X-Git-Url: https://git.camperquake.de/gitweb.cgi?a=blobdiff_plain;f=module%2Fzfs%2Fspa.c;h=0b9649723c904cb3506cd0d9d901cf8adebe6feb;hb=a38718a63d79116d6cb614dd2821e2a3955e5c8c;hp=64f5e067ca97bdd99faa5cb469314439fd7fe618;hpb=e75c13c353571efaa9b4e047f16969ec13a518f5;p=zfs.git diff --git a/module/zfs/spa.c b/module/zfs/spa.c index 64f5e06..0b96497 100644 --- a/module/zfs/spa.c +++ b/module/zfs/spa.c @@ -22,6 +22,9 @@ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. */ +/* + * Copyright 2011 Nexenta Systems, Inc. All rights reserved. + */ /* * This file contains all the routines used when modifying on-disk SPA state. @@ -40,6 +43,7 @@ #include #include #include +#include #include #include #include @@ -94,7 +98,7 @@ typedef struct zio_taskq_info { } zio_taskq_info_t; static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = { - "issue", "issue_high", "intr", "intr_high" + "iss", "iss_h", "int", "int_h" }; /* @@ -105,7 +109,7 @@ const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = { /* ISSUE ISSUE_HIGH INTR INTR_HIGH */ { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, { ZTI_FIX(8), ZTI_NULL, ZTI_BATCH, ZTI_NULL }, - { ZTI_BATCH, ZTI_FIX(5), ZTI_FIX(8), ZTI_FIX(5) }, + { ZTI_BATCH, ZTI_FIX(5), ZTI_FIX(16), ZTI_FIX(5) }, { ZTI_FIX(100), ZTI_NULL, ZTI_ONE, ZTI_NULL }, { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, { ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, @@ -113,7 +117,7 @@ const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = { static dsl_syncfunc_t spa_sync_props; static boolean_t spa_has_active_shared_spare(spa_t *spa); -static int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config, +static inline int spa_load_impl(spa_t *spa, uint64_t, nvlist_t *config, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig, char **ereport); static void spa_vdev_resilver_done(spa_t *spa); @@ -229,7 +233,9 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp) zap_attribute_t za; int err; - VERIFY(nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0); + err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP); + if (err) + return err; mutex_enter(&spa->spa_props_lock); @@ -241,7 +247,7 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp) /* If no pool property object, no more prop to get. */ if (mos == NULL || spa->spa_pool_props_object == 0) { mutex_exit(&spa->spa_props_lock); - return (0); + goto out; } /* @@ -271,8 +277,8 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp) dp = spa_get_dsl(spa); rw_enter(&dp->dp_config_rwlock, RW_READER); - if (err = dsl_dataset_hold_obj(dp, - za.za_first_integer, FTAG, &ds)) { + if ((err = dsl_dataset_hold_obj(dp, + za.za_first_integer, FTAG, &ds))) { rw_exit(&dp->dp_config_rwlock); break; } @@ -334,7 +340,7 @@ spa_prop_validate(spa_t *spa, nvlist_t *props) { nvpair_t *elem; int error = 0, reset_bootfs = 0; - uint64_t objnum; + uint64_t objnum = 0; elem = NULL; while ((elem = nvlist_next_nvpair(props, elem)) != NULL) { @@ -398,7 +404,7 @@ spa_prop_validate(spa_t *spa, nvlist_t *props) break; } - if (error = dmu_objset_hold(strval, FTAG, &os)) + if ((error = dmu_objset_hold(strval,FTAG,&os))) break; /* Must be ZPL and not gzip compressed. */ @@ -611,9 +617,8 @@ spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub) static taskq_t * spa_taskq_create(spa_t *spa, const char *name, enum zti_modes mode, - uint_t value) + uint_t value, uint_t flags) { - uint_t flags = TASKQ_PREPOPULATE; boolean_t batch = B_FALSE; switch (mode) { @@ -663,18 +668,22 @@ spa_create_zio_taskqs(spa_t *spa) const zio_taskq_info_t *ztip = &zio_taskqs[t][q]; enum zti_modes mode = ztip->zti_mode; uint_t value = ztip->zti_value; + uint_t flags = 0; char name[32]; + if (t == ZIO_TYPE_WRITE) + flags |= TASKQ_NORECLAIM; + (void) snprintf(name, sizeof (name), "%s_%s", zio_type_name[t], zio_taskq_types[q]); spa->spa_zio_taskq[t][q] = - spa_taskq_create(spa, name, mode, value); + spa_taskq_create(spa, name, mode, value, flags); } } } -#ifdef _KERNEL +#if defined(_KERNEL) && defined(HAVE_SPA_THREAD) static void spa_thread(void *arg) { @@ -764,6 +773,7 @@ spa_activate(spa_t *spa, int mode) ASSERT(spa->spa_proc == &p0); spa->spa_did = 0; +#ifdef HAVE_SPA_THREAD /* Only create a process if we're going to be around a while. */ if (spa_create_process && strcmp(spa->spa_name, TRYIMPORT_NAME) != 0) { if (newproc(spa_thread, (caddr_t)spa, syscid, maxclsyspri, @@ -784,6 +794,7 @@ spa_activate(spa_t *spa, int mode) #endif } } +#endif /* HAVE_SPA_THREAD */ mutex_exit(&spa->spa_proc_lock); /* If we didn't create a process, we need to create our taskqs. */ @@ -1144,7 +1155,7 @@ spa_load_l2cache(spa_t *spa) uint_t nl2cache; int i, j, oldnvdevs; uint64_t guid; - vdev_t *vd, **oldvdevs, **newvdevs; + vdev_t *vd, **oldvdevs, **newvdevs = NULL; spa_aux_vdev_t *sav = &spa->spa_l2cache; ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL); @@ -1270,7 +1281,7 @@ load_nvlist(spa_t *spa, uint64_t obj, nvlist_t **value) nvsize = *(uint64_t *)db->db_data; dmu_buf_rele(db, FTAG); - packed = kmem_alloc(nvsize, KM_SLEEP); + packed = kmem_alloc(nvsize, KM_SLEEP | KM_NODEBUG); error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed, DMU_READ_PREFETCH); if (error == 0) @@ -1293,8 +1304,9 @@ spa_check_removed(vdev_t *vd) spa_check_removed(vd->vdev_child[c]); if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd)) { - zfs_post_autoreplace(vd->vdev_spa, vd); - spa_event_notify(vd->vdev_spa, vd, ESC_ZFS_VDEV_CHECK); + zfs_ereport_post(FM_EREPORT_RESOURCE_AUTOREPLACE, + vd->vdev_spa, vd, NULL, 0, 0); + spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_CHECK); } } @@ -1790,7 +1802,8 @@ spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type, * Load an existing storage pool, using the pool's builtin spa_config as a * source of configuration information. */ -static int +__attribute__((always_inline)) +static inline int spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config, spa_load_state_t state, spa_import_type_t type, boolean_t mosconfig, char **ereport) @@ -1966,7 +1979,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config, cmn_err(CE_WARN, "pool '%s' could not be " "loaded as it was last accessed by " "another system (host: %s hostid: 0x%lx). " - "See: http://www.sun.com/msg/ZFS-8000-EY", + "See: http://zfsonlinux.org/msg/ZFS-8000-EY", spa_name(spa), hostname, (unsigned long)hostid); return (EBADF); @@ -2171,7 +2184,7 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config, * to start pushing transactions. */ if (state != SPA_LOAD_TRYIMPORT) { - if (error = spa_load_verify(spa)) + if ((error = spa_load_verify(spa))) return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error)); } @@ -3638,7 +3651,7 @@ spa_export_common(char *pool, int new_state, nvlist_t **oldconfig, } } - spa_event_notify(spa, NULL, ESC_ZFS_POOL_DESTROY); + spa_event_notify(spa, NULL, FM_EREPORT_ZFS_POOL_DESTROY); if (spa->spa_state != POOL_STATE_UNINITIALIZED) { spa_unload(spa); @@ -3816,7 +3829,7 @@ int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing) { uint64_t txg, dtl_max_txg; - vdev_t *rvd = spa->spa_root_vdev; + ASSERTV(vdev_t *rvd = spa->spa_root_vdev;) vdev_t *oldvd, *newvd, *newrootvd, *pvd, *tvd; vdev_ops_t *pvops; char *oldvdpath, *newvdpath; @@ -3969,7 +3982,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing) if (newvd->vdev_isspare) { spa_spare_activate(newvd); - spa_event_notify(spa, newvd, ESC_ZFS_VDEV_SPARE); + spa_event_notify(spa, newvd, FM_EREPORT_ZFS_DEVICE_SPARE); } oldvdpath = spa_strdup(oldvd->vdev_path); @@ -4001,7 +4014,7 @@ spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing) spa_strfree(newvdpath); if (spa->spa_bootfs) - spa_event_notify(spa, newvd, ESC_ZFS_BOOTFS_VDEV_ATTACH); + spa_event_notify(spa, newvd, FM_EREPORT_ZFS_BOOTFS_VDEV_ATTACH); return (0); } @@ -4016,10 +4029,10 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done) { uint64_t txg; int error; - vdev_t *rvd = spa->spa_root_vdev; + ASSERTV(vdev_t *rvd = spa->spa_root_vdev;) vdev_t *vd, *pvd, *cvd, *tvd; boolean_t unspare = B_FALSE; - uint64_t unspare_guid; + uint64_t unspare_guid = 0; char *vdpath; int c, t; @@ -4202,7 +4215,7 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done) vd->vdev_detached = B_TRUE; vdev_dirty(tvd, VDD_DTL, vd, txg); - spa_event_notify(spa, vd, ESC_ZFS_VDEV_REMOVE); + spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_REMOVE); /* hang on to the spa before we release the lock */ spa_open_ref(spa, FTAG); @@ -5033,9 +5046,6 @@ spa_async_probe(spa_t *spa, vdev_t *vd) static void spa_async_autoexpand(spa_t *spa, vdev_t *vd) { - sysevent_id_t eid; - nvlist_t *attr; - char *physpath; int c; if (!spa->spa_autoexpand) @@ -5049,17 +5059,7 @@ spa_async_autoexpand(spa_t *spa, vdev_t *vd) if (!vd->vdev_ops->vdev_op_leaf || vd->vdev_physpath == NULL) return; - physpath = kmem_zalloc(MAXPATHLEN, KM_SLEEP); - (void) snprintf(physpath, MAXPATHLEN, "/devices%s", vd->vdev_physpath); - - VERIFY(nvlist_alloc(&attr, NV_UNIQUE_NAME, KM_SLEEP) == 0); - VERIFY(nvlist_add_string(attr, DEV_PHYS_PATH, physpath) == 0); - - (void) ddi_log_sysevent(zfs_dip, SUNW_VENDOR, EC_DEV_STATUS, - ESC_DEV_DLE, attr, &eid, DDI_SLEEP); - - nvlist_free(attr); - kmem_free(physpath, MAXPATHLEN); + spa_event_notify(vd->vdev_spa, vd, FM_EREPORT_ZFS_DEVICE_AUTOEXPAND); } static void @@ -5228,7 +5228,7 @@ spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx) * saves us a pre-read to get data we don't actually care about. */ bufsize = P2ROUNDUP(nvsize, SPA_CONFIG_BLOCKSIZE); - packed = kmem_alloc(bufsize, KM_SLEEP); + packed = vmem_alloc(bufsize, KM_SLEEP); VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR, KM_SLEEP) == 0); @@ -5236,7 +5236,7 @@ spa_sync_nvlist(spa_t *spa, uint64_t obj, nvlist_t *nv, dmu_tx_t *tx) dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx); - kmem_free(packed, bufsize); + vmem_free(packed, bufsize); VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db)); dmu_buf_will_dirty(db, tx); @@ -5596,7 +5596,7 @@ spa_sync(spa_t *spa, uint64_t txg) ddt_sync(spa, txg); dsl_scan_sync(dp, tx); - while (vd = txg_list_remove(&spa->spa_vdev_txg_list, txg)) + while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, txg))) vdev_sync(vd, txg); if (pass == 1) @@ -5678,7 +5678,7 @@ spa_sync(spa_t *spa, uint64_t txg) /* * Update usable space statistics. */ - while (vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg))) + while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg)))) vdev_sync_done(vd, txg); spa_update_dspace(spa); @@ -5857,8 +5857,7 @@ spa_has_active_shared_spare(spa_t *spa) } /* - * Post a sysevent corresponding to the given event. The 'name' must be one of - * the event definitions in sys/sysevent/eventdefs.h. The payload will be + * Post a FM_EREPORT_ZFS_* event from sys/fm/fs/zfs.h. The payload will be * filled in from the spa and (optionally) the vdev. This doesn't do anything * in the userland libzpool, as we don't want consumers to misinterpret ztest * or zdb as real changes. @@ -5867,49 +5866,65 @@ void spa_event_notify(spa_t *spa, vdev_t *vd, const char *name) { #ifdef _KERNEL - sysevent_t *ev; - sysevent_attr_list_t *attr = NULL; - sysevent_value_t value; - sysevent_id_t eid; - - ev = sysevent_alloc(EC_ZFS, (char *)name, SUNW_KERN_PUB "zfs", - SE_SLEEP); - - value.value_type = SE_DATA_TYPE_STRING; - value.value.sv_string = spa_name(spa); - if (sysevent_add_attr(&attr, ZFS_EV_POOL_NAME, &value, SE_SLEEP) != 0) - goto done; - - value.value_type = SE_DATA_TYPE_UINT64; - value.value.sv_uint64 = spa_guid(spa); - if (sysevent_add_attr(&attr, ZFS_EV_POOL_GUID, &value, SE_SLEEP) != 0) - goto done; - - if (vd) { - value.value_type = SE_DATA_TYPE_UINT64; - value.value.sv_uint64 = vd->vdev_guid; - if (sysevent_add_attr(&attr, ZFS_EV_VDEV_GUID, &value, - SE_SLEEP) != 0) - goto done; - - if (vd->vdev_path) { - value.value_type = SE_DATA_TYPE_STRING; - value.value.sv_string = vd->vdev_path; - if (sysevent_add_attr(&attr, ZFS_EV_VDEV_PATH, - &value, SE_SLEEP) != 0) - goto done; - } - } - - if (sysevent_attach_attributes(ev, attr) != 0) - goto done; - attr = NULL; - - (void) log_sysevent(ev, SE_SLEEP, &eid); - -done: - if (attr) - sysevent_free_attr(attr); - sysevent_free(ev); + zfs_ereport_post(name, spa, vd, NULL, 0, 0); #endif } + +#if defined(_KERNEL) && defined(HAVE_SPL) +/* state manipulation functions */ +EXPORT_SYMBOL(spa_open); +EXPORT_SYMBOL(spa_open_rewind); +EXPORT_SYMBOL(spa_get_stats); +EXPORT_SYMBOL(spa_create); +EXPORT_SYMBOL(spa_import_rootpool); +EXPORT_SYMBOL(spa_import); +EXPORT_SYMBOL(spa_tryimport); +EXPORT_SYMBOL(spa_destroy); +EXPORT_SYMBOL(spa_export); +EXPORT_SYMBOL(spa_reset); +EXPORT_SYMBOL(spa_async_request); +EXPORT_SYMBOL(spa_async_suspend); +EXPORT_SYMBOL(spa_async_resume); +EXPORT_SYMBOL(spa_inject_addref); +EXPORT_SYMBOL(spa_inject_delref); +EXPORT_SYMBOL(spa_scan_stat_init); +EXPORT_SYMBOL(spa_scan_get_stats); + +/* device maniion */ +EXPORT_SYMBOL(spa_vdev_add); +EXPORT_SYMBOL(spa_vdev_attach); +EXPORT_SYMBOL(spa_vdev_detach); +EXPORT_SYMBOL(spa_vdev_remove); +EXPORT_SYMBOL(spa_vdev_setpath); +EXPORT_SYMBOL(spa_vdev_setfru); +EXPORT_SYMBOL(spa_vdev_split_mirror); + +/* spare statech is global across all pools) */ +EXPORT_SYMBOL(spa_spare_add); +EXPORT_SYMBOL(spa_spare_remove); +EXPORT_SYMBOL(spa_spare_exists); +EXPORT_SYMBOL(spa_spare_activate); + +/* L2ARC statech is global across all pools) */ +EXPORT_SYMBOL(spa_l2cache_add); +EXPORT_SYMBOL(spa_l2cache_remove); +EXPORT_SYMBOL(spa_l2cache_exists); +EXPORT_SYMBOL(spa_l2cache_activate); +EXPORT_SYMBOL(spa_l2cache_drop); + +/* scanning */ +EXPORT_SYMBOL(spa_scan); +EXPORT_SYMBOL(spa_scan_stop); + +/* spa syncing */ +EXPORT_SYMBOL(spa_sync); /* only for DMU use */ +EXPORT_SYMBOL(spa_sync_allpools); + +/* properties */ +EXPORT_SYMBOL(spa_prop_set); +EXPORT_SYMBOL(spa_prop_get); +EXPORT_SYMBOL(spa_prop_clear_bootfs); + +/* asynchronous event notification */ +EXPORT_SYMBOL(spa_event_notify); +#endif