3(lumv3); + *lmm_magic = LOV_USER_MAGIC_V3; + } else if (*lmm_magic != LOV_USER_MAGIC_V1) { + CDEBUG(D_IOCTL, + "bad userland LOV MAGIC: %#08x != %#08x nor %#08x\n", + *lmm_magic, LOV_USER_MAGIC_V1, LOV_USER_MAGIC_V3); + return -EINVAL; + } + return 0; +} + void lov_stripe_lock(struct lov_stripe_md *md); void lov_stripe_unlock(struct lov_stripe_md *md); @@ -213,7 +244,7 @@ struct obd_type { char *typ_name; int typ_refcnt; struct lu_device_type *typ_lu; - cfs_spinlock_t obd_type_lock; + spinlock_t obd_type_lock; }; struct brw_page { @@ -227,19 +258,27 @@ struct brw_page { struct ost_server_data; +struct osd_properties { + size_t osd_max_ea_size; +}; + +#define OBT_MAGIC 0xBDDECEAE /* hold common fields for "target" device */ struct obd_device_target { + __u32 obt_magic; + __u32 obt_instance; struct super_block *obt_sb; /** last_rcvd file */ struct file *obt_rcvd_filp; +#ifdef HAVE_SERVER_SUPPORT struct lu_target *obt_lut; +#endif __u64 obt_mount_count; - cfs_semaphore_t obt_quotachecking; - struct lustre_quota_ctxt obt_qctxt; - lustre_quota_version_t obt_qfmt; - cfs_rw_semaphore_t obt_rwsem; + struct rw_semaphore obt_rwsem; struct vfsmount *obt_vfsmnt; struct file *obt_health_check_filp; + struct osd_properties obt_osd_properties; + struct obd_job_stats obt_jobstats; }; /* llog contexts */ @@ -283,13 +322,13 @@ struct filter_obd { cfs_dentry_t *fo_dentry_O; cfs_dentry_t **fo_dentry_O_groups; struct filter_subdirs *fo_dentry_O_sub; - cfs_semaphore_t fo_init_lock; /* group initialization lock */ - int fo_committed_group; + struct mutex fo_init_lock; /* group initialization lock*/ + int fo_committed_group; - cfs_spinlock_t fo_objidlock; /* protect fo_lastobjid */ + spinlock_t fo_objidlock; /* protect fo_lastobjid */ - unsigned long fo_destroys_in_progress; - cfs_semaphore_t fo_create_locks[FILTER_SUBDIR_COUNT]; + unsigned long fo_destroys_in_progress; + struct mutex fo_create_locks[FILTER_SUBDIR_COUNT]; cfs_list_t fo_export_list; int fo_subdir_count; @@ -300,7 +339,8 @@ struct filter_obd { int fo_tot_granted_clients; obd_size fo_readcache_max_filesize; - int fo_read_cache:1, /**< enable read-only cache */ + spinlock_t fo_flags_lock; + unsigned int fo_read_cache:1, /**< enable read-only cache */ fo_writethrough_cache:1,/**< read cache writes */ fo_mds_ost_sync:1, /**< MDS-OST orphan recovery*/ fo_raid_degraded:1;/**< RAID device degraded */ @@ -312,39 +352,37 @@ struct filter_obd { __u64 *fo_last_objids; /* last created objid for groups, * protected by fo_objidlock */ - cfs_semaphore_t fo_alloc_lock; + struct mutex fo_alloc_lock; cfs_atomic_t fo_r_in_flight; cfs_atomic_t fo_w_in_flight; - /* - * per-filter pool of kiobuf's allocated by filter_common_setup() and - * torn down by filter_cleanup(). Contains OST_NUM_THREADS elements of - * which ->fo_iobuf_count were allocated. - * - * This pool contains kiobuf used by - * filter_{prep,commit}rw_{read,write}() and is shared by all OST - * threads. - * - * Locking: none, each OST thread uses only one element, determined by - * its "ordinal number", ->t_id. - */ - struct filter_iobuf **fo_iobuf_pool; - int fo_iobuf_count; - - cfs_list_t fo_llog_list; - cfs_spinlock_t fo_llog_list_lock; + /* + * per-filter pool of kiobuf's allocated by filter_common_setup() and + * torn down by filter_cleanup(). + * + * This pool contains kiobuf used by + * filter_{prep,commit}rw_{read,write}() and is shared by all OST + * threads. + * + * Locking: protected by internal lock of cfs_hash, pool can be + * found from this hash table by t_id of ptlrpc_thread. + */ + struct cfs_hash *fo_iobuf_hash; + + cfs_list_t fo_llog_list; + spinlock_t fo_llog_list_lock; struct brw_stats fo_filter_stats; - struct lustre_quota_ctxt fo_quota_ctxt; - cfs_spinlock_t fo_quotacheck_lock; - cfs_atomic_t fo_quotachecking; int fo_fmd_max_num; /* per exp filter_mod_data */ int fo_fmd_max_age; /* jiffies to fmd expiry */ + unsigned long fo_syncjournal:1, /* sync journal on writes */ + fo_sync_lock_cancel:2;/* sync on lock cancel */ + /* sptlrpc stuff */ - cfs_rwlock_t fo_sptlrpc_lock; + rwlock_t fo_sptlrpc_lock; struct sptlrpc_rule_set fo_sptlrpc_rset; /* capability related */ @@ -365,18 +403,27 @@ struct timeout_item { }; #define OSC_MAX_RIF_DEFAULT 8 +#define MDS_OSC_MAX_RIF_DEFAULT 50 #define OSC_MAX_RIF_MAX 256 #define OSC_MAX_DIRTY_DEFAULT (OSC_MAX_RIF_DEFAULT * 4) #define OSC_MAX_DIRTY_MB_MAX 2048 /* arbitrary, but < MAX_LONG bytes */ #define OSC_DEFAULT_RESENDS 10 +/* possible values for fo_sync_lock_cancel */ +enum { + NEVER_SYNC_ON_CANCEL = 0, + BLOCKING_SYNC_ON_CANCEL = 1, + ALWAYS_SYNC_ON_CANCEL = 2, + NUM_SYNC_ON_CANCEL_STATES +}; + #define MDC_MAX_RIF_DEFAULT 8 #define MDC_MAX_RIF_MAX 512 struct mdc_rpc_lock; struct obd_import; struct client_obd { - cfs_rw_semaphore_t cl_sem; + struct rw_semaphore cl_sem; struct obd_uuid cl_target_uuid; struct obd_import *cl_import; /* ptlrpc connection state */ int cl_conn_count; @@ -399,11 +446,22 @@ struct client_obd { long cl_dirty_transit; /* dirty synchronous */ long cl_avail_grant; /* bytes of credit for ost */ long cl_lost_grant; /* lost credits (trunc) */ - cfs_list_t cl_cache_waiters; /* waiting for cache/grant */ - cfs_time_t cl_next_shrink_grant; /* jiffies */ - cfs_list_t cl_grant_shrink_list; /* Timeout event list */ - cfs_semaphore_t cl_grant_sem; /*grant shrink list cfs_semaphore*/ - int cl_grant_shrink_interval; /* seconds */ + + /* since we allocate grant by blocks, we don't know how many grant will + * be used to add a page into cache. As a solution, we reserve maximum + * grant before trying to dirty a page and unreserve the rest. + * See osc_{reserve|unreserve}_grant for details. */ + long cl_reserved_grant; + cfs_list_t cl_cache_waiters; /* waiting for cache/grant */ + cfs_time_t cl_next_shrink_grant; /* jiffies */ + cfs_list_t cl_grant_shrink_list; /* Timeout event list */ + int cl_grant_shrink_interval; /* seconds */ + + /* A chunk is an optimal size used by osc_extent to determine + * the extent size. A chunk is max(CFS_PAGE_SIZE, OST block size) */ + int cl_chunkbits; + int cl_chunk; + int cl_extent_tax; /* extent overhead, by bytes */ /* keep track of objects that have lois that contain pages which * have been queued for async brw. this lock also protects the @@ -422,7 +480,10 @@ struct client_obd { * Exact type of ->cl_loi_list_lock is defined in arch/obd.h together * with client_obd_list_{un,}lock() and * client_obd_list_lock_{init,done}() functions. - */ + * + * NB by Jinshan: though field names are still _loi_, but actually + * osc_object{}s are in the list. + */ client_obd_lock_t cl_loi_list_lock; cfs_list_t cl_loi_ready_list; cfs_list_t cl_loi_hp_ready_list; @@ -431,9 +492,9 @@ struct client_obd { int cl_r_in_flight; int cl_w_in_flight; /* just a sum of the loi/lop pending numbers to be exported by /proc */ - int cl_pending_w_pages; - int cl_pending_r_pages; - int cl_max_pages_per_rpc; + cfs_atomic_t cl_pending_w_pages; + cfs_atomic_t cl_pending_r_pages; + int cl_max_pages_per_rpc; int cl_max_rpcs_in_flight; struct obd_histogram cl_read_rpc_hist; struct obd_histogram cl_write_rpc_hist; @@ -442,17 +503,25 @@ struct client_obd { struct obd_histogram cl_read_offset_hist; struct obd_histogram cl_write_offset_hist; + /* lru for osc caching pages */ + struct cl_client_cache *cl_cache; + cfs_list_t cl_lru_osc; /* member of cl_cache->ccc_lru */ + cfs_atomic_t *cl_lru_left; + cfs_atomic_t cl_lru_busy; + cfs_atomic_t cl_lru_shrinkers; + cfs_atomic_t cl_lru_in_list; + cfs_list_t cl_lru_list; /* lru page list */ + client_obd_lock_t cl_lru_list_lock; /* page list protector */ + /* number of in flight destroy rpcs is limited to max_rpcs_in_flight */ cfs_atomic_t cl_destroy_in_flight; cfs_waitq_t cl_destroy_waitq; struct mdc_rpc_lock *cl_rpc_lock; - struct mdc_rpc_lock *cl_setattr_lock; struct mdc_rpc_lock *cl_close_lock; - struct osc_creator cl_oscc; /* mgc datastruct */ - cfs_semaphore_t cl_mgc_sem; + struct semaphore cl_mgc_sem; struct vfsmount *cl_mgc_vfsmnt; struct dentry *cl_mgc_configs_dir; cfs_atomic_t cl_mgc_refcount; @@ -468,95 +537,24 @@ struct client_obd { /* also protected by the poorly named _loi_list_lock lock above */ struct osc_async_rc cl_ar; - /* used by quotacheck */ - int cl_qchk_stat; /* quotacheck stat of the peer */ + /* used by quotacheck when the servers are older than 2.4 */ + int cl_qchk_stat; /* quotacheck stat of the peer */ +#define CL_NOT_QUOTACHECKED 1 /* client->cl_qchk_stat init value */ +#if LUSTRE_VERSION_CODE >= OBD_OCD_VERSION(2, 7, 50, 0) +#warning "please consider removing quotacheck compatibility code" +#endif /* sequence manager */ struct lu_client_seq *cl_seq; cfs_atomic_t cl_resends; /* resend count */ -}; -#define obd2cli_tgt(obd) ((char *)(obd)->u.cli.cl_target_uuid.uuid) -#define CL_NOT_QUOTACHECKED 1 /* client->cl_qchk_stat init value */ - -struct mgs_obd { - struct ptlrpc_service *mgs_service; - struct vfsmount *mgs_vfsmnt; - struct super_block *mgs_sb; - struct dentry *mgs_configs_dir; - struct dentry *mgs_fid_de; - cfs_list_t mgs_fs_db_list; - cfs_semaphore_t mgs_sem; - cfs_proc_dir_entry_t *mgs_proc_live; + /* ptlrpc work for writeback in ptlrpcd context */ + void *cl_writeback_work; + /* hash tables for osc_quota_info */ + cfs_hash_t *cl_quota_hash[MAXQUOTAS]; }; - -struct mds_obd { - /* NB this field MUST be first */ - struct obd_device_target mds_obt; - struct ptlrpc_service *mds_service; - struct ptlrpc_service *mds_setattr_service; - struct ptlrpc_service *mds_readpage_service; - cfs_dentry_t *mds_fid_de; - int mds_max_mdsize; - int mds_max_cookiesize; - __u64 mds_io_epoch; - unsigned long mds_atime_diff; - cfs_semaphore_t mds_epoch_sem; - struct ll_fid mds_rootfid; - cfs_dentry_t *mds_pending_dir; - cfs_dentry_t *mds_logs_dir; - cfs_dentry_t *mds_objects_dir; - struct llog_handle *mds_cfg_llh; - struct obd_device *mds_osc_obd; /* XXX lov_obd */ - struct obd_uuid mds_lov_uuid; - char *mds_profile; - struct obd_export *mds_osc_exp; /* XXX lov_exp */ - struct lov_desc mds_lov_desc; - __u32 mds_id; - - /* mark pages dirty for write. */ - cfs_bitmap_t *mds_lov_page_dirty; - /* array for store pages with obd_id */ - void **mds_lov_page_array; - /* file for store objid */ - struct file *mds_lov_objid_filp; - __u32 mds_lov_objid_count; - __u32 mds_lov_objid_max_index; - __u32 mds_lov_objid_lastpage; - __u32 mds_lov_objid_lastidx; - - - struct lustre_quota_info mds_quota_info; - cfs_rw_semaphore_t mds_qonoff_sem; - cfs_semaphore_t mds_health_sem; - unsigned long mds_fl_user_xattr:1, - mds_fl_acl:1, - mds_evict_ost_nids:1, - mds_fl_cfglog:1, - mds_fl_synced:1, - mds_quota:1, - mds_fl_target:1; /* mds have one or - * more targets */ - - struct upcall_cache *mds_identity_cache; - - /* for capability keys update */ - struct lustre_capa_key *mds_capa_keys; - cfs_rw_semaphore_t mds_notify_lock; -}; - -/* lov objid */ -extern __u32 mds_max_ost_index; - -#define MDS_LOV_ALLOC_SIZE (CFS_PAGE_SIZE) - -#define OBJID_PER_PAGE() (MDS_LOV_ALLOC_SIZE / sizeof(obd_id)) - -#define MDS_LOV_OBJID_PAGES_COUNT (mds_max_ost_index/OBJID_PER_PAGE()) - -extern int mds_lov_init_objids(struct obd_device *obd); -extern void mds_lov_destroy_objids(struct obd_device *obd); +#define obd2cli_tgt(obd) ((char *)(obd)->u.cli.cl_target_uuid.uuid) struct obd_id_info { __u32 idx; @@ -566,23 +564,25 @@ struct obd_id_info { /* */ struct echo_obd { - struct obdo eo_oa; - cfs_spinlock_t eo_lock; - __u64 eo_lastino; - struct lustre_handle eo_nl_lock; - cfs_atomic_t eo_prep; + struct obd_device_target eo_obt; + struct obdo eo_oa; + spinlock_t eo_lock; + __u64 eo_lastino; + struct lustre_handle eo_nl_lock; + cfs_atomic_t eo_prep; }; struct ost_obd { - struct ptlrpc_service *ost_service; - struct ptlrpc_service *ost_create_service; - struct ptlrpc_service *ost_io_service; - cfs_semaphore_t ost_health_sem; + struct ptlrpc_service *ost_service; + struct ptlrpc_service *ost_create_service; + struct ptlrpc_service *ost_io_service; + struct ptlrpc_service *ost_seq_service; + struct mutex ost_health_mutex; }; struct echo_client_obd { - struct obd_export *ec_exp; /* the local connection to osc/lov */ - cfs_spinlock_t ec_lock; + struct obd_export *ec_exp; /* the local connection to osc/lov */ + spinlock_t ec_lock; cfs_list_t ec_objects; cfs_list_t ec_locks; int ec_nstripes; @@ -614,7 +614,7 @@ struct ost_pool { lov_obd->lov_tgts */ unsigned int op_count; /* number of OSTs in the array */ unsigned int op_size; /* allocated size of lp_array */ - cfs_rw_semaphore_t op_rw_sem; /* to protect ost_pool use */ + struct rw_semaphore op_rw_sem; /* to protect ost_pool use */ }; /* Round-robin allocator data */ @@ -626,6 +626,9 @@ struct lov_qos_rr { unsigned long lqr_dirty:1; /* recalc round-robin list */ }; +/* allow statfs data caching for 1 second */ +#define OBD_STATFS_CACHE_SECONDS 1 + struct lov_statfs_data { struct obd_info lsd_oi; struct obd_statfs lsd_statfs; @@ -633,7 +636,7 @@ struct lov_statfs_data { /* Stripe placement optimization */ struct lov_qos { cfs_list_t lq_oss_list; /* list of OSSs that targets use */ - cfs_rw_semaphore_t lq_rw_sem; + struct rw_semaphore lq_rw_sem; __u32 lq_active_oss_count; unsigned int lq_prio_free; /* priority for free space */ unsigned int lq_threshold_rr;/* priority for rr */ @@ -668,7 +671,6 @@ struct lov_tgt_desc { #define pool_tgt_count(_p) _p->pool_obds.op_count #define pool_tgt_array(_p) _p->pool_obds.op_array #define pool_tgt_rw_sem(_p) _p->pool_obds.op_rw_sem -#define pool_tgt(_p, _i) _p->pool_lov->lov_tgts[_p->pool_obds.op_array[_i]] struct pool_desc { char pool_name[LOV_MAXPOOLNAME + 1]; /* name of pool */ @@ -678,8 +680,8 @@ struct pool_desc { cfs_hlist_node_t pool_hash; /* access by poolname */ cfs_list_t pool_list; /* serial access */ cfs_proc_dir_entry_t *pool_proc_entry; /* file in /proc */ - struct lov_obd *pool_lov; /* lov obd to which this - pool belong */ + struct obd_device *pool_lobd; /* obd of the lov/lod to which + * this pool belongs */ }; struct lov_obd { @@ -687,9 +689,8 @@ struct lov_obd { struct lov_tgt_desc **lov_tgts; /* sparse array */ struct ost_pool lov_packed; /* all OSTs in a packed array */ - cfs_semaphore_t lov_lock; + struct mutex lov_lock; struct obd_connect_data lov_ocd; - struct lov_qos lov_qos; /* qos info per lov */ cfs_atomic_t lov_refcount; __u32 lov_tgt_count; /* how many OBD's */ __u32 lov_active_tgt_count; /* how many active */ @@ -701,6 +702,9 @@ struct lov_obd { cfs_list_t lov_pool_list; /* used for sequential access */ cfs_proc_dir_entry_t *lov_pool_proc_entry; enum lustre_sec_part lov_sp_me; + + /* Cached LRU pages from upper layer */ + void *lov_cache; }; struct lmv_tgt_desc { @@ -708,7 +712,7 @@ struct lmv_tgt_desc { struct obd_export *ltd_exp; int ltd_active; /* is this target up for requests */ int ltd_idx; - cfs_semaphore_t ltd_fid_sem; + struct mutex ltd_fid_mutex; }; enum placement_policy { @@ -721,9 +725,9 @@ enum placement_policy { typedef enum placement_policy placement_policy_t; struct lmv_obd { - int refcount; - struct lu_client_fld lmv_fld; - cfs_spinlock_t lmv_lock; + int refcount; + struct lu_client_fld lmv_fld; + spinlock_t lmv_lock; placement_policy_t lmv_placement; struct lmv_desc desc; struct obd_uuid cluuid; @@ -734,7 +738,7 @@ struct lmv_obd { int max_def_easize; int max_cookiesize; int server_timeout; - cfs_semaphore_t init_sem; + struct mutex init_mutex; struct lmv_tgt_desc *tgts; int tgts_size; @@ -746,7 +750,8 @@ struct lmv_obd { }; struct niobuf_local { - __u64 offset; + __u64 lnb_file_offset; + __u32 lnb_page_offset; __u32 len; __u32 flags; cfs_page_t *page; @@ -760,11 +765,14 @@ struct niobuf_local { #define LUSTRE_CMM_NAME "cmm" #define LUSTRE_MDD_NAME "mdd" -#define LUSTRE_OSD_NAME "osd" +#define LUSTRE_OSD_LDISKFS_NAME "osd-ldiskfs" +#define LUSTRE_OSD_ZFS_NAME "osd-zfs" #define LUSTRE_VVP_NAME "vvp" #define LUSTRE_LMV_NAME "lmv" #define LUSTRE_CMM_MDC_NAME "cmm-mdc" #define LUSTRE_SLP_NAME "slp" +#define LUSTRE_LOD_NAME "lod" +#define LUSTRE_OSP_NAME "osp" /* obd device type names */ /* FIXME all the references to LUSTRE_MDS_NAME should be swapped with LUSTRE_MDT_NAME */ @@ -781,6 +789,7 @@ struct niobuf_local { #define LUSTRE_CACHEOBD_NAME "cobd" #define LUSTRE_ECHO_NAME "obdecho" #define LUSTRE_ECHO_CLIENT_NAME "echo_client" +#define LUSTRE_QMT_NAME "qmt" /* Constant obd names (post-rename) */ #define LUSTRE_MDS_OBDNAME "MDS" @@ -791,6 +800,33 @@ struct niobuf_local { /* Don't conflict with on-wire flags OBD_BRW_WRITE, etc */ #define N_LOCAL_TEMP_PAGE 0x10000000 +static inline int is_osp_on_ost(char *name) +{ + char *ptr; + + ptr = strrchr(name, '-'); + if (ptr == NULL) { + CERROR("%s is not a obdname\n", name); + return 0; + } + + if (strncmp(ptr + 1, "OST", 3) != 0 && strncmp(ptr + 1, "MDT", 3) != 0) + return 0; + + /* match the "-osp" */ + if (ptr - name < strlen(LUSTRE_OSP_NAME) + 1) + return 0; + + ptr -= (strlen(LUSTRE_OSP_NAME) + 1); + if (*ptr != '-') + return 0; + + if (strncmp(ptr + 1, LUSTRE_OSP_NAME, strlen(LUSTRE_OSP_NAME)) != 0) + return 0; + + return 1; +} + struct obd_trans_info { __u64 oti_transno; __u64 oti_xid; @@ -803,12 +839,16 @@ struct obd_trans_info { struct llog_cookie oti_onecookie; struct llog_cookie *oti_logcookies; int oti_numcookies; + /** synchronous write is needed */ + long oti_sync_write:1; /* initial thread handling transaction */ struct ptlrpc_thread * oti_thread; __u32 oti_conn_cnt; /** VBR: versions */ __u64 oti_pre_version; + /** JobID */ + char *oti_jobid; struct obd_uuid *oti_ost_uuid; }; @@ -848,8 +888,8 @@ static inline void oti_alloc_cookies(struct obd_trans_info *oti,int num_cookies) if (num_cookies == 1) oti->oti_logcookies = &oti->oti_onecookie; else - OBD_ALLOC(oti->oti_logcookies, - num_cookies * sizeof(oti->oti_onecookie)); + OBD_ALLOC_LARGE(oti->oti_logcookies, + num_cookies * sizeof(oti->oti_onecookie)); oti->oti_numcookies = num_cookies; } @@ -862,8 +902,8 @@ static inline void oti_free_cookies(struct obd_trans_info *oti) if (oti->oti_logcookies == &oti->oti_onecookie) LASSERT(oti->oti_numcookies == 1); else - OBD_FREE(oti->oti_logcookies, - oti->oti_numcookies * sizeof(oti->oti_onecookie)); + OBD_FREE_LARGE(oti->oti_logcookies, + oti->oti_numcookies*sizeof(oti->oti_onecookie)); oti->oti_logcookies = NULL; oti->oti_numcookies = 0; } @@ -872,6 +912,8 @@ static inline void oti_free_cookies(struct obd_trans_info *oti) * Events signalled through obd_notify() upcall-chain. */ enum obd_notify_event { + /* target added */ + OBD_NOTIFY_CREATE, /* Device connect start */ OBD_NOTIFY_CONNECT, /* Device activated */ @@ -887,8 +929,9 @@ enum obd_notify_event { OBD_NOTIFY_SYNC, /* Configuration event */ OBD_NOTIFY_CONFIG, - /* Trigger quota recovery */ - OBD_NOTIFY_QUOTA + /* Administratively deactivate/activate event */ + OBD_NOTIFY_DEACTIVATE, + OBD_NOTIFY_ACTIVATE }; /* bit-mask flags for config events */ @@ -910,17 +953,10 @@ struct obd_notify_upcall { }; struct target_recovery_data { - svc_handler_t trd_recovery_handler; - pid_t trd_processing_task; - cfs_completion_t trd_starting; - cfs_completion_t trd_finishing; -}; - -enum filter_groups { - FILTER_GROUP_MDS0 = 0, - FILTER_GROUP_LLOG = 1, - FILTER_GROUP_ECHO = 2 , - FILTER_GROUP_MDS1_N_BASE = 3 + svc_handler_t trd_recovery_handler; + pid_t trd_processing_task; + struct completion trd_starting; + struct completion trd_finishing; }; /** @@ -933,60 +969,42 @@ enum filter_groups { * 2. The group number indexing starts from 0 instead of 3 */ -static inline int filter_group_is_mds(obd_gr group) -{ - return (group == FILTER_GROUP_MDS0 || - group >= FILTER_GROUP_MDS1_N_BASE); -} - -#define LASSERT_MDS_GROUP(group) LASSERT(filter_group_is_mds(group)) +#define LASSERT_SEQ_IS_MDT(seq) LASSERT(fid_seq_is_mdt(seq)) -static inline __u64 objgrp_to_mdsno(obd_gr group) +static inline __u64 objseq_to_mdsno(obd_seq seq) { - LASSERT(filter_group_is_mds(group)); - if (group == FILTER_GROUP_MDS0) + LASSERT_SEQ_IS_MDT(seq); + if (seq == FID_SEQ_OST_MDT0) return 0; - return group - FILTER_GROUP_MDS1_N_BASE + 1; + return seq - FID_SEQ_OST_MDT1 + 1; } -static inline int mdt_to_obd_objgrp(int mdtid) +static inline int mdt_to_obd_objseq(int mdtid) { /** - * MDS0 uses group 0 always, other MDSes will use groups from - * FILTER_GROUP_MDS1_N_BASE + * MDS0 uses seq 0 pre FID-on-OST, other MDSes will use seq from + * FID_SEQ_OST_MDT1 */ if (mdtid) - return FILTER_GROUP_MDS1_N_BASE + mdtid - 1; + return FID_SEQ_OST_MDT1 + mdtid - 1; return 0; } -static inline __u64 obdo_mdsno(struct obdo *oa) -{ - LASSERT((oa->o_valid & OBD_MD_FLGROUP)); - return objgrp_to_mdsno(oa->o_gr); -} - -static inline int obdo_is_mds(struct obdo *oa) -{ - LASSERT(oa->o_valid & OBD_MD_FLGROUP); - return filter_group_is_mds(oa->o_gr); -} - struct obd_llog_group { cfs_list_t olg_list; - int olg_group; + int olg_seq; struct llog_ctxt *olg_ctxts[LLOG_MAX_CTXTS]; cfs_waitq_t olg_waitq; - cfs_spinlock_t olg_lock; - struct obd_export *olg_exp; - int olg_initializing; - cfs_semaphore_t olg_cat_processing; + spinlock_t olg_lock; + struct obd_export *olg_exp; + int olg_initializing; + struct mutex olg_cat_processing; }; /* corresponds to one of the obd's */ -#define MAX_OBD_NAME 128 #define OBD_DEVICE_MAGIC 0XAB5CD6EF #define OBD_DEV_BY_DEVNAME 0xffffd0de + struct obd_device { struct obd_type *obd_type; __u32 obd_magic; @@ -998,16 +1016,15 @@ struct obd_device { struct lu_device *obd_lu_dev; int obd_minor; + /* bitfield modification is protected by obd_dev_lock */ unsigned long obd_attached:1, /* finished attach */ obd_set_up:1, /* finished setup */ obd_recovering:1, /* there are recoverable clients */ obd_abort_recovery:1,/* recovery expired */ obd_version_recov:1, /* obd uses version checking */ - obd_recovery_expired:1, obd_replayable:1, /* recovery is enabled; inform clients */ obd_no_transno:1, /* no committed-transno notification */ obd_no_recov:1, /* fail instead of retry messages */ - obd_req_replaying:1, /* replaying requests */ obd_stopping:1, /* started cleanup */ obd_starting:1, /* started setup */ obd_force:1, /* cleanup with > 0 obd refcount */ @@ -1016,7 +1033,11 @@ struct obd_device { obd_no_conn:1, /* deny new connections */ obd_inactive:1, /* device active/inactive * (for /proc/status only!!) */ + obd_no_ir:1, /* no imperative recovery. */ obd_process_conf:1; /* device is processing mgs config */ + /* use separate field as it is set in interrupt to don't mess with + * protection of other bits using _bh lock */ + unsigned long obd_recovery_expired:1; /* uuid-export hash body */ cfs_hash_t *obd_uuid_hash; /* nid-export hash body */ @@ -1030,21 +1051,21 @@ struct obd_device { cfs_list_t obd_unlinked_exports; cfs_list_t obd_delayed_exports; int obd_num_exports; - cfs_spinlock_t obd_nid_lock; - struct ldlm_namespace *obd_namespace; - struct ptlrpc_client obd_ldlm_client; /* XXX OST/MDS only */ - /* a spinlock is OK for what we do now, may need a semaphore later */ - cfs_spinlock_t obd_dev_lock; - cfs_semaphore_t obd_dev_sem; - __u64 obd_last_committed; - struct fsfilt_operations *obd_fsops; - cfs_spinlock_t obd_osfs_lock; - struct obd_statfs obd_osfs; /* locked by obd_osfs_lock */ - __u64 obd_osfs_age; - struct lvfs_run_ctxt obd_lvfs_ctxt; - struct obd_llog_group obd_olg; /* default llog group */ - struct obd_device *obd_observer; - cfs_rw_semaphore_t obd_observer_link_sem; + spinlock_t obd_nid_lock; + struct ldlm_namespace *obd_namespace; + struct ptlrpc_client obd_ldlm_client; /* XXX OST/MDS only */ + /* a spinlock is OK for what we do now, may need a semaphore later */ + spinlock_t obd_dev_lock; /* protect OBD bitfield above */ + struct mutex obd_dev_mutex; + __u64 obd_last_committed; + struct fsfilt_operations *obd_fsops; + spinlock_t obd_osfs_lock; + struct obd_statfs obd_osfs; /* locked by obd_osfs_lock */ + __u64 obd_osfs_age; + struct lvfs_run_ctxt obd_lvfs_ctxt; + struct obd_llog_group obd_olg; /* default llog group */ + struct obd_device *obd_observer; + struct rw_semaphore obd_observer_link_sem; struct obd_notify_upcall obd_upcall; struct obd_export *obd_self_export; /* list of exports in LRU order, for ping evictor, with obd_dev_lock */ @@ -1052,25 +1073,30 @@ struct obd_device { time_t obd_eviction_timer; /* for ping evictor */ int obd_max_recoverable_clients; - int obd_connected_clients; + cfs_atomic_t obd_connected_clients; int obd_stale_clients; int obd_delayed_clients; - cfs_spinlock_t obd_processing_task_lock; /* BH lock (timer) */ + /* this lock protects all recovery list_heads, timer and + * obd_next_recovery_transno value */ + spinlock_t obd_recovery_task_lock; __u64 obd_next_recovery_transno; int obd_replayed_requests; int obd_requests_queued_for_recovery; cfs_waitq_t obd_next_transno_waitq; + /* protected by obd_recovery_task_lock */ cfs_timer_t obd_recovery_timer; time_t obd_recovery_start; /* seconds */ time_t obd_recovery_end; /* seconds, for lprocfs_status */ - time_t obd_recovery_time_hard; + int obd_recovery_time_hard; int obd_recovery_timeout; + int obd_recovery_ir_factor; /* new recovery stuff from CMD2 */ struct target_recovery_data obd_recovery_data; int obd_replayed_locks; cfs_atomic_t obd_req_replay_clients; cfs_atomic_t obd_lock_replay_clients; + /* all lists are protected by obd_recovery_task_lock */ cfs_list_t obd_req_replay_queue; cfs_list_t obd_lock_replay_queue; cfs_list_t obd_final_req_queue; @@ -1079,14 +1105,12 @@ struct obd_device { union { struct obd_device_target obt; struct filter_obd filter; - struct mds_obd mds; struct client_obd cli; struct ost_obd ost; struct echo_client_obd echo_client; struct echo_obd echo; struct lov_obd lov; struct lmv_obd lmv; - struct mgs_obd mgs; } u; /* Fields used by LProcFS */ unsigned int obd_cntr_base; @@ -1106,7 +1130,7 @@ struct obd_device { /** * Ldlm pool part. Save last calculated SLV and Limit. */ - cfs_rwlock_t obd_pool_lock; + rwlock_t obd_pool_lock; int obd_pool_limit; __u64 obd_pool_slv; @@ -1115,9 +1139,12 @@ struct obd_device { * debugging. */ struct lu_ref obd_reference; + + int obd_conn_inprogress; }; #define OBD_LLOG_FL_SENDNOW 0x0001 +#define OBD_LLOG_FL_EXIT 0x0002 enum obd_cleanup_stage { /* Special case hack for MDS LOVs */ @@ -1154,20 +1181,40 @@ enum obd_cleanup_stage { #define KEY_NEXT_ID "next_id" #define KEY_READ_ONLY "read-only" #define KEY_REGISTER_TARGET "register_target" -#define KEY_REVIMP_UPD "revimp_update" #define KEY_SET_FS "set_fs" +#define KEY_TGT_COUNT "tgt_count" /* KEY_SET_INFO in lustre_idl.h */ #define KEY_SPTLRPC_CONF "sptlrpc_conf" +#define KEY_CONNECT_FLAG "connect_flags" +#define KEY_SYNC_LOCK_CANCEL "sync_lock_cancel" +#define KEY_CACHE_SET "cache_set" +#define KEY_CACHE_LRU_SHRINK "cache_lru_shrink" struct lu_context; +/* /!\ must be coherent with include/linux/namei.h on patched kernel */ +#define IT_OPEN (1 << 0) +#define IT_CREAT (1 << 1) +#define IT_READDIR (1 << 2) +#define IT_GETATTR (1 << 3) +#define IT_LOOKUP (1 << 4) +#define IT_UNLINK (1 << 5) +#define IT_TRUNC (1 << 6) +#define IT_GETXATTR (1 << 7) +#define IT_EXEC (1 << 8) +#define IT_PIN (1 << 9) +#define IT_LAYOUT (1 << 10) +#define IT_QUOTA_DQACQ (1 << 11) +#define IT_QUOTA_CONN (1 << 12) + static inline int it_to_lock_mode(struct lookup_intent *it) { /* CREAT needs to be tested before open (both could be set) */ if (it->it_op & IT_CREAT) return LCK_CW; - else if (it->it_op & (IT_READDIR | IT_GETATTR | IT_OPEN | IT_LOOKUP)) + else if (it->it_op & (IT_READDIR | IT_GETATTR | IT_OPEN | IT_LOOKUP | + IT_LAYOUT)) return LCK_CR; LASSERTF(0, "Invalid it_op: %d\n", it->it_op); @@ -1181,7 +1228,7 @@ struct md_op_data { struct lu_fid op_fid4; /* to the operation locks. */ mdsno_t op_mds; /* what mds server open will go to */ struct lustre_handle op_handle; - __u64 op_mod_time; + obd_time op_mod_time; const char *op_name; int op_namelen; __u32 op_mode; @@ -1196,9 +1243,7 @@ struct md_op_data { /* iattr fields and blocks. */ struct iattr op_attr; #ifdef __KERNEL__ -#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,14) - unsigned int op_attr_flags; -#endif + unsigned int op_attr_flags; #endif __u64 op_valid; loff_t op_attr_blocks; @@ -1216,6 +1261,10 @@ struct md_op_data { /* Operation type */ __u32 op_opc; + + /* Used by readdir */ + __u32 op_npages; + __u64 op_offset; }; struct md_enqueue_info; @@ -1224,24 +1273,31 @@ typedef int (* md_enqueue_cb_t)(struct ptlrpc_request *req, struct md_enqueue_info *minfo, int rc); +/* seq client type */ +enum lu_cli_type { + LUSTRE_SEQ_METADATA = 1, + LUSTRE_SEQ_DATA +}; + struct md_enqueue_info { struct md_op_data mi_data; struct lookup_intent mi_it; struct lustre_handle mi_lockh; - struct dentry *mi_dentry; struct inode *mi_dir; md_enqueue_cb_t mi_cb; + __u64 mi_cbdata; unsigned int mi_generation; - void *mi_cbdata; }; struct obd_ops { cfs_module_t *o_owner; int (*o_iocontrol)(unsigned int cmd, struct obd_export *exp, int len, void *karg, void *uarg); - int (*o_get_info)(struct obd_export *, __u32 keylen, void *key, - __u32 *vallen, void *val, struct lov_stripe_md *lsm); - int (*o_set_info_async)(struct obd_export *, __u32 keylen, void *key, + int (*o_get_info)(const struct lu_env *env, struct obd_export *, + __u32 keylen, void *key, __u32 *vallen, void *val, + struct lov_stripe_md *lsm); + int (*o_set_info_async)(const struct lu_env *, struct obd_export *, + __u32 keylen, void *key, __u32 vallen, void *val, struct ptlrpc_request_set *set); int (*o_attach)(struct obd_device *dev, obd_count len, void *data); @@ -1272,7 +1328,7 @@ struct obd_ops { int (*o_disconnect)(struct obd_export *exp); /* Initialize/finalize fids infrastructure. */ - int (*o_fid_init)(struct obd_export *exp); + int (*o_fid_init)(struct obd_export *exp, enum lu_cli_type type); int (*o_fid_fini)(struct obd_export *exp); /* Allocate new fid according to passed @hint. */ @@ -1283,11 +1339,9 @@ struct obd_ops { * Object with @fid is getting deleted, we may want to do something * about this. */ - int (*o_fid_delete)(struct obd_export *exp, const struct lu_fid *fid); - - int (*o_statfs)(struct obd_device *obd, struct obd_statfs *osfs, - __u64 max_age, __u32 flags); - int (*o_statfs_async)(struct obd_device *obd, struct obd_info *oinfo, + int (*o_statfs)(const struct lu_env *, struct obd_export *exp, + struct obd_statfs *osfs, __u64 max_age, __u32 flags); + int (*o_statfs_async)(struct obd_export *exp, struct obd_info *oinfo, __u64 max_age, struct ptlrpc_request_set *set); int (*o_packmd)(struct obd_export *exp, struct lov_mds_md **disk_tgt, struct lov_stripe_md *mem_src); @@ -1297,20 +1351,23 @@ struct obd_ops { obd_id *ids); /* FIXME: add fid capability support for create & destroy! */ int (*o_precreate)(struct obd_export *exp); - int (*o_create)(struct obd_export *exp, struct obdo *oa, - struct lov_stripe_md **ea, struct obd_trans_info *oti); + int (*o_create)(const struct lu_env *env, struct obd_export *exp, + struct obdo *oa, struct lov_stripe_md **ea, + struct obd_trans_info *oti); int (*o_create_async)(struct obd_export *exp, struct obd_info *oinfo, struct lov_stripe_md **ea, struct obd_trans_info *oti); - int (*o_destroy)(struct obd_export *exp, struct obdo *oa, - struct lov_stripe_md *ea, struct obd_trans_info *oti, - struct obd_export *md_exp, void *capa); - int (*o_setattr)(struct obd_export *exp, struct obd_info *oinfo, - struct obd_trans_info *oti); + int (*o_destroy)(const struct lu_env *env, struct obd_export *exp, + struct obdo *oa, struct lov_stripe_md *ea, + struct obd_trans_info *oti, struct obd_export *md_exp, + void *capa); + int (*o_setattr)(const struct lu_env *, struct obd_export *exp, + struct obd_info *oinfo, struct obd_trans_info *oti); int (*o_setattr_async)(struct obd_export *exp, struct obd_info *oinfo, struct obd_trans_info *oti, struct ptlrpc_request_set *rqset); - int (*o_getattr)(struct obd_export *exp, struct obd_info *oinfo); + int (*o_getattr)(const struct lu_env *env, struct obd_export *exp, + struct obd_info *oinfo); int (*o_getattr_async)(struct obd_export *exp, struct obd_info *oinfo, struct ptlrpc_request_set *set); int (*o_brw)(int rw, struct obd_export *exp, struct obd_info *oinfo, @@ -1320,12 +1377,12 @@ struct obd_ops { struct ost_lvb *lvb, int kms_only); int (*o_adjust_kms)(struct obd_export *exp, struct lov_stripe_md *lsm, obd_off size, int shrink); - int (*o_punch)(struct obd_export *exp, struct obd_info *oinfo, - struct obd_trans_info *oti, + int (*o_punch)(const struct lu_env *, struct obd_export *exp, + struct obd_info *oinfo, struct obd_trans_info *oti, struct ptlrpc_request_set *rqset); - int (*o_sync)(struct obd_export *exp, struct obdo *oa, - struct lov_stripe_md *ea, obd_size start, obd_size end, - void *capa); + int (*o_sync)(const struct lu_env *env, struct obd_export *exp, + struct obd_info *oinfo, obd_size start, obd_size end, + struct ptlrpc_request_set *set); int (*o_migrate)(struct lustre_handle *conn, struct lov_stripe_md *dst, struct lov_stripe_md *src, obd_size start, obd_size end, struct obd_trans_info *oti); @@ -1333,15 +1390,15 @@ struct obd_ops { struct lustre_handle *srconn, struct lov_stripe_md *src, obd_size start, obd_size end, struct obd_trans_info *); int (*o_iterate)(struct lustre_handle *conn, - int (*)(obd_id, obd_gr, void *), - obd_id *startid, obd_gr group, void *data); - int (*o_preprw)(int cmd, struct obd_export *exp, struct obdo *oa, - int objcount, struct obd_ioobj *obj, - struct niobuf_remote *remote, int *nr_pages, - struct niobuf_local *local, - struct obd_trans_info *oti, - struct lustre_capa *capa); - int (*o_commitrw)(int cmd, struct obd_export *exp, struct obdo *oa, + int (*)(obd_id, obd_seq, void *), + obd_id *startid, obd_seq seq, void *data); + int (*o_preprw)(const struct lu_env *env, int cmd, + struct obd_export *exp, struct obdo *oa, int objcount, + struct obd_ioobj *obj, struct niobuf_remote *remote, + int *nr_pages, struct niobuf_local *local, + struct obd_trans_info *oti, struct lustre_capa *capa); + int (*o_commitrw)(const struct lu_env *env, int cmd, + struct obd_export *exp, struct obdo *oa, int objcount, struct obd_ioobj *obj, struct niobuf_remote *remote, int pages, struct niobuf_local *local, @@ -1379,7 +1436,7 @@ struct obd_ops { int (*o_notify)(struct obd_device *obd, struct obd_device *watched, enum obd_notify_event ev, void *data); - int (*o_health_check)(struct obd_device *); + int (*o_health_check)(const struct lu_env *env, struct obd_device *); struct obd_uuid *(*o_get_uuid) (struct obd_export *exp); /* quota methods */ @@ -1387,12 +1444,8 @@ struct obd_ops { struct obd_quotactl *); int (*o_quotactl)(struct obd_device *, struct obd_export *, struct obd_quotactl *); - int (*o_quota_adjust_qunit)(struct obd_export *exp, - struct quota_adjust_qunit *oqaq, - struct lustre_quota_ctxt *qctxt); - - int (*o_ping)(struct obd_export *exp); + int (*o_ping)(const struct lu_env *, struct obd_export *exp); /* pools methods */ int (*o_pool_new)(struct obd_device *obd, char *poolname); @@ -1464,7 +1517,7 @@ struct md_ops { int (*m_enqueue)(struct obd_export *, struct ldlm_enqueue_info *, struct lookup_intent *, struct md_op_data *, struct lustre_handle *, void *, int, - struct ptlrpc_request **, int); + struct ptlrpc_request **, __u64); int (*m_getattr)(struct obd_export *, struct md_op_data *, struct ptlrpc_request **); int (*m_getattr_name)(struct obd_export *, struct md_op_data *, @@ -1472,7 +1525,7 @@ struct md_ops { int (*m_intent_lock)(struct obd_export *, struct md_op_data *, void *, int, struct lookup_intent *, int, struct ptlrpc_request **, - ldlm_blocking_callback, int); + ldlm_blocking_callback, __u64); int (*m_link)(struct obd_export *, struct md_op_data *, struct ptlrpc_request **); int (*m_rename)(struct obd_export *, struct md_op_data *, @@ -1486,9 +1539,8 @@ struct md_ops { struct md_open_data **mod); int (*m_sync)(struct obd_export *, const struct lu_fid *, struct obd_capa *, struct ptlrpc_request **); - int (*m_readpage)(struct obd_export *, const struct lu_fid *, - struct obd_capa *, __u64, struct page *, - struct ptlrpc_request **); + int (*m_readpage)(struct obd_export *, struct md_op_data *, + struct page **, struct ptlrpc_request **); int (*m_unlink)(struct obd_export *, struct md_op_data *, struct ptlrpc_request **); @@ -1516,9 +1568,9 @@ struct md_ops { struct ptlrpc_request *); int (*m_clear_open_replay_data)(struct obd_export *, struct obd_client_handle *); - int (*m_set_lock_data)(struct obd_export *, __u64 *, void *, __u32 *); + int (*m_set_lock_data)(struct obd_export *, __u64 *, void *, __u64 *); - ldlm_mode_t (*m_lock_match)(struct obd_export *, int, + ldlm_mode_t (*m_lock_match)(struct obd_export *, __u64, const struct lu_fid *, ldlm_type_t, ldlm_policy_data_t *, ldlm_mode_t, struct lustre_handle *); @@ -1540,7 +1592,7 @@ struct md_ops { struct ldlm_enqueue_info *); int (*m_revalidate_lock)(struct obd_export *, struct lookup_intent *, - struct lu_fid *, __u32 *); + struct lu_fid *, __u64 *bits); /* * NOTE: If adding ops, add another LPROCFS_MD_OP_INIT() line to @@ -1558,7 +1610,7 @@ struct lsm_operations { void (*lsm_stripe_by_offset)(struct lov_stripe_md *, int *, obd_off *, obd_off *); int (*lsm_lmm_verify) (struct lov_mds_md *lmm, int lmm_bytes, - int *stripe_count); + __u16 *stripe_count); int (*lsm_unpackmd) (struct lov_obd *lov, struct lov_stripe_md *lsm, struct lov_mds_md *lmm); }; @@ -1584,39 +1636,6 @@ int lvfs_check_io_health(struct obd_device *obd, struct file *file); #define OBD_CALC_STRIPE_START 1 #define OBD_CALC_STRIPE_END 2 -static inline void obd_transno_commit_cb(struct obd_device *obd, __u64 transno, - struct obd_export *exp, int error) -{ - if (error) { - CERROR("%s: transno "LPU64" commit error: %d\n", - obd->obd_name, transno, error); - return; - } - if (exp && transno > exp->exp_last_committed) { - CDEBUG(D_HA, "%s: transno "LPU64" committed\n", - obd->obd_name, transno); - exp->exp_last_committed = transno; - ptlrpc_commit_replies(exp); - } else { - CDEBUG(D_INFO, "%s: transno "LPU64" committed\n", - obd->obd_name, transno); - } - if (transno > obd->obd_last_committed) - obd->obd_last_committed = transno; -} - -static inline void init_obd_quota_ops(quota_interface_t *interface, - struct obd_ops *obd_ops) -{ - if (!interface) - return; - - LASSERT(obd_ops); - obd_ops->o_quotacheck = QUOTA_OP(interface, check); - obd_ops->o_quotactl = QUOTA_OP(interface, ctl); - obd_ops->o_quota_adjust_qunit = QUOTA_OP(interface, adjust_qunit); -} - static inline struct lustre_capa *oinfo_capa(struct obd_info *oinfo) { return oinfo->oi_capa; @@ -1642,4 +1661,26 @@ static inline struct md_open_data *obd_mod_alloc(void) } \ }) +void obdo_from_inode(struct obdo *dst, struct inode *src, obd_flag valid); +void obdo_set_parent_fid(struct obdo *dst, const struct lu_fid *parent); + +/* return 1 if client should be resend request */ +static inline int client_should_resend(int resend, struct client_obd *cli) +{ + return cfs_atomic_read(&cli->cl_resends) ? + cfs_atomic_read(&cli->cl_resends) > resend : 1; +} + +/** + * Return device name for this device + * + * XXX: lu_device is declared before obd_device, while a pointer pointing + * back to obd_device in lu_device, so this helper function defines here + * instead of in lu_object.h + */ +static inline const char *lu_dev_name(const struct lu_device *lu_dev) +{ + return lu_dev->ld_obd->obd_name; +} + #endif /* __OBD_H */