X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Finclude%2Fobd.h;h=94f9d8c32aadaf88b7c8c26a6cdb0e7e2be7f657;hp=be65f34c856a92c338e011286bf41ab42c27ca69;hb=0d8c5ccc4ecfe7c0d10a0a4f92fd291320a97190;hpb=03f537c50b760b9319052f5600cbdb6ab2c2c94d diff --git a/lustre/include/obd.h b/lustre/include/obd.h index be65f34..94f9d8c 100644 --- a/lustre/include/obd.h +++ b/lustre/include/obd.h @@ -27,7 +27,7 @@ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2012, Whamcloud, Inc. + * Copyright (c) 2011, 2013, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -36,6 +36,9 @@ #ifndef __OBD_H #define __OBD_H +#ifndef __KERNEL__ +# error "userspace should not include " +#endif #if defined(__linux__) #include @@ -47,27 +50,19 @@ #error Unsupported operating system. #endif -#define IOC_OSC_TYPE 'h' -#define IOC_OSC_MIN_NR 20 -#define IOC_OSC_SET_ACTIVE _IOWR(IOC_OSC_TYPE, 21, struct obd_device *) -#define IOC_OSC_MAX_NR 50 - -#define IOC_MDC_TYPE 'i' -#define IOC_MDC_MIN_NR 20 -#define IOC_MDC_MAX_NR 50 - #include +#include +#include #ifdef HAVE_SERVER_SUPPORT # include +# include #endif #include -#include #include +#include #include #include - -#include - +#include #define MAX_OBD_DEVICES 8192 @@ -78,17 +73,15 @@ struct osc_async_rc { }; struct lov_oinfo { /* per-stripe data structure */ - 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 */ - - unsigned long loi_kms_valid:1; - __u64 loi_kms; /* known minimum size */ - struct ost_lvb loi_lvb; - struct osc_async_rc loi_ar; + 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 */ + + 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) { @@ -101,17 +94,16 @@ static inline void loi_init(struct lov_oinfo *loi) } struct lov_stripe_md { - cfs_atomic_t lsm_refc; - cfs_spinlock_t lsm_lock; - pid_t lsm_lock_owner; /* debugging */ + 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; + /* 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_seq; /* lov object seq */ + struct ost_id lw_object_oi; /* lov object id/seq */ /* LOV-private members start here -- only for use in lov/. */ __u32 lw_magic; @@ -125,8 +117,7 @@ struct lov_stripe_md { struct lov_oinfo *lsm_oinfo[0]; }; -#define lsm_object_id lsm_wire.lw_object_id -#define lsm_object_seq lsm_wire.lw_object_seq +#define lsm_oi lsm_wire.lw_object_oi #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 @@ -134,6 +125,27 @@ struct lov_stripe_md { #define lsm_stripe_count lsm_wire.lw_stripe_count #define lsm_pool_name lsm_wire.lw_pool_name +static inline bool lsm_is_released(struct lov_stripe_md *lsm) +{ + return !!(lsm->lsm_pattern & LOV_PATTERN_F_RELEASED); +} + +static inline bool lsm_has_objects(struct lov_stripe_md *lsm) +{ + if (lsm == NULL) + return false; + if (lsm_is_released(lsm)) + return false; + return true; +} + +static inline int lov_stripe_md_size(unsigned int stripe_count) +{ + struct lov_stripe_md lsm; + + return sizeof(lsm) + stripe_count * sizeof(lsm.lsm_oinfo[0]); +} + struct obd_info; typedef int (*obd_enqueue_update_f)(void *cookie, int rc); @@ -171,226 +183,45 @@ struct obd_info { char *oi_jobid; }; -/* compare all relevant fields. */ -static inline int lov_stripe_md_cmp(struct lov_stripe_md *m1, - struct lov_stripe_md *m2) -{ - /* - * ->lsm_wire contains padding, but it should be zeroed out during - * allocation. - */ - 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); - struct obd_type { - cfs_list_t typ_chain; - struct obd_ops *typ_dt_ops; - struct md_ops *typ_md_ops; - cfs_proc_dir_entry_t *typ_procroot; - char *typ_name; - int typ_refcnt; - struct lu_device_type *typ_lu; - cfs_spinlock_t obd_type_lock; + struct list_head typ_chain; + struct obd_ops *typ_dt_ops; + struct md_ops *typ_md_ops; + struct proc_dir_entry *typ_procroot; + struct proc_dir_entry *typ_procsym; + __u32 typ_sym_filter; + char *typ_name; + int typ_refcnt; + struct lu_device_type *typ_lu; + spinlock_t obd_type_lock; }; struct brw_page { - obd_off off; - cfs_page_t *pg; - int count; - obd_flag flag; -}; - -/* Individual type definitions */ - -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_rw_semaphore_t obt_rwsem; - struct vfsmount *obt_vfsmnt; - struct file *obt_health_check_filp; - struct osd_properties obt_osd_properties; - struct obd_job_stats obt_jobstats; + obd_off off; + struct page *pg; + int count; + obd_flag flag; }; /* llog contexts */ enum llog_ctxt_id { - LLOG_CONFIG_ORIG_CTXT = 0, - LLOG_CONFIG_REPL_CTXT, - LLOG_MDS_OST_ORIG_CTXT, - LLOG_MDS_OST_REPL_CTXT, - LLOG_SIZE_ORIG_CTXT, - LLOG_SIZE_REPL_CTXT, - LLOG_RD1_ORIG_CTXT, - LLOG_RD1_REPL_CTXT, - LLOG_TEST_ORIG_CTXT, - LLOG_TEST_REPL_CTXT, - LLOG_LOVEA_ORIG_CTXT, - LLOG_LOVEA_REPL_CTXT, - LLOG_CHANGELOG_ORIG_CTXT, /**< changelog generation on mdd */ - LLOG_CHANGELOG_REPL_CTXT, /**< changelog access on clients */ - LLOG_CHANGELOG_USER_ORIG_CTXT, /**< for multiple changelog consumers */ - LLOG_MAX_CTXTS -}; - -#define FILTER_SUBDIR_COUNT 32 /* set to zero for no subdirs */ - -struct filter_subdirs { - cfs_dentry_t *dentry[FILTER_SUBDIR_COUNT]; -}; - - -struct filter_ext { - __u64 fe_start; - __u64 fe_end; -}; - -struct filter_obd { - /* NB this field MUST be first */ - struct obd_device_target fo_obt; - const char *fo_fstype; - - int fo_group_count; - cfs_dentry_t *fo_dentry_O; - cfs_dentry_t **fo_dentry_O_groups; - struct filter_subdirs *fo_dentry_O_sub; - cfs_mutex_t fo_init_lock; /* group initialization lock */ - int fo_committed_group; - - cfs_spinlock_t fo_objidlock; /* protect fo_lastobjid */ - - unsigned long fo_destroys_in_progress; - cfs_mutex_t fo_create_locks[FILTER_SUBDIR_COUNT]; - - cfs_list_t fo_export_list; - int fo_subdir_count; - - obd_size fo_tot_dirty; /* protected by obd_osfs_lock */ - obd_size fo_tot_granted; /* all values in bytes */ - obd_size fo_tot_pending; - int fo_tot_granted_clients; - - obd_size fo_readcache_max_filesize; - cfs_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 */ - - struct obd_import *fo_mdc_imp; - struct obd_uuid fo_mdc_uuid; - struct lustre_handle fo_mdc_conn; - struct file **fo_last_objid_files; - __u64 *fo_last_objids; /* last created objid for groups, - * protected by fo_objidlock */ - - cfs_mutex_t 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(). - * - * 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; - cfs_spinlock_t fo_llog_list_lock; - - struct brw_stats fo_filter_stats; - - 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; - struct sptlrpc_rule_set fo_sptlrpc_rset; - - /* capability related */ - unsigned int fo_fl_oss_capa; - cfs_list_t fo_capa_keys; - cfs_hlist_head_t *fo_capa_hash; - struct llog_commit_master *fo_lcm; - int fo_sec_level; + LLOG_CONFIG_ORIG_CTXT = 0, + LLOG_CONFIG_REPL_CTXT, + LLOG_MDS_OST_ORIG_CTXT, + LLOG_MDS_OST_REPL_CTXT, + LLOG_SIZE_ORIG_CTXT, + LLOG_SIZE_REPL_CTXT, + LLOG_RD1_ORIG_CTXT, + LLOG_RD1_REPL_CTXT, + LLOG_TEST_ORIG_CTXT, + LLOG_TEST_REPL_CTXT, + LLOG_LOVEA_ORIG_CTXT, + LLOG_LOVEA_REPL_CTXT, + LLOG_CHANGELOG_ORIG_CTXT, /**< changelog generation on mdd */ + LLOG_CHANGELOG_REPL_CTXT, /**< changelog access on clients */ + LLOG_CHANGELOG_USER_ORIG_CTXT, /**< for multiple changelog consumers */ + LLOG_AGENT_ORIG_CTXT, /**< agent requests generation on cdt */ + LLOG_MAX_CTXTS }; struct timeout_item { @@ -402,12 +233,12 @@ struct timeout_item { cfs_list_t ti_chain; }; -#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 +#define OBD_MAX_RIF_DEFAULT 8 +#define OBD_MAX_RIF_MAX 512 +#define OSC_MAX_RIF_MAX 256 +#define OSC_MAX_DIRTY_DEFAULT (OBD_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 { @@ -417,29 +248,24 @@ enum { 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; - /* max_mds_easize is purely a performance thing so we don't have to - * call obd_size_diskmd() all the time. */ - int cl_default_mds_easize; - int cl_max_mds_easize; - int cl_max_mds_cookiesize; + /* max_mds_easize is purely a performance thing so we don't have to + * call obd_size_diskmd() all the time. */ + int cl_default_mds_easize; + int cl_max_mds_easize; + int cl_default_mds_cookiesize; + int cl_max_mds_cookiesize; enum lustre_sec_part cl_sp_me; enum lustre_sec_part cl_sp_to; struct sptlrpc_flavor cl_flvr_mgc; /* fixed flavor of mgc->mgs */ - //struct llog_canceld_ctxt *cl_llcd; /* it's included by obd_llog_ctxt */ - void *cl_llcd_offset; - /* the grant values are protected by loi_list_lock below */ long cl_dirty; /* all _dirty_ in bytes */ long cl_dirty_max; /* allowed w/o rpc */ @@ -458,14 +284,14 @@ struct client_obd { 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) */ + * the extent size. A chunk is max(PAGE_CACHE_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 - * lists of osc_client_pages that hang off of the loi */ + /* keep track of objects that have lois that contain pages which + * have been queued for async brw. this lock also protects the + * lists of osc_client_pages that hang off of the loi */ /* * ->cl_loi_list_lock protects consistency of * ->cl_loi_{ready,read,write}_list. ->ap_make_ready() and @@ -489,43 +315,44 @@ struct client_obd { cfs_list_t cl_loi_hp_ready_list; cfs_list_t cl_loi_write_list; cfs_list_t cl_loi_read_list; - int cl_r_in_flight; - int cl_w_in_flight; - /* just a sum of the loi/lop pending numbers to be exported by /proc */ - 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; - struct obd_histogram cl_read_page_hist; - struct obd_histogram cl_write_page_hist; - struct obd_histogram cl_read_offset_hist; - struct obd_histogram cl_write_offset_hist; + int cl_r_in_flight; + int cl_w_in_flight; + /* just a sum of the loi/lop pending numbers to be exported by /proc */ + atomic_t cl_pending_w_pages; + atomic_t cl_pending_r_pages; + __u32 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; + struct obd_histogram cl_read_page_hist; + struct obd_histogram cl_write_page_hist; + struct obd_histogram cl_read_offset_hist; + struct obd_histogram cl_write_offset_hist; /* lru for osc caching pages */ - struct cl_client_lru *cl_lru; - cfs_list_t cl_lru_osc; /* member of cl_lru->ccl_list */ - 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; + struct cl_client_cache *cl_cache; + cfs_list_t cl_lru_osc; /* member of cl_cache->ccc_lru */ + atomic_t *cl_lru_left; + atomic_t cl_lru_busy; + atomic_t cl_lru_shrinkers; + 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 */ + atomic_t cl_unstable_count; - /* 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; + /* number of in flight destroy rpcs is limited to max_rpcs_in_flight */ + atomic_t cl_destroy_in_flight; + wait_queue_head_t cl_destroy_waitq; struct mdc_rpc_lock *cl_rpc_lock; struct mdc_rpc_lock *cl_close_lock; /* mgc datastruct */ - cfs_semaphore_t cl_mgc_sem; - struct vfsmount *cl_mgc_vfsmnt; - struct dentry *cl_mgc_configs_dir; - cfs_atomic_t cl_mgc_refcount; - struct obd_export *cl_mgc_mgsexp; + struct mutex cl_mgc_mutex; + struct local_oid_storage *cl_mgc_los; + struct dt_object *cl_mgc_configs_dir; + atomic_t cl_mgc_refcount; + struct obd_export *cl_mgc_mgsexp; /* checksumming for data sent over the network */ unsigned int cl_checksum:1; /* 0 = disabled, 1 = enabled */ @@ -544,15 +371,16 @@ struct client_obd { #warning "please consider removing quotacheck compatibility code" #endif - /* sequence manager */ - struct lu_client_seq *cl_seq; + /* sequence manager */ + struct lu_client_seq *cl_seq; - cfs_atomic_t cl_resends; /* resend count */ + atomic_t cl_resends; /* resend count */ - /* ptlrpc work for writeback in ptlrpcd context */ - void *cl_writeback_work; + /* ptlrpc work for writeback in ptlrpcd context */ + void *cl_writeback_work; + void *cl_lru_work; /* hash tables for osc_quota_info */ - cfs_hash_t *cl_quota_hash[MAXQUOTAS]; + cfs_hash_t *cl_quota_hash[MAXQUOTAS]; }; #define obd2cli_tgt(obd) ((char *)(obd)->u.cli.cl_target_uuid.uuid) @@ -561,103 +389,32 @@ struct obd_id_info { obd_id *data; }; -/* */ - -struct echo_obd { - struct obd_device_target eo_obt; - struct obdo eo_oa; - cfs_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_mutex_t 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; __u64 ec_unique; }; -struct lov_qos_oss { - struct obd_uuid lqo_uuid; /* ptlrpc's c_remote_uuid */ - cfs_list_t lqo_oss_list; /* link to lov_qos */ - __u64 lqo_bavail; /* total bytes avail on OSS */ - __u64 lqo_penalty; /* current penalty */ - __u64 lqo_penalty_per_obj;/* penalty decrease every obj*/ - time_t lqo_used; /* last used time, seconds */ - __u32 lqo_ost_count; /* number of osts on this oss */ -}; - -struct ltd_qos { - struct lov_qos_oss *ltq_oss; /* oss info */ - __u64 ltq_penalty; /* current penalty */ - __u64 ltq_penalty_per_obj; /* penalty decrease every obj*/ - __u64 ltq_weight; /* net weighting */ - time_t ltq_used; /* last used time, seconds */ - unsigned int ltq_usable:1; /* usable for striping */ -}; - /* Generic subset of OSTs */ struct ost_pool { __u32 *op_array; /* array of index of 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 */ -}; - -/* Round-robin allocator data */ -struct lov_qos_rr { - __u32 lqr_start_idx; /* start index of new inode */ - __u32 lqr_offset_idx; /* aliasing for start_idx */ - int lqr_start_count; /* reseed counter */ - struct ost_pool lqr_pool; /* round-robin optimized list */ - unsigned long lqr_dirty:1; /* recalc round-robin list */ + struct rw_semaphore op_rw_sem; /* to protect ost_pool use */ }; /* 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; -}; -/* 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; - __u32 lq_active_oss_count; - unsigned int lq_prio_free; /* priority for free space */ - unsigned int lq_threshold_rr;/* priority for rr */ - struct lov_qos_rr lq_rr; /* round robin qos data */ - unsigned long lq_dirty:1, /* recalc qos data */ - lq_same_space:1,/* the ost's all have approx. - the same space avail */ - lq_reset:1, /* zero current penalties */ - lq_statfs_in_progress:1; /* statfs op in - progress */ - /* qos statfs data */ - struct lov_statfs_data *lq_statfs_data; - cfs_waitq_t lq_statfs_waitq; /* waitqueue to notify statfs - * requests completion */ -}; - struct lov_tgt_desc { cfs_list_t ltd_kill; struct obd_uuid ltd_uuid; struct obd_device *ltd_obd; struct obd_export *ltd_exp; - struct ltd_qos ltd_qos; /* qos info per target */ __u32 ltd_gen; __u32 ltd_index; /* index in lov_obd->tgts */ unsigned long ltd_active:1,/* is this target up for requests */ @@ -665,53 +422,36 @@ struct lov_tgt_desc { ltd_reap:1; /* should this target be deleted */ }; -/* Pool metadata */ -#define pool_tgt_size(_p) _p->pool_obds.op_size -#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 - -struct pool_desc { - char pool_name[LOV_MAXPOOLNAME + 1]; /* name of pool */ - struct ost_pool pool_obds; /* pool members */ - cfs_atomic_t pool_refcount; /* pool ref. counter */ - struct lov_qos_rr pool_rr; /* round robin qos */ - 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 obd_device *pool_lobd; /* obd of the lov/lod to which - * this pool belongs */ -}; - struct lov_obd { - struct lov_desc desc; - struct lov_tgt_desc **lov_tgts; /* sparse array */ - struct ost_pool lov_packed; /* all OSTs in a packed - array */ - cfs_mutex_t lov_lock; - struct obd_connect_data lov_ocd; - cfs_atomic_t lov_refcount; - __u32 lov_tgt_count; /* how many OBD's */ - __u32 lov_active_tgt_count; /* how many active */ - __u32 lov_death_row;/* tgts scheduled to be deleted */ - __u32 lov_tgt_size; /* size of tgts array */ - int lov_connects; - int lov_pool_count; - cfs_hash_t *lov_pools_hash_body; /* used for key access */ - 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 data from upper layer */ - void *lov_lru; + struct lov_desc desc; + struct lov_tgt_desc **lov_tgts; /* sparse array */ + struct ost_pool lov_packed; /* all OSTs in a packed + array */ + struct mutex lov_lock; + struct obd_connect_data lov_ocd; + struct proc_dir_entry *targets_proc_entry; + atomic_t lov_refcount; + __u32 lov_death_row; /* tgts scheduled to be deleted */ + __u32 lov_tgt_size; /* size of tgts array */ + int lov_connects; + int lov_pool_count; + cfs_hash_t *lov_pools_hash_body; /* used for key access */ + struct list_head lov_pool_list; /* used for sequential access */ + struct proc_dir_entry *lov_pool_proc_entry; + enum lustre_sec_part lov_sp_me; + + /* Cached LRU and unstable data from upper layer */ + void *lov_cache; + + struct rw_semaphore lov_notify_lock; }; struct lmv_tgt_desc { - struct obd_uuid ltd_uuid; - struct obd_export *ltd_exp; - int ltd_active; /* is this target up for requests */ - int ltd_idx; - cfs_mutex_t ltd_fid_mutex; + struct obd_uuid ltd_uuid; + struct obd_export *ltd_exp; + int ltd_idx; + struct mutex ltd_fid_mutex; + unsigned long ltd_active:1; /* target up for requests */ }; enum placement_policy { @@ -724,54 +464,52 @@ 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; - placement_policy_t lmv_placement; - struct lmv_desc desc; - struct obd_uuid cluuid; - struct obd_export *exp; - - int connected; - int max_easize; - int max_def_easize; - int max_cookiesize; - int server_timeout; - cfs_mutex_t init_mutex; - - struct lmv_tgt_desc *tgts; - int tgts_size; - - struct obd_connect_data *datas; - int datas_size; - - struct obd_connect_data conn_data; + 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; + struct obd_export *exp; + struct proc_dir_entry *targets_proc_entry; + + struct mutex init_mutex; + int connected; + int max_easize; + int max_def_easize; + int max_cookiesize; + int max_def_cookiesize; + int server_timeout; + + int tgts_size; /* size of tgts array */ + struct lmv_tgt_desc **tgts; + + struct obd_connect_data conn_data; }; struct niobuf_local { - __u64 lnb_file_offset; - __u32 lnb_page_offset; - __u32 len; - __u32 flags; - cfs_page_t *page; - cfs_dentry_t *dentry; - int lnb_grant_used; - int rc; + __u64 lnb_file_offset; + __u32 lnb_page_offset; + __u32 len; + __u32 flags; + struct page *page; + struct dentry *dentry; + int lnb_grant_used; + int rc; }; #define LUSTRE_FLD_NAME "fld" #define LUSTRE_SEQ_NAME "seq" -#define LUSTRE_CMM_NAME "cmm" #define LUSTRE_MDD_NAME "mdd" #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" +#define LUSTRE_LWP_NAME "lwp" /* obd device type names */ /* FIXME all the references to LUSTRE_MDS_NAME should be swapped with LUSTRE_MDT_NAME */ @@ -785,7 +523,6 @@ struct niobuf_local { #define LUSTRE_MGS_NAME "mgs" #define LUSTRE_MGC_NAME "mgc" -#define LUSTRE_CACHEOBD_NAME "cobd" #define LUSTRE_ECHO_NAME "obdecho" #define LUSTRE_ECHO_CLIENT_NAME "echo_client" #define LUSTRE_QMT_NAME "qmt" @@ -796,10 +533,7 @@ struct niobuf_local { #define LUSTRE_MGS_OBDNAME "MGS" #define LUSTRE_MGC_OBDNAME "MGC" -/* 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) +static inline int is_osp_on_mdt(char *name) { char *ptr; @@ -809,23 +543,28 @@ static inline int is_osp_on_ost(char *name) return 0; } - if (strncmp(ptr + 1, "OST", 3) != 0 && strncmp(ptr + 1, "MDT", 3) != 0) - return 0; + /* 1.8 OSC/OSP name on MDT is fsname-OSTxxxx-osc */ + if (strncmp(ptr + 1, "osc", 3) == 0) + return 1; - /* match the "-osp" */ - if (ptr - name < strlen(LUSTRE_OSP_NAME) + 1) + if (strncmp(ptr + 1, "MDT", 3) != 0) return 0; - ptr -= (strlen(LUSTRE_OSP_NAME) + 1); - if (*ptr != '-') + while (*(--ptr) != '-' && ptr != name); + + if (ptr == name) return 0; - if (strncmp(ptr + 1, LUSTRE_OSP_NAME, strlen(LUSTRE_OSP_NAME)) != 0) + if (strncmp(ptr + 1, LUSTRE_OSP_NAME, strlen(LUSTRE_OSP_NAME)) != 0 && + strncmp(ptr + 1, LUSTRE_OSC_NAME, strlen(LUSTRE_OSC_NAME)) != 0) return 0; return 1; } +/* Don't conflict with on-wire flags OBD_BRW_WRITE, etc */ +#define N_LOCAL_TEMP_PAGE 0x10000000 + struct obd_trans_info { __u64 oti_transno; __u64 oti_xid; @@ -839,7 +578,7 @@ struct obd_trans_info { struct llog_cookie *oti_logcookies; int oti_numcookies; /** synchronous write is needed */ - long oti_sync_write:1; + unsigned long oti_sync_write:1; /* initial thread handling transaction */ struct ptlrpc_thread * oti_thread; @@ -952,88 +691,56 @@ 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; + svc_handler_t trd_recovery_handler; + pid_t trd_processing_task; + struct completion trd_starting; + struct completion trd_finishing; }; -/** - * In HEAD for CMD, the object is created in group number which is 3>= - * or indexing starts from 3. To test this assertions are added to disallow - * group 0. But to run 2.0 mds server on 1.8.x disk format (i.e. interop_mode) - * object in group 0 needs to be allowed. - * So for interop mode following changes needs to be done: - * 1. No need to assert on group 0 or allow group 0 - * 2. The group number indexing starts from 0 instead of 3 - */ - -#define LASSERT_SEQ_IS_MDT(seq) LASSERT(fid_seq_is_mdt(seq)) - -static inline __u64 objseq_to_mdsno(obd_seq seq) -{ - LASSERT_SEQ_IS_MDT(seq); - if (seq == FID_SEQ_OST_MDT0) - return 0; - return seq - FID_SEQ_OST_MDT1 + 1; -} - -static inline int mdt_to_obd_objseq(int mdtid) -{ - /** - * MDS0 uses seq 0 pre FID-on-OST, other MDSes will use seq from - * FID_SEQ_OST_MDT1 - */ - if (mdtid) - return FID_SEQ_OST_MDT1 + mdtid - 1; - return 0; -} - struct obd_llog_group { - cfs_list_t olg_list; - 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_mutex_t olg_cat_processing; + int olg_seq; + struct llog_ctxt *olg_ctxts[LLOG_MAX_CTXTS]; + wait_queue_head_t olg_waitq; + spinlock_t olg_lock; + struct mutex olg_cat_processing; }; /* corresponds to one of the obd's */ #define OBD_DEVICE_MAGIC 0XAB5CD6EF -#define OBD_DEV_BY_DEVNAME 0xffffd0de struct obd_device { - struct obd_type *obd_type; - __u32 obd_magic; + struct obd_type *obd_type; + __u32 obd_magic; /* common and UUID name of this device */ - char obd_name[MAX_OBD_NAME]; - struct obd_uuid obd_uuid; - - 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_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_stopping:1, /* started cleanup */ - obd_starting:1, /* started setup */ - obd_force:1, /* cleanup with > 0 obd refcount */ - obd_fail:1, /* cleanup with failover */ - obd_async_recov:1, /* allow asynchronous orphan cleanup */ - 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 */ + char obd_name[MAX_OBD_NAME]; + struct obd_uuid obd_uuid; + int obd_minor; + struct lu_device *obd_lu_dev; + + /* 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_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_stopping:1, /* started cleanup */ + obd_starting:1, /* started setup */ + obd_force:1, /* cleanup with > 0 obd refcount */ + obd_fail:1, /* cleanup with failover */ + obd_async_recov:1, /* allow asynchronous orphan cleanup */ + 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 */ + obd_uses_nid_stats:1; /* maintain per-client OBD stats */ + /* 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; @@ -1041,95 +748,97 @@ struct obd_device { cfs_hash_t *obd_uuid_hash; /* nid-export hash body */ cfs_hash_t *obd_nid_hash; - /* nid stats body */ - cfs_hash_t *obd_nid_stats_hash; - cfs_list_t obd_nid_stats; - cfs_atomic_t obd_refcount; - cfs_waitq_t obd_refcount_waitq; - cfs_list_t obd_exports; - 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; /* protects obd bitfield above */ - cfs_mutex_t obd_dev_mutex; - __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; + /* nid stats body */ + cfs_hash_t *obd_nid_stats_hash; + cfs_list_t obd_nid_stats; + atomic_t obd_refcount; + cfs_list_t obd_exports; + cfs_list_t obd_unlinked_exports; + cfs_list_t obd_delayed_exports; + struct list_head obd_lwp_list; + int obd_num_exports; + 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; + 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; + struct obd_export *obd_lwp_export; /* list of exports in LRU order, for ping evictor, with obd_dev_lock */ cfs_list_t obd_exports_timed; time_t obd_eviction_timer; /* for ping evictor */ - int obd_max_recoverable_clients; - cfs_atomic_t obd_connected_clients; - int obd_stale_clients; - int obd_delayed_clients; + int obd_max_recoverable_clients; + atomic_t obd_connected_clients; + int obd_stale_clients; + int obd_delayed_clients; /* this lock protects all recovery list_heads, timer and * obd_next_recovery_transno value */ - cfs_spinlock_t obd_recovery_task_lock; + 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 */ - 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; - int obd_recovery_stage; - - union { - struct obd_device_target obt; - struct filter_obd filter; - 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; - } u; - /* Fields used by LProcFS */ - unsigned int obd_cntr_base; - struct lprocfs_stats *obd_stats; - - unsigned int md_cntr_base; - struct lprocfs_stats *md_stats; - - cfs_proc_dir_entry_t *obd_proc_entry; - cfs_proc_dir_entry_t *obd_proc_exports_entry; - cfs_proc_dir_entry_t *obd_svc_procroot; - struct lprocfs_stats *obd_svc_stats; - cfs_atomic_t obd_evict_inprogress; - cfs_waitq_t obd_evict_inprogress_waitq; - cfs_list_t obd_evict_list; /* protected with pet_lock */ + int obd_replayed_requests; + int obd_requests_queued_for_recovery; + wait_queue_head_t obd_next_transno_waitq; + /* protected by obd_recovery_task_lock */ + struct timer_list obd_recovery_timer; + time_t obd_recovery_start; /* seconds */ + time_t obd_recovery_end; /* seconds, for lprocfs_status */ + 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; + atomic_t obd_req_replay_clients; + 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; + + union { +#ifdef HAVE_SERVER_SUPPORT + struct obd_device_target obt; + struct filter_obd filter; + struct ost_obd ost; + struct echo_obd echo; +#endif + struct client_obd cli; + struct echo_client_obd echo_client; + struct lov_obd lov; + struct lmv_obd lmv; + } u; + /* Fields used by LProcFS */ + unsigned int obd_cntr_base; + struct lprocfs_stats *obd_stats; + + unsigned int obd_md_cntr_base; + struct lprocfs_stats *obd_md_stats; + + struct proc_dir_entry *obd_proc_entry; + struct proc_dir_entry *obd_proc_exports_entry; + struct proc_dir_entry *obd_svc_procroot; + struct lprocfs_stats *obd_svc_stats; + struct lprocfs_seq_vars *obd_vars; + atomic_t obd_evict_inprogress; + wait_queue_head_t obd_evict_inprogress_waitq; + cfs_list_t obd_evict_list; /* protected with pet_lock */ /** * 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; @@ -1171,10 +880,14 @@ enum obd_cleanup_stage { #define KEY_INIT_RECOV "initial_recov" #define KEY_INTERMDS "inter_mds" #define KEY_LAST_ID "last_id" +#define KEY_LAST_FID "last_fid" #define KEY_LOCK_TO_STRIPE "lock_to_stripe" #define KEY_LOVDESC "lovdesc" #define KEY_LOV_IDX "lov_idx" -#define KEY_MAX_EASIZE "max_easize" +#define KEY_MAX_EASIZE "max_easize" +#define KEY_DEFAULT_EASIZE "default_easize" +#define KEY_MAX_COOKIESIZE "max_cookiesize" +#define KEY_DEFAULT_COOKIESIZE "default_cookiesize" #define KEY_MDS_CONN "mds_conn" #define KEY_MGSSEC "mgssec" #define KEY_NEXT_ID "next_id" @@ -1187,8 +900,10 @@ enum obd_cleanup_stage { #define KEY_CONNECT_FLAG "connect_flags" #define KEY_SYNC_LOCK_CANCEL "sync_lock_cancel" -#define KEY_LRU_SET "lru_set" -#define KEY_LRU_SHRINK "lru_shrink" +#define KEY_CACHE_SET "cache_set" +#define KEY_CACHE_LRU_SHRINK "cache_lru_shrink" +#define KEY_CHANGELOG_INDEX "changelog_index" +#define KEY_OSP_CONNECTED "osp_connected" struct lu_context; @@ -1206,18 +921,25 @@ struct lu_context; #define IT_LAYOUT (1 << 10) #define IT_QUOTA_DQACQ (1 << 11) #define IT_QUOTA_CONN (1 << 12) +#define IT_SETXATTR (1 << 13) 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 | - IT_LAYOUT)) - return LCK_CR; - - LASSERTF(0, "Invalid it_op: %d\n", it->it_op); - return -EINVAL; + /* 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_GETATTR | IT_OPEN | IT_LOOKUP | + IT_LAYOUT)) + return LCK_CR; + else if (it->it_op & IT_READDIR) + return LCK_PR; + else if (it->it_op & IT_GETXATTR) + return LCK_PR; + else if (it->it_op & IT_SETXATTR) + return LCK_PW; + + LASSERTF(0, "Invalid it_op: %d\n", it->it_op); + return -EINVAL; } struct md_op_data { @@ -1238,32 +960,53 @@ struct md_op_data { __u32 op_fsgid; cfs_cap_t op_cap; void *op_data; + size_t op_data_size; /* iattr fields and blocks. */ struct iattr op_attr; #ifdef __KERNEL__ unsigned int op_attr_flags; #endif - __u64 op_valid; - loff_t op_attr_blocks; + __u64 op_valid; + loff_t op_attr_blocks; + + /* Size-on-MDS epoch and flags. */ + __u64 op_ioepoch; + __u32 op_flags; + + /* Capa fields */ + struct obd_capa *op_capa1; + struct obd_capa *op_capa2; + + /* Various operation flags. */ + enum mds_op_bias op_bias; - /* Size-on-MDS epoch and flags. */ - __u64 op_ioepoch; - __u32 op_flags; + /* Used by readdir */ + __u64 op_hash_offset; - /* Capa fields */ - struct obd_capa *op_capa1; - struct obd_capa *op_capa2; + /* The offset of dir entry among the same hash entries, which + * is used to resolve the hash conflict. */ + __u32 op_same_hash_offset; - /* Various operation flags. */ - __u32 op_bias; + /* Used by readdir */ + __u32 op_npages; - /* Operation type */ - __u32 op_opc; + /* used to transfer info between the stacks of MD client + * see enum op_cli_flags */ + __u32 op_cli_flags; - /* Used by readdir */ - __u32 op_npages; - __u64 op_offset; + /* File object data version for HSM release, on client */ + __u64 op_data_version; + struct lustre_handle op_lease_handle; +}; + +#define op_stripe_offset op_ioepoch +#define op_max_pages op_valid + +struct md_callback { + int (*md_blocking_ast)(struct ldlm_lock *lock, + struct ldlm_lock_desc *desc, + void *data, int flag); }; struct md_enqueue_info; @@ -1283,9 +1026,9 @@ struct md_enqueue_info { }; struct obd_ops { - cfs_module_t *o_owner; - int (*o_iocontrol)(unsigned int cmd, struct obd_export *exp, int len, - void *karg, void *uarg); + struct module *o_owner; + int (*o_iocontrol)(unsigned int cmd, struct obd_export *exp, int len, + void *karg, void *uarg); 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); @@ -1320,20 +1063,19 @@ struct obd_ops { void *localdata); int (*o_disconnect)(struct obd_export *exp); - /* Initialize/finalize fids infrastructure. */ - int (*o_fid_init)(struct obd_export *exp); - int (*o_fid_fini)(struct obd_export *exp); + /* Initialize/finalize fids infrastructure. */ + int (*o_fid_init)(struct obd_device *obd, + struct obd_export *exp, enum lu_cli_type type); + int (*o_fid_fini)(struct obd_device *obd); - /* Allocate new fid according to passed @hint. */ - int (*o_fid_alloc)(struct obd_export *exp, struct lu_fid *fid, - struct md_op_data *op_data); + /* Allocate new fid according to passed @hint. */ + int (*o_fid_alloc)(const struct lu_env *env, struct obd_export *exp, + struct lu_fid *fid, struct md_op_data *op_data); /* * 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)(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, @@ -1342,10 +1084,6 @@ struct obd_ops { struct lov_stripe_md *mem_src); int (*o_unpackmd)(struct obd_export *exp,struct lov_stripe_md **mem_tgt, struct lov_mds_md *disk_src, int disk_len); - int (*o_preallocate)(struct lustre_handle *, obd_count *req, - obd_id *ids); - /* FIXME: add fid capability support for create & destroy! */ - int (*o_precreate)(struct obd_export *exp); 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); @@ -1365,28 +1103,6 @@ struct obd_ops { 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, - obd_count oa_bufs, struct brw_page *pgarr, - struct obd_trans_info *oti); - int (*o_merge_lvb)(struct obd_export *exp, struct lov_stripe_md *lsm, - 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)(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)(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); - int (*o_copy)(struct lustre_handle *dstconn, struct lov_stripe_md *dst, - 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_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, @@ -1398,21 +1114,12 @@ struct obd_ops { struct niobuf_remote *remote, int pages, struct niobuf_local *local, struct obd_trans_info *oti, int rc); - int (*o_enqueue)(struct obd_export *, struct obd_info *oinfo, - struct ldlm_enqueue_info *einfo, - struct ptlrpc_request_set *rqset); int (*o_change_cbdata)(struct obd_export *, struct lov_stripe_md *, ldlm_iterator_t it, void *data); int (*o_find_cbdata)(struct obd_export *, struct lov_stripe_md *, ldlm_iterator_t it, void *data); - int (*o_cancel)(struct obd_export *, struct lov_stripe_md *md, - __u32 mode, struct lustre_handle *); - int (*o_cancel_unused)(struct obd_export *, struct lov_stripe_md *, - ldlm_cancel_flags_t flags, void *opaque); int (*o_init_export)(struct obd_export *exp); int (*o_destroy_export)(struct obd_export *exp); - int (*o_extent_calc)(struct obd_export *, struct lov_stripe_md *, - int cmd, obd_off *); /* llog related obd_methods */ int (*o_llog_init)(struct obd_device *obd, struct obd_llog_group *grp, @@ -1457,130 +1164,97 @@ struct obd_ops { * Also, add a wrapper function in include/linux/obd_class.h. */ }; -enum { - LUSTRE_OPC_MKDIR = (1 << 0), - LUSTRE_OPC_SYMLINK = (1 << 1), - LUSTRE_OPC_MKNOD = (1 << 2), - LUSTRE_OPC_CREATE = (1 << 3), - LUSTRE_OPC_ANY = (1 << 4) -}; - /* lmv structures */ -#define MEA_MAGIC_LAST_CHAR 0xb2221ca1 -#define MEA_MAGIC_ALL_CHARS 0xb222a11c -#define MEA_MAGIC_HASH_SEGMENT 0xb222a11b - -#define MAX_HASH_SIZE_32 0x7fffffffUL -#define MAX_HASH_SIZE 0x7fffffffffffffffULL -#define MAX_HASH_HIGHEST_BIT 0x1000000000000000ULL - struct lustre_md { - struct mdt_body *body; - struct lov_stripe_md *lsm; - struct lmv_stripe_md *mea; + struct mdt_body *body; + struct lov_stripe_md *lsm; + struct lmv_stripe_md *lmv; #ifdef CONFIG_FS_POSIX_ACL - struct posix_acl *posix_acl; + struct posix_acl *posix_acl; #endif - struct mdt_remote_perm *remote_perm; - struct obd_capa *mds_capa; - struct obd_capa *oss_capa; + struct mdt_remote_perm *remote_perm; + struct obd_capa *mds_capa; + struct obd_capa *oss_capa; + __u64 lm_flags; }; struct md_open_data { - struct obd_client_handle *mod_och; - struct ptlrpc_request *mod_open_req; - struct ptlrpc_request *mod_close_req; - cfs_atomic_t mod_refcount; + struct obd_client_handle *mod_och; + struct ptlrpc_request *mod_open_req; + struct ptlrpc_request *mod_close_req; + atomic_t mod_refcount; + bool mod_is_create; }; struct lookup_intent; +struct cl_attr; struct md_ops { - int (*m_getstatus)(struct obd_export *, struct lu_fid *, - struct obd_capa **); - int (*m_change_cbdata)(struct obd_export *, const struct lu_fid *, - ldlm_iterator_t, void *); - int (*m_find_cbdata)(struct obd_export *, const struct lu_fid *, - ldlm_iterator_t, void *); - int (*m_close)(struct obd_export *, struct md_op_data *, - struct md_open_data *, struct ptlrpc_request **); - int (*m_create)(struct obd_export *, struct md_op_data *, - const void *, int, int, __u32, __u32, cfs_cap_t, - __u64, struct ptlrpc_request **); - int (*m_done_writing)(struct obd_export *, struct md_op_data *, - struct md_open_data *); - 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 **, __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 *, - struct ptlrpc_request **); - int (*m_intent_lock)(struct obd_export *, struct md_op_data *, - void *, int, struct lookup_intent *, int, - struct ptlrpc_request **, + /* Every operation from MD_STATS_FIRST_OP up to and including + * MD_STATS_LAST_OP will be counted by EXP_MD_OP_INCREMENT() + * and will appear in /proc/fs/lustre/{lmv,mdc}/.../md_stats. + * Operations after MD_STATS_LAST_OP are excluded from stats. + * There are a few reasons for doing this: we prune the 17 + * counters which will be of minimal use in understanding + * metadata utilization, we save memory by allocating 15 + * instead of 32 counters, we save cycles by not counting. + * + * MD_STATS_FIRST_OP must be the first member of md_ops. + */ +#define MD_STATS_FIRST_OP m_close + int (*m_close)(struct obd_export *, struct md_op_data *, + struct md_open_data *, struct ptlrpc_request **); + + int (*m_create)(struct obd_export *, struct md_op_data *, + const void *, int, int, __u32, __u32, cfs_cap_t, + __u64, struct ptlrpc_request **); + + int (*m_enqueue)(struct obd_export *, struct ldlm_enqueue_info *, + const union ldlm_policy_data *, + struct lookup_intent *, struct md_op_data *, + struct lustre_handle *, __u64); + + int (*m_getattr)(struct obd_export *, struct md_op_data *, + struct ptlrpc_request **); + + int (*m_intent_lock)(struct obd_export *, struct md_op_data *, + struct lookup_intent *, + struct ptlrpc_request **, 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 *, - const char *, int, const char *, int, - struct ptlrpc_request **); - int (*m_is_subdir)(struct obd_export *, const struct lu_fid *, - const struct lu_fid *, - struct ptlrpc_request **); - int (*m_setattr)(struct obd_export *, struct md_op_data *, void *, - int , void *, int, struct ptlrpc_request **, - 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 *, struct md_op_data *, - struct page **, struct ptlrpc_request **); - - int (*m_unlink)(struct obd_export *, struct md_op_data *, - struct ptlrpc_request **); - - int (*m_setxattr)(struct obd_export *, const struct lu_fid *, - struct obd_capa *, obd_valid, const char *, - const char *, int, int, int, __u32, - struct ptlrpc_request **); - - int (*m_getxattr)(struct obd_export *, const struct lu_fid *, - struct obd_capa *, obd_valid, const char *, - const char *, int, int, int, - struct ptlrpc_request **); - - int (*m_init_ea_size)(struct obd_export *, int, int, int); - - int (*m_get_lustre_md)(struct obd_export *, struct ptlrpc_request *, - struct obd_export *, struct obd_export *, - struct lustre_md *); - - int (*m_free_lustre_md)(struct obd_export *, struct lustre_md *); - - int (*m_set_open_replay_data)(struct obd_export *, - struct obd_client_handle *, - 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 *, __u64 *); - 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 *); - - int (*m_cancel_unused)(struct obd_export *, const struct lu_fid *, - ldlm_policy_data_t *, ldlm_mode_t, - ldlm_cancel_flags_t flags, void *opaque); - int (*m_renew_capa)(struct obd_export *, struct obd_capa *oc, - renew_capa_cb_t cb); - int (*m_unpack_capa)(struct obd_export *, struct ptlrpc_request *, - const struct req_msg_field *, struct obd_capa **); - - int (*m_get_remote_perm)(struct obd_export *, const struct lu_fid *, - struct obd_capa *, __u32, - struct ptlrpc_request **); + int (*m_link)(struct obd_export *, struct md_op_data *, + struct ptlrpc_request **); + + int (*m_rename)(struct obd_export *, struct md_op_data *, + const char *, int, const char *, int, + struct ptlrpc_request **); + + int (*m_setattr)(struct obd_export *, struct md_op_data *, void *, + int , void *, int, struct ptlrpc_request **, + struct md_open_data **mod); + + int (*m_fsync)(struct obd_export *, const struct lu_fid *, + struct obd_capa *, struct ptlrpc_request **); + + int (*m_readpage)(struct obd_export *, struct md_op_data *, + struct page **, struct ptlrpc_request **); + + int (*m_read_entry)(struct obd_export *, struct md_op_data *, + struct md_callback *cb_op, struct lu_dirent **ld, + struct page **ppage); + + int (*m_unlink)(struct obd_export *, struct md_op_data *, + struct ptlrpc_request **); + + int (*m_setxattr)(struct obd_export *, const struct lu_fid *, + struct obd_capa *, obd_valid, const char *, + const char *, int, int, int, __u32, + struct ptlrpc_request **); + + int (*m_getxattr)(struct obd_export *, const struct lu_fid *, + struct obd_capa *, obd_valid, const char *, + const char *, int, int, int, + struct ptlrpc_request **); int (*m_intent_getattr_async)(struct obd_export *, struct md_enqueue_info *, @@ -1588,12 +1262,68 @@ struct md_ops { int (*m_revalidate_lock)(struct obd_export *, struct lookup_intent *, struct lu_fid *, __u64 *bits); +#define MD_STATS_LAST_OP m_revalidate_lock - /* - * NOTE: If adding ops, add another LPROCFS_MD_OP_INIT() line to - * lprocfs_alloc_md_stats() in obdclass/lprocfs_status.c. Also, add a - * wrapper function in include/linux/obd_class.h. - */ + int (*m_getstatus)(struct obd_export *, struct lu_fid *, + struct obd_capa **); + + int (*m_null_inode)(struct obd_export *, const struct lu_fid *); + + int (*m_find_cbdata)(struct obd_export *, const struct lu_fid *, + ldlm_iterator_t, void *); + + int (*m_done_writing)(struct obd_export *, struct md_op_data *, + struct md_open_data *); + + int (*m_getattr_name)(struct obd_export *, struct md_op_data *, + struct ptlrpc_request **); + + int (*m_is_subdir)(struct obd_export *, const struct lu_fid *, + const struct lu_fid *, + struct ptlrpc_request **); + + int (*m_init_ea_size)(struct obd_export *, int, int, int, int); + + int (*m_get_lustre_md)(struct obd_export *, struct ptlrpc_request *, + struct obd_export *, struct obd_export *, + struct lustre_md *); + + int (*m_free_lustre_md)(struct obd_export *, struct lustre_md *); + + int (*m_merge_attr)(struct obd_export *, + const struct lmv_stripe_md *lsm, + struct cl_attr *attr); + + int (*m_update_lsm_md)(struct obd_export *, struct lmv_stripe_md *lsm, + struct mdt_body *, ldlm_blocking_callback); + + int (*m_set_open_replay_data)(struct obd_export *, + struct obd_client_handle *, + struct lookup_intent *); + + int (*m_clear_open_replay_data)(struct obd_export *, + struct obd_client_handle *); + + int (*m_set_lock_data)(struct obd_export *, __u64 *, void *, __u64 *); + + 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 *); + + int (*m_cancel_unused)(struct obd_export *, const struct lu_fid *, + ldlm_policy_data_t *, ldlm_mode_t, + ldlm_cancel_flags_t flags, void *opaque); + + int (*m_renew_capa)(struct obd_export *, struct obd_capa *oc, + renew_capa_cb_t cb); + + int (*m_unpack_capa)(struct obd_export *, struct ptlrpc_request *, + const struct req_msg_field *, struct obd_capa **); + + int (*m_get_remote_perm)(struct obd_export *, const struct lu_fid *, + struct obd_capa *, __u32, + struct ptlrpc_request **); }; struct lsm_operations { @@ -1625,8 +1355,6 @@ static inline const struct lsm_operations *lsm_op_find(int magic) } } -int lvfs_check_io_health(struct obd_device *obd, struct file *file); - /* Requests for obd_extent_calc() */ #define OBD_CALC_STRIPE_START 1 #define OBD_CALC_STRIPE_END 2 @@ -1638,22 +1366,22 @@ static inline struct lustre_capa *oinfo_capa(struct obd_info *oinfo) static inline struct md_open_data *obd_mod_alloc(void) { - struct md_open_data *mod; - OBD_ALLOC_PTR(mod); - if (mod == NULL) - return NULL; - cfs_atomic_set(&mod->mod_refcount, 1); - return mod; + struct md_open_data *mod; + OBD_ALLOC_PTR(mod); + if (mod == NULL) + return NULL; + atomic_set(&mod->mod_refcount, 1); + return mod; } -#define obd_mod_get(mod) cfs_atomic_inc(&(mod)->mod_refcount) -#define obd_mod_put(mod) \ -({ \ - if (cfs_atomic_dec_and_test(&(mod)->mod_refcount)) { \ - if ((mod)->mod_open_req) \ - ptlrpc_req_finished((mod)->mod_open_req); \ - OBD_FREE_PTR(mod); \ - } \ +#define obd_mod_get(mod) atomic_inc(&(mod)->mod_refcount) +#define obd_mod_put(mod) \ +({ \ + if (atomic_dec_and_test(&(mod)->mod_refcount)) { \ + if ((mod)->mod_open_req) \ + ptlrpc_req_finished((mod)->mod_open_req); \ + OBD_FREE_PTR(mod); \ + } \ }) void obdo_from_inode(struct obdo *dst, struct inode *src, obd_flag valid); @@ -1662,8 +1390,8 @@ 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 atomic_read(&cli->cl_resends) ? + atomic_read(&cli->cl_resends) > resend : 1; } /** @@ -1678,4 +1406,64 @@ static inline const char *lu_dev_name(const struct lu_device *lu_dev) return lu_dev->ld_obd->obd_name; } +static inline bool filename_is_volatile(const char *name, int namelen, int *idx) +{ + const char *start; + char *end; + + if (strncmp(name, LUSTRE_VOLATILE_HDR, LUSTRE_VOLATILE_HDR_LEN) != 0) + return false; + + /* caller does not care of idx */ + if (idx == NULL) + return true; + + /* volatile file, the MDT can be set from name */ + /* name format is LUSTRE_VOLATILE_HDR:[idx]: */ + /* if no MDT is specified, use std way */ + if (namelen < LUSTRE_VOLATILE_HDR_LEN + 2) + goto bad_format; + /* test for no MDT idx case */ + if ((*(name + LUSTRE_VOLATILE_HDR_LEN) == ':') && + (*(name + LUSTRE_VOLATILE_HDR_LEN + 1) == ':')) { + *idx = -1; + return true; + } + /* we have an idx, read it */ + start = name + LUSTRE_VOLATILE_HDR_LEN + 1; + *idx = strtoul(start, &end, 16); + /* error cases: + * no digit, no trailing :, negative value + */ + if (((*idx == 0) && (end == start)) || + (*end != ':') || (*idx < 0)) + goto bad_format; + + return true; +bad_format: + /* bad format of mdt idx, we cannot return an error + * to caller so we use hash algo */ + CERROR("Bad volatile file name format: %s\n", + name + LUSTRE_VOLATILE_HDR_LEN); + return false; +} + +static inline int cli_brw_size(struct obd_device *obd) +{ + LASSERT(obd != NULL); + return obd->u.cli.cl_max_pages_per_rpc << PAGE_CACHE_SHIFT; +} + +static inline void client_adjust_max_dirty(struct client_obd *cli) +{ + /* initializing */ + if (cli->cl_dirty_max <= 0) + cli->cl_dirty_max = OSC_MAX_DIRTY_DEFAULT * 1024 * 1024; + else + cli->cl_dirty_max = cli->cl_max_rpcs_in_flight * + (cli->cl_max_pages_per_rpc << PAGE_CACHE_SHIFT); + if (cli->cl_dirty_max >> PAGE_CACHE_SHIFT > totalram_pages / 8) + cli->cl_dirty_max = totalram_pages << (PAGE_CACHE_SHIFT - 3); +} + #endif /* __OBD_H */